JP2004054732A - Human resource utilization support system and human resource utilization support program - Google Patents

Abstract

<P>PROBLEM TO BE SOLVED: To support revitalization of human resources by providing an analytical environment about functions of persons and jobs required by a company. <P>SOLUTION: A human resource utilization support system comprises individual analyzing answer information receiving means 11 for receiving individual analyzing answer information including a current position and answer sentences to questions, answer information analyzing means 14 for relating frequencies of appearance of answer words in the answer sentences to the answer words, job identification word storing means 16 for defining identification words related to jobs required by each position, position compatibility determining means 17 for determining matching between the current position and a position based on identification words corresponding to answer words of a given frequency of appearance or higher, evaluation parameter generating means 19 for generating evaluation parameters based on the frequencies of appearance of the answer words, comparison parameter storing means 21 for storing comparison parameters as comparison standards of the evaluation parameters, job compatibility determining means 22 for comparing the parameters, and information outputting means 24 for outputting the determination result. <P>COPYRIGHT: (C)2004,JPO

Description

【表２】

【００７６】
そして、表１および表２に示したように、本実施形態では、上述した分析区分に加え、基準単位単語の属性に鑑みて定められた抽出区分を規定する。ここで、抽出区分は、一括して抽出すべき分析区分の単語を特定するために設けられた便宜的な区分であり、第２階層区分の所定の区分に応じて、ａ，ｂ，ｃの３つの区分が指定されている。例えば、キーワードとして分解した単語を選別する際に、区分ａと区分ｂに該当する単語を抽出する際のインデックスとして用いる。
【００７７】
ここで、区分ａは、内容を修飾または補足する単語群をまとめた区分を意味する。区分ｂは、内容を具体的に示す単語群をまとめた区分を意味する。区分ｃは、その単語だけでは意味をなさない単語群をまとめた区分を意味する。
【００７８】
なお、これらの分析区分の階層構造データは、分解後に、分析区分に基づくカウントが行われる際の利便性を考慮して、記憶手段の所定の領域に、分析用辞書記憶手段１３に記憶される基準単位単語に関連付けられる分析区分を規定するデータとして記憶される。記憶の形態は、例えば、データベース構造などを採用し、各辞書の分析区分の階層構造に応じて、階層区分と分析区分コードとの対応付けを規定するとともに、高速検索のための検索インデックスとして抽出区分との対応付けも規定する。
【００７９】
以上説明した分析区分と関連付けられた基準単位単語を記憶する各辞書およびその他の辞書について詳述する。
【００８０】
体言辞書記憶領域は、一般的な国文法では普通名詞、固有名詞、代名詞、数詞（主に単位など）、副詞、連体詞、感動詞に分類される基準単位単語を記憶する。ここで、本発明に特徴的なことは、一般的な国文法では体言に分類される接続詞について、独立したかたちで接続詞辞書を設けていることと、連語を体言辞書に登録していることである。
【００８１】
これは、接続詞辞書が、実際に文章表現内容の分析を行う際に高い頻度で利用されることに鑑み、体言辞書に含めず独立させて設ける構成を採用したものであり、自然文を分解して単語にする際に用いることとは離れて、異なる言語分析を行う構成を付加的に備えた場合などを想定し、文章表現内容の分析に兼用しやすいようにしたものである。また、連語も二つ以上の単語が結合してはいるが、ひとまとまりの観念を表す語であるので、体言辞書に記憶している。
【００８２】
用言辞書記憶領域は、一般的な国文法では動詞、形容詞に分類される基準単位単語を記憶し、同様に、助詞辞書記憶領域は、助詞に分類される基準単位単語を、助動詞記憶領域は、助動詞に分類される基準単位単語をそれぞれ記憶する。
【００８３】
また、活用辞書記憶領域は、一般的な国文法では動詞、形容詞に分類される基準単位単語の活用として語尾変化のパターンなどを記憶しており、自然文の分解に際して、その単語が用言であるか否かを認定する際に用言辞書と併用される。
【００８４】
句点識別辞書記憶領域は、自然文の末尾を認識するために、句点（。）を記憶する。ただし、必要に応じて、句点識別辞書記憶領域は、さらに、読点（、）や疑問符（？）、感嘆符（！）などを記憶してもよい。
【００８５】
削除記号辞書記憶領域は、分解や分析に関係しない記号で、削除しても問題がない記号を記憶する。例えば、文章中で見出しなどに使う「○」「●」「◎」「§」「※」などの記号が挙げられる。
【００８６】
表記辞書記憶領域は、電子文書などで特殊記号に分類される自然文中の記号を変換するために、変換元情報と変換先情報を関連付けて記憶する。例えば、百分率記号（％）と「パーセント」という単語、ドル記号（＄）と「ドル」という単語、円（￥）と「円」または「エン」などを関連付けて記憶する。
【００８７】
感想単語辞書記憶領域は、回答文を回答単語に分解する段階では使用しないが、後述する中心キーワード比率や中心補完単語比率などの分析値を生成する際に必要となる感想単語を格納する。詳細には、感想単語辞書記憶領域は、本発明独自の区分として、感想単語という分析用の特別区分を設け、これに該当する単語の数を差し引いた分析値を生成する処理を行うために、その感想単語に該当する基準単位単語を格納し、一致判断の際の参照辞書とする辞書記憶領域である。
【００８８】
ここで、感想単語に該当するのは、「私は…と思う」と表現される場合の、「私」「思う」など、自然文制作者の主観的な感想として位置づけられる、主体を自分として述べるときに用いられる単語である。例えば、感想名詞に該当する単語として、「自ら」「自分」「私」「僕」「当社」「我社」など、感想動詞に該当する単語として、「思う」「できる」「対する」「考える」「感じる」などが挙げられる。
【００８９】
なお、本実施形態では、体言辞書記憶領域、用言辞書記憶領域、接続詞辞書記憶領域、など、概ね分析区分と一致する構成として、回答文を回答単語に分解する際の参照の高速化を図れるようにしたが、これに限られるものではない。例えば、分析区分と対応付けられ、自然文の分解に際して基準単位単語として参照する分析手順を実現できるものであれば、独立した単語群を形成するように記憶されたデータベース構造に代えて、単語がその分析区分や品詞種別に基づきまとめられることなく混在したデータベース構造を採用してもよい。
【００９０】
また、必ずしも単一の分析用辞書記憶手段１３が、図３に示したように、複数の辞書を統括して記憶する構成に限られるものではなく、複数の記憶手段から分析用辞書記憶手段１３が構成され、複数にグループ化された辞書群を分散して記憶する構成を採用してもよい。また、辞書も上述した機能別の辞書を複数備える構成に限るものではなく、例えば、用言辞書と活用辞書を統合した複合機能を果たす辞書を構成要素としてもよい。
【００９１】
続いて、回答情報分析手段１４は、回答文が有する回答単語を分析し、回答単語の出現頻度を当該回答単語に関連付けるとともに、個人の現在の所属についての情報を回答情報から取得する手段であり、回答文を単語単位の情報を含む個人分析用情報に変換する機能を果たす。
【００９２】
具体的には、回答情報分析手段１４は、個人分析用回答情報記憶手段１２から個人分析用回答情報にかか回答文を読み出し、読み出した回答文について、分析用辞書記憶手段１３が記憶する基準単位単語を参照して、回答文が有する回答単語と基準単位単語との一致を判別する分析を行い、一致をもって、回答文が有する回答単語を特定し、回答文を回答単語単位に分解する処理を行う。その後、回答情報分析手段１４は、分解した回答単語について、同一単語の重複数をカウント値として取得し、回答単語の出現回数を求め、この出現回数を出現頻度として、回答単語と関連付ける。また、回答情報分析手段１４は、個人分析用回答情報が有する回答者個人情報もしくは回答文の付帯情報から、分析対象となる個人の現在の所属を取得する。
【００９３】
加えて、回答情報分析手段１４は、この分析の結果得られた回答単語と回答単語の出現頻度とを、前述した個人識別情報と関連付けて分析結果記憶手段１５に記憶する手段として、また、現在の所属とその個人識別情報とを関連付けて分析結果記憶手段１５に記憶する手段としても、機能する。
【００９４】
ここで、回答文の回答単語への分解について、より詳細に説明する。回答情報分析手段１４は、句点識別辞書記憶領域を参照しながら、個人分析用回答情報記憶手段１２に記憶された分析対象となる回答文をセンテンス単位で読み出し、読み出した自然文について、所定の文字数単位で自然文の先頭から順番に文字列を抜き出し、分析用辞書記憶手段１３の体言辞書記憶領域、用言辞書記憶領域、接続詞辞書記憶領域など、所定の順序で各辞書を参照し、辞書内に記憶された基準単位単語と比較して分解される単位となる単語を特定することにより単語に分解する手段として機能する。
【００９５】
また、回答情報分析手段１４は、複数の単語に自然文を分解するに際し、分解した単語に前述した分析区分を関連付ける手段としても機能する。分析用辞書記憶手段１３に記憶される基準単位単語に関連付けられた分析区分を、分解した単語に関連付けるべき分析区分として用いることにより、分解した単語の属性に応じた分析区分を付与する手段として機能する。
【００９６】
なお、回答情報分析手段１４が実現する、個人分析用回答情報記憶手段１２に記憶された回答文を回答単語に分解するプロセスは、上述の態様に限られるものではなく、少なくとも単語とそれに応じた分析区分を関連付けて分解記憶することができるものであれば足りる。
【００９７】
分析結果記憶手段１５は、前述したように、回答情報分析手段１４が回答文を単語レベルに分解して得られた回答単語を記憶する手段であり、図４に示したデータベース構造を採用する。自然文を分解して得られた回答単語と、その回答単語の属性に応じた分析区分と、前述した出現頻度とを関連付けて記憶する機能を呈する。
【００９８】
加えて、分析結果記憶手段１５は、回答単語の出現頻度の履歴を、前述した個人識別情報と関連付けて記憶する機能も備える。詳細には、図４に示したように、個人分析用回答情報を日、月、年など、定期的に受け付ける場合の、回答単語の出現頻度や回答単語に関連する情報の履歴を記憶可能な構造を採用し、人材能力の変化や推移を含めた分析を実現可能にする。
【００９９】
職務識別単語記憶手段１６は、企業が備える所属について、各所属に必要とされる職務と関連付けられる識別単語を規定した記憶手段であり、具体的には、図５に示したようなデータベース構造を採用する。ここで、図５に示した職務識別単語記憶手段１６は、所属に必要とされる職務を表す定義的な単語をそのまま識別単語として設定した例を示す。例えば、図５に示したように、管理部であれば、経理、会計、営業部であれば販売、企画という普通名詞が識別単語として規定される。
【０１００】
ただし、各所属の職務についてみると、所属において必要とされる職務は通常複数有り、その所属に固有の職務のみならず、各所属に共通する職務なども存在するので、識別単語は各所属を代表する定義的な単語に限られるものではなく、複数の所属に共通する職務である交渉、渉外、折衝などの単語も、それらを職務として必要とする所属と関連付けられる識別単語となる。
【０１０１】
また、本実施形態では、職務を定義づける単語をそのまま職務と関連付けられる識別単語としたが、直接的に職務を表す単語を用いるのではなく、所属に必要とされる職務を示唆する単語を、間接的に職務を表す識別単語として用いてもよい。そのような場合には、識別単語は名詞だけに限られず、動詞なども該当することが生じ得る。
【０１０２】
なお、複数の企業に共通する認識に基づく識別単語を事前に規定し記憶するだけでなく、処理手段３３が、その企業の風土や、社内用語を踏まえた企業独自の単語を識別単語として受け付けて職務識別単語記憶手段１６に記憶する処理や、各所属のベテラン社員から得た個人分析用回答情報より抽出した回答単語を識別単語に加える処理を行うことにより、より的確に職務を表す識別単語を規定する職務識別単語記憶手段１６を提供できる。
【０１０３】
続いて、所属適合判定手段１７は、職務識別単語記憶手段１６を参照し、所定の出現頻度以上の回答単語に対応する識別単語に基づき得られた所属と現在の所属とを比較し、一致するかを判定する手段である。
【０１０４】
具体的には、所属適合判定手段１７は、所属と、所属に必要な職務を表す単語との相関に基づく回帰的手法を用いて個人の所属の適否を判断するものであり、個人識別情報をインデックスとして、その個人が回答文に記載した回答単語と、その個人の現在の所属とを分析結果記憶手段１５から取得する機能と、回答単語に関連付けられた出現頻度に基づいて、分析に用いる回答単語のフィルタリングを行う機能を備え、さらに、フィルタリングした回答単語群と所属に必要とされる職務に関連付けられる識別単語群とを比較し、完全一致、もしくは所定の割合以上の部分一致をもって一致するものと判断し、その一致するものと判断した個人に適していると位置づけられる、識別単語群に関連付けられた所属を好適所属として取得する機能を果たす。
【０１０５】
また、所属適合判定手段１７は、上述した一の回答情報に含まれる回答単語の出現頻度をフィルタリングに用いるだけでなく、分析結果記憶手段１５に記憶された回答単語やその出現頻度などの関連情報についての履歴に基づき抽出された、所定の時間サイクル以上で用いられる回答単語に対応する識別単語に基づき得られた好適所属を比較する機能をも備える。
【０１０６】
具体的には、所属適合判定手段１７は、回答文の中に出てくる記載上の出現頻度だけでなく、日、月、年など経時的サイクルの観点から得られる経時的な出現頻度についてもフィルタリングの要素として用いる。例えば、季節によって業務上の職務が変化する場合には、連続３ヶ月という時間サイクルを条件として、適合判定に用いる回答単語とする、あるいは、一年間の平均、一ヶ月の平均、一週間の平均という時間サイクルを条件とするなど、所定の期間内に一定の出現頻度を示した回答単語に対応する識別単語に基づき好適所属を取得した上で、上述した所属比較する。
【０１０７】
このように、所属適合判定手段１７が、回答単語の出現頻度の履歴に基づき、所定の時間サイクル以上で用いられる回答単語を抽出し、抽出した回答単語に対応する識別単語に基づき得られた所属を比較する構成を採用することにより、個人が頻繁に活用している職務に基づき判定することができ、より個人の特質に沿った判定が可能となる。
【０１０８】
加えて、所属適合判定手段１７は、一致すると判断した好適所属と現在の所属とを比較し、一致するか否かを判定し、その判定した所属適合判定結果を所属適合判定結果記憶手段１８に記憶する手段としても機能する。
【０１０９】
ここで、所属適合判定結果記憶手段１８は、前述した所属適合判定手段１７が判定した所属適合判定結果を記憶する手段であるが、具体的には、図６に示したように、個人に固有の個人識別情報、所属情報、および、所属適合判定結果とを関連付けて記憶するデータベースの形態を採用する。
【０１１０】
以上説明した、個人分析用回答情報受付手段１１、個人分析用回答情報記憶手段１２、分析用辞書記憶手段１３、回答情報分析手段１４、分析結果記憶手段１５、職務識別単語記憶手段１６、所属適合判定手段１７、所属適合判定結果記憶手段１８が有機的に結合し機能することにより、受け付けた個人分析用回答情報に基づき、現在の所属と個人の備える職務との適合性を判定する人材活用支援システム１が実現する。
【０１１１】
このような構成を採用することにより、第三者の恣意が入り込むことのない客観的な判断材料に基づく判定結果を得ることが可能となり、ひいては、企業の人材活用の一助となる資料を提供できる人材活用支援システムを実現できる。
【０１１２】
ここで、現在、各個人が遂行している職務は、個人の有する職能によりその実現が図られるものであり、回答単語から得られる職務の適合性を判断することは、ひいては、個人の有する職能の適合性を判断することにもなる。例えば、営業にいながら、遂行している業務が管理の職務と完全に一致しており、逆に、営業の職務と不一致であるという結果を人材活用支援システム１が提供した場合には、その結果が配置転換などによる人材活用のための判断材料となる。
【０１１３】
続いて、職務適応性を判定する処理にかかわる、評価パラメータ生成手段１９、評価パラメータ記憶手段２０、比較パラメータ記憶手段２１、職務適応性判定手段２２、職務適応性判定結果記憶手段２３について説明する。
【０１１４】
まず、評価パラメータ生成手段１９は、回答単語の出現頻度に基づく個人分析用評価パラメータを生成する手段である。ここで、本実施形態において生成される評価パラメータとしては、回答単語などの数から得られる個人の能力を表す比率などの数値が所定の基準に対してどの程度乖離しているかを示す乖離値と、その乖離値に基づき得られる顕在能力を評価する顕在能力評価値とが挙げられる。したがって、いずれも、日本語表現における単語使用のバランスを客観的に数値化した乖離値に左右されるパラメータであるといえる。
【０１１５】
より具体的には、評価パラメータ生成手段１９は、まず、分析用辞書記憶手段１３に記憶された回答文を単語に分解する際の基準となる基準単位単語と分析区分との関連付けを参照して付与された回答単語と分析区分との関連付けと、当該分析区分に属する前記回答単語の出現頻度や回答単語の種類数など回答単語に関連する数に基づき得られる数値を用いて普遍化した乖離値である評価パラメータを生成する手段として機能する。加えて、評価パラメータ生成手段１９は、前述した個人のコンピテンシーを評価できる、生成した複数の乖離値に基づく顕在能力評価値である評価パラメータを生成する手段としても機能する。
【０１１６】
以下、評価パラメータ生成手段１９の評価パラメータ生成機能につき、詳細に説明する。概略的には、評価パラメータ生成手段１９は、上述したいずれの機能を発揮する場合においても、評価パラメータを生成するために、分析結果記憶手段に記憶された回答単語に関連するカウント値を用いて普遍化した比率を分析値とし、この分析値の基準分析値に対する乖離の程度を示す乖離値を判定のためのパラメータとして生成するという段階的生成シーケンスを採用するものである。ゆえに、理解容易のため、分析値の生成までと乖離値の生成とを分けて説明する。なお、詳細は後述するが、乖離値の生成に用いる基準分析値は、原則として、日本の文化的背景を分析に織り込むべく、社説や論文などの文献から得た分析値を基準分析値とする。
【０１１７】
評価パラメータ生成手段１９は、個人単位で、分析結果記憶手段１５が記憶した複数の回答単語を参照し、少なくとも分解した回答単語にかかわる数をカウント値として取得し、このカウント値に基づき各種分析値を生成する。ここで、「少なくとも」としたのは、回答情報分析手段１４が直接的に分解した単語数のみではなく、必要に応じて、分析結果記憶手段１５のみならず、個人分析用回答情報記憶手段１２なども参照し、対象となる自然文の総文字数や、センテンス数、各分析区分に該当する単語の文字数を合計した数、などについてもカウントし、そのカウント値を取得するからである。なお、評価パラメータ生成手段１９により生成された分析値などに基づく再カウントなども、必要に応じて行う。
【０１１８】
本発明にかかる分析値の代表的なものを表３に示す。表３には、回答文を書いた個人の能力を、その回答文の内容に基づき評価できる分析項目内容と、分析項目内容に対応する分析値、および、その分析値の算出則を記している。
【表３】

【０１１９】
例えば、表３に示したように、回答文を記載した個人の能力評価の一要素として、文章整備力なら総頻度比率を評価指標の基礎データとして用いることができる。また、その算出則は総頻度数を計算文字数で除すというものである。なお、算出則に用いられる各要素の詳細は後述する。
【０１２０】
ここで、表３の分析項目内容に記載した各能力に対応付けられた分析値は、発明者の長年の研究および分析から得られた経験に基づき、文章の内容を評価する際の各能力を数値評価するに適していると認定されたものである。ここで、分析値として得られた数値を絶対比較しても、その能力に関する他の文章との優劣を判断することが可能である。なお、表３に示した分析値および分析値算出則は一例であり、それ以外の観点から算出則を規定し、その規定した算出則に基づき得られる分析値を用いた評価を行う場合でも、好適に本発明を用いることができることはいうまでもない。
【０１２１】
そして、本実施形態においては、表３に示したような分析値算出則記憶領域への分析値算出則の記憶形態として、プログラム関数もしくはモジュールの演算機能のかたちで記憶手段３１に記憶される構成を採用しており、評価パラメータ生成手段１９が記憶手段３１の所定の記憶領域に記憶されたプログラム関数またはモジュールを利用して分析値を生成するという方式を採る。
【０１２２】
ただし、分析値算出則の記憶は、上述したように、関数として繰り返し用いることができるようにサブルーチン的に記憶する形態に限られるものではなく、シーケンス制御の手順として、重複して逐次規定され記憶される形態であってもよい。
【０１２３】
以下、上述した分析値自体の生成に用いるカウント値の取得について詳細に説明する。評価パラメータ生成手段１９は、上述したような形態で分析結果記憶手段１５が記憶する分解された単語群に基づき、単語に関連付けられた分析区分を検索インデックスとするなどしながら単語の数をカウントし、そのカウント値を取得する。
【０１２４】
ここで、分析結果記憶手段１５に記憶された単語群に基づくカウント値の例としては、第１階層区分、第２階層区分、第３階層区分など、各区分が付与された単語の総数を区分毎にカウントしたカウント値が挙げられる。カウント値として得られるデータ項目のうち、第１階層区分の体言（第１階層区分コード０１）に該当する単語の数をカウントして得られた体言数、同様に、用言（第１階層区分コード０２）に該当する単語の数をカウントして得られた用言数、付属語（第１階層区分コード０３）に該当する単語の数をカウントして得られた付属語数などがこれに該当する。
【０１２５】
ただし、評価パラメータ生成手段１９が取得するカウント値は、単に特定の分析区分に該当する単語の総数に限られるものではなく、重複する同一単語はひとまとまり（一つ）としてカウントして求められる単語の種類数（以下、出現単語数と称する）などであってもよい。また、分析区分という枠内に該当する単語のみをカウント対象とするものではなく、その分析区分をまたがって単語数をカウントしてもよい。
【０１２６】
例えば、分析対象となる文章中に出現した回数（以下、この出現回数を頻度数と称する）が同じである単語の数をカウントして求められる頻度数が同じ単語の数（以下、同頻度数と称する）、さらには、頻度数が一である単語（分析対象文章中で一度しか出現していない単語）の数（以下、１頻度単語数と称する）、などをカウント値として評価パラメータ生成手段１９が取得する態様も採り得る。
【０１２７】
続いて、評価パラメータ生成手段１９は、各種カウント値を用いて分析値を生成するが、分析値の生成にあたっては、記憶手段３１の所定の領域に設けられた不図示の分析値算出則記憶領域に記憶された、表３に示したような分析値算出則を参照して算出する。
【０１２８】
算出される分析値のうち、直接的にカウント値を用いて得られる分析値の具体例としては、表３に示された総頻度数、除数詞名詞数やセンテンス用言率などが挙げられる。総頻度数は、分析対象となる複数の自然文に含まれる分析区分のうち、抽出区分ａおよびｂに該当する分析区分毎の単語の数をカウントして得られた総和をいう。また、除数詞名詞数は、第２階層区分が普通名詞、固有名詞、代名詞に該当する単語の数の総和から数詞に該当する単語の数を差し引いた単語数をいう。そして、センテンス用言率は、用言数をセンテンス数で除した数値をいう。
【０１２９】
ここで、本実施形態において、総頻度数をすべての単語の数にせず、抽出区分ａ，ｂに限ったのは、経験則から、抽出区分ｃに該当する単語の数を除外して分析を行うことにより、専門家が経験的によい文章であると判断した場合の人的分析評価に沿った、数値分析評価が得られると判明したからである。
【０１３０】
一方、カウント値を加工して得られた二次的な数値を用いて生成される分析値の具体例としては、総頻度比率、体言率、などの比率または率で表される、計算文字数でカウント値等を除して求められる分析値や、中心キーワード比率、中心補完単語比率、中心１頻度単語比率、第１単語頻度比率、第２単語頻度比率、第３単語重複率のように、カウント値を用いて算出したパラメータを計算文字数で除した分析値などが挙げられる。
【０１３１】
ここでいう計算文字数は、例えば固有名詞の文字数の多寡（外国の地名などの有無）や、対象となる文章自体の多寡など、物理的または表面的な要素により分析値が大きく影響を受けることがないように導入したパラメータであり、本発明に特徴的な概念である。
【０１３２】
本実施形態における計算文字数の算出則は、（計算文字数）＝（分析対象となる文章の総文字数）×（総頻度数）／（総頻度数でカウントされた分析区分に該当する単語の文字数の総和）である。ゆえに、比率を算出する際の分母となるため文字数と称したが、単語の数である総頻度数と、文字数から得られた無次元係数とから得られる頻度数とを用いて、分析値を生成しているものであり、これにより、文字量の影響を極力排除せんとするものである。
【０１３３】
より具体的には、冗長でわかりにくい文章と簡明で短い文章とを比較した際に、その文字量により左右されにくい分析値を得ることができるようにするため、単語の数に対して文字数に基づく補正をかけ、分析値の普遍化を実現する普遍化パラメータとして上述の計算文字数を選択したものである。
【０１３４】
また、中心キーワード比率、中心補完単語比率、中心１頻度単語比率などは、分解した単語のうちキーワードと認定される単語数が重要なパラメータとなる。
【０１３５】
先にも触れたが、キーワードとは、同頻度件数を同頻度のパターン数で除した同頻度比率が０．７以下の単語とされる。ここで、同頻度件数とは、特定の出現回数につき、その出現回数が同じである単語の種類数をいう。例えば、出現回数が４回の単語が３種類あったら、頻度数が４の同頻度件数は３とされる。また、同頻度のパターン数とは、同頻度と認定された単語群の数をいい、具体的には、出現回数が同じである単語群がいくつできるかをカウントしたものである。例えば、出現回数が１回の単語群、出現回数が２回の単語群、出現回数が３回の単語群、出現回数が４回の単語群、出現回数が６回の単語群、出現回数が１２回の単語群がある場合には、同頻度のパターン数は６となる。このとき、頻度数が４の単語群については、前述したように同頻度件数が３であるので、同頻度比率が０．５となり、キーワードとして認定されることになる。なお、参考までに、優秀な文章とされる社説のうちの１つについて分析した結果をあげると、頻度数１２が同頻度件数１件、頻度数６が同頻度件数１件、頻度数４が同頻度件数３件、頻度数３が同頻度件数５件、頻度数２が同頻度件数１８件、頻度数１が同頻度件数１７８件、となり、頻度数１２、頻度数６、頻度数４までに該当する単語群がキーワードとして認定された。
【０１３６】
このキーワード認定のために、評価パラメータ生成手段１９は、分解された単語もしくは単語群がキーワードに該当するか否かを、各カウント値に基づき算出した同頻度比率が０．７以上か否かにより判断する機能も備える。このとき、その判断結果に基づき、キーワードに該当する単語数（以下、キーワード数と称する）を評価パラメータ生成手段１９がカウントし、さらに、そのカウント値に基づき、評価パラメータ生成手段１９が分析値を生成する。
【０１３７】
具体的には、以上説明したキーワードに該当する単語数に基づき生成される分析値が、補完単語数であり、キーワード数から、前述した一頻度単語数を差し引いた数値をいう。
【０１３８】
さらに、評価パラメータ生成手段１９は、キーワードに該当すると判断したキーワード群や補完単語に該当すると判断した補完単語群の中に、感想単語が存在するか否かを判断する機能を備える。具体的には、前述した分析用辞書記憶手段１３に記憶されている不図示の感想単語辞書記憶領域を参照して、その感想単語として記憶されている基準単位単語との一致をみることにより判断する。
【０１３９】
評価パラメータ生成手段１９は、この感想単語の認定に基づき、中心キーワード数として、キーワード数から感想単語数を差し引いた数値を分析値として生成する。また、中心補完単語数として、補完単語数から感想単語を差し引いた数値を分析値として生成する。さらに、中心一頻度単語数として、一頻度単語に該当する単語から感想単語を差し引いた数値を分析値として生成する。その上で、評価パラメータ生成手段１９は、これら、中心キーワード数、中心補完単語数、中心一頻度単語数をそれぞれ計算文字数で除して、中心キーワード比率、中心補完単語比率、中心１頻度単語比率を生成し、分析値として記憶手段３１の所定の領域に記憶する。
【０１４０】
ただし、本実施形態におけるキーワードの認定方法として、上述した同頻度件数を同頻度のパターン数で除したものを同頻度比率としたが、キーワードの概念は、前述したように、ある一定回数以上の出現頻度で使用されている単語というものであり、この区別を行うための指標の一例として上述の同頻度比率を用いた構成を採用したものである。したがって、キーワードを認定するための基準となる頻度を定められるものであれば、これに限られるものではなく、キーワードを認定するための出現頻度のしきい値として、例えば、同頻度件数を総同頻度件数で除した値を同頻度比率として用いてもよい。
【０１４１】
他の分析値としては、第１単語頻度比率、第２単語頻度比率、第３単語重複率が挙げられる。ここで、第１単語、第２単語、第３単語、第４単語は、評価パラメータ生成手段１９が取得したカウント値に基づき、頻度数（出現回数）の多い単語を上位から４つ抽出して特定する。ゆえに、それぞれの単語の頻度数は、第１単語数、第２単語数、第３単語数、第４単語数、として取り扱われ、各々を計算文字数で除した値が第１単語頻度比率、第２単語頻度比率、第３単語頻度比率、第４単語頻度比率という分析値になる。また、第３単語重複率とは、第１単語、第２単語、第３単語とされた単語について、その種類数の和をとり、３で除した値をいう。
【０１４２】
そして、各分析値は、表３に示したように分析対象の文章の内容を表す数値となる。具体的には、文章整備力を表す数値として総頻度比率が、語彙力を表す数値として除数詞名詞率が、テーマ設定力を表す数値として中心キーワード比率が、　テーマ説明力を表す数値として中心補完単語比率が、テーマ補足力を表す数値として中心１頻度単語比率が、単語構成力を表す数値として出現単語比率が、センテンス構成力を表す数値としてセンテンス比率が、センテンス主張力を表す数値としてセンテンス頻度平均値が、主張力を表す数値として第１単語頻度比率が、　主張補足力を表す数値として第２単語頻度比率が、主張構成力を表す数値として第３単語重複率が、論理展開力を表す数値として同頻度数比率が、論理強制力を表す数値として単語回帰値が、行動表現力を表す数値としてセンテンス用言率が採用される。
【０１４３】
ここで、センテンス頻度平均値とは、各センテンスの区分ｂに該当する単語について、その単語重量値の総和をとり、そのセンテンス毎に得られた単語重量値の総和をさらに合計した値を、センテンス数で除した値である。ここでいう単語重量値は、単語の頻度比率であり、出現回数（頻度数）を計算文字数で除したものである。
【０１４４】
また、回帰値とはデータ分布を一つの直線に表したときの傾きを示す値であるが、単語回帰値の本来の意義は、すべての単語の頻度比率を分布図に表したときの直線の傾きをいう。ただし、本実施形態では、近似的に傾きを取得する構成を採用している。具体的には、縦軸のパラメータとして理想値を用い、横軸のパラメータとして分析対象文章の分析値をとり、理想値と合致する場合には４５度の傾きをなす分布直線を想定する。そして、単語の頻度比率としては、第２単語頻度比率を第１単語頻度比率で除した値、第３単語頻度比率を第１単語頻度比率で除した値、第４単語頻度比率を第１単語頻度比率で除した値の総計３つの値を用いて、これらのパラメータを用いて傾きを求める。なお、第１単語頻度比率で除したのは、回帰値を得るための傾きを単調増加する直線に近似できるようにするためである。
【０１４５】
詳細には、縦軸には、平均値±標準偏差の範囲に入ったときの値から算出された再平均値を用いて算出した理想値をとるとともに、横軸には、各分析対象文章の現実値をとり、これら理想値と現実値の３対のデータから最小自乗法により傾きを算出する。ここで、最小自乗法による算出には、公知のプログラム関数などを用いる。また、単語回帰値の傾きが表す傾向としては、傾きが４５度よりもなだらかな場合には、冗長であいまいな表現となっており、４５度より急な場合には命令張の断定的な表現となっているということが、経験的に得られている。
【０１４６】
以上、評価パラメータ生成手段１９が乖離値生成に先立ち、取得、生成するカウント値や分析値について説明したが、これらカウント値および分析値は、適宜、記憶手段３１の所定の記憶領域に、個人識別情報と関連付けて記憶される。そして、評価パラメータ生成手段１９は、必要に応じてその所定の記憶領域から読み出し、乖離値の生成に用いる。
【０１４７】
次に、乖離値の生成について詳細に説明する。評価パラメータ生成手段１９は、前述したように、乖離値の生成にあたっては、その乖離の程度を表す場合に用いる中央値としての基準分析値を用い、この基準分析値を、日本の文化的背景を分析に織り込むべく、社説や論文などの文献から得た分析値とするものである。逆に言えば、基準分析値の取り方により、乖離の基準となる価値観というものを設定できることになり、本実施形態では社説などから得られる文化をその基準とするが、場合によっては、企業が任意に規定したモデルケースを基準としてもよいし、現実に勤務する優秀な社員から取得したデータを基準としてもよい。なお、基準分析値は、記憶手段３１の所定の記憶領域に記憶され、乖離値生成の際に評価パラメータ生成手段１９が読み出す。
【０１４８】
ここで、基準分析値の設定にあたっては、社説など、優秀だと各ジャンルで認められる文章をサンプルとして選定し、数百から数千のサンプルを分析し、その分析値の平均を基準分析値とする。これは、文章を分析するという本質から考えると、基準分析値が、日本語という文化を反映したものであることが望ましいといえるからであり、前述したバランスという観点から、日本の文化として集束する値を求めて基準分析値とするものである。ただし、各部署における優秀なスタッフをグループサンプルとして部署毎に分析値を求め、平均値を基準分析値とした場合には、基準分析値自体が企業の社風や部署の特性などを色濃く反映したものとなる。
【０１４９】
ただし、平均値としては、例えば、処理手段３３が、複数の文章毎に生成された複数の分析値を用いて標準偏差を算出し、算出した標準偏差を用いて再度分析値をフィルタリングして抽出した分析値集団の平均値が好適である。具体的には、フィルタリングとして、分析対象となる文章毎に生成された各分析値を用いて、複数の分析値について平均値と標準偏差を算出し、その上で、（平均値）±（標準偏差）の範囲に入っていない分析値を除外し、再度、平均値を求める処理を行う。この再度求められた平均値が再平均値であり、この再平均値を基準分析値、すなわち理想値として位置づけ、文章の分析を行う際の数値基準として用いる手順を採用することがより好ましいといえる。
【０１５０】
このように、標準偏差を用いた処理を行うのは、分析対象が文章という個人差が大きくなりやすいものであるため、ばらつきの度合いが大きい標本が分析対象に混じる可能性が高いことに起因し、標準偏差を用いて画一的に除外し、外乱的な要素を排除した客観性の高い評価基準を得る構成とするものである。なお、多少の外乱的な要素を含む基準としても良い場合には、直接的に複数の分析値から得られる平均値を、上述したような数値操作を行うことなく、基準分析値としてもよい。
【０１５１】
そして、評価パラメータ生成手段１９は、上述した基準分析値と生成した分析値とに基づき、読み出した基準分析値に対する分析値の乖離の程度を示す乖離値を生成する機能を呈するが、ここで、乖離値を算出するための算出則を次式（１）に示す。この算出則は、プログラム関数やモジュールなどのかたちで、前述した分析値算出則と同様に、記憶手段３１の所定の領域に設けられた不図示の乖離値算出則記憶領域に記憶される。
【０１５２】
式（１）において、Ａを補正係数、Ｂを段階係数、Ｃを分散係数と称する。これらの係数は、記憶手段３１の乖離値算出用の係数を記憶する所定の係数記憶領域に分析値の種類毎に記憶される。
【数１】

【０１５３】
以下に、各係数の意義とともに、各係数の算出手順について参考までに説明する。まず、補正係数Ａは、分析値毎に桁数異なるため、分析値を指標として用いる際に値を揃えて普遍化しやすいように、分析値を修正する係数である。式（１）において、原則として分析値に補正係数Ａを乗じた値が、１付近を中央値とするように補正係数Ａを定める。ゆえに、この補正係数Ａは、基準分析値で１を除すという手順で得られるものであり、基準分析値に基づく乖離値の生成を行う処理の根幹をなす。例えば、総頻度比率の基準分析値で１を除した商が、総頻度比率の補正係数Ａとなる。具体的には、０．７４４５である場合、この０．７４４５で１を除した商の１．３４３１が補正係数Ａとなる。
【０１５４】
また、段階係数Ｂは、乖離値が、満点が最適という評価概念ではなく、理想値に近いと最適という評価概念に基づくものであるという性質に鑑み、乖離の程度を分析結果として見たときに、評価数値の整数部分の数で把握しやすいように、算出結果を修正する係数である。本実施形態では、段階係数Ｂを５に設定している。
【０１５５】
具体的には、本発明で用いる乖離値は、基準となる値に対して、どの程度乖離しているかという程度を表すものである。したがって、例えば０を基準とした場合を考えると、＋５であっても、−５であっても、乖離の程度は同じとなる。この場合では、０で表される「ほどほど」のところが最も良いとしたときの、その最善値からの離れ具合を示す段階を５としたものといえ、評価の内容によっては、３段階、１０段階などのほうが感覚的に分かりやすい、または、評価結果を利用しやすい場合があるので、固定値にせず、任意に採り得る係数としている。このため、段階係数Ｂは、処理手段３３による算出結果で得るのではなく、事前に所定の記憶領域に記憶される構成を採用する。
【０１５６】
そして、分散係数Ｃは、乖離値の乖離の程度を示すことになる幅を修正するものである。単に補正係数Ａをかけて分析値の桁数を揃えただけでは、数値の変化の幅が普遍化されていないので、乖離値毎の相対比較を行う場合には不適である。このために用いられるのが、式（１）における分散係数Ｃである。
【０１５７】
本実施形態では、分散係数Ｃを、基準分析値を求める際に算出可能な標準偏差として、乖離値の普遍化を図っている。具体的には、基準分析値の再平均値を算出した際のデータについて、各分析値とその再平均値を用いて算出した標準偏差を用いてもよいし、単なる平均値から求めた標準偏差を用いてもよい。
【０１５８】
ここで、式（１）の意義について見れば、処理手段３３が、乖離値を生成する際に、特に、式（１）において、この補正係数Ａと分析値の積から１を引くことにより、０を基準とした乖離の程度を示す数値とし、段階係数Ｂを乗じることにより、この乖離の程度を分析結果として見たときに、認識しやすいように修正し、この分散係数Ｃで除することにより、算出結果が標準偏差に対する比率という意義を与えて、乖離値の最大値と最小値の幅を相対比較の可能なスケールに修正することにより、評価パラメータ生成手段１９が実現され、乖離値が生成される。
【０１５９】
このような乖離値を用いて評価する手法を採用したのは、理解しやすい文章の場合には、分析値が特定の一の値に集束するという経験的に得られた傾向に基づくものである。
【０１６０】
また、乖離値の段階係数Ｂによる乖離値の変動幅の設定は、求められる評価のばらつきに応じた段階評価を踏まえて設定されるものであり、事前に設定されるものであってもよいし、分析途中で、人材活用支援システム１が入力手段３４などを介して何段階評価を行うかを受け付ける構成としてもよい。
【０１６１】
さらに、評価パラメータ生成手段１９は、生成した乖離値を、評価パラメータ記憶手段２０に記憶する機能も備える。ゆえに、評価パラメータ生成手段１９は、乖離値生成の観点からみれば、記憶手段３１に設けられた不図示の乖離値算出則記憶領域、および所定の係数記憶領域とを参照し、生成した分析値と、乖離値算出則記憶領域から読み出した前述した式（１）で表される乖離値算出則、および係数記憶領域から読み出した各係数と、に基づき、乖離値を生成する手段として機能するものであるといえる。そして、この係数に基づき乖離値を生成することにより、間接的にではあるが基準分析値に基づき乖離値を生成する手段として機能する。
【０１６２】
以下は、評価パラメータ生成手段１９が上述した乖離値を用いて、顕在能力評価値を生成する手段として機能する面について説明する。評価パラメータ生成手段１９が生成した複数の乖離値に基づき生成する顕在能力評価値の例としては、理解する力の評価要素として理解力と把握力が、分析する力の評価要素として分析力、洞察力、および問題発見力が、考える力の評価要素として創造力が、表現する力の評価要素として表現力と説得力が、応用する力の評価要素として応用力、行動力、および対応力が、それぞれ挙げられる。このように、それぞれ異なる観点からなる複数の乖離値を用いて評価要素となる顕在能力が算出される。
【０１６３】
なお、処理手段３３が実現する顕在能力評価値の算出方式としては、複数の乖離値の平均値を用いる方式や、乖離値の理想値と比較した際の、理想値と乖離値との差の絶対値をとり、この絶対値に基づいて理想値にどれくらい及ばないかを、理想値を満点とする減点方式で表す方式など、任意の算出方式を採り得る。この算出則も、記憶手段３１の所定の領域に設けられた不図示の顕在能力評価値算出則記憶領域に記憶される。
【０１６４】
言い換えれば、評価パラメータ生成手段１９は、他の手段の生成手順と同じく、詳細には、生成した乖離値と、不図示の顕在能力評価値算出則記憶領域から読み出した顕在能力評価値算出則と、に基づき、顕在能力評価値を生成する手段として機能する。
【０１６５】
具体的には、本実施形態において、理解する力の評価要素となる理解力の評価は、体言率、テーマ説明力、単語構成力、センテンス構成力、主張補足力という観点から複合的に判断される。体言率、中心補完単語比率、出現単語比率、センテンス比率、第２単語頻度比率、の５つの乖離値の平均をとって数値化する。
【０１６６】
理解する力の評価要素となる把握力の評価は、用言率、テーマ説明力、単語構成力、主張構成力、論理展開力という観点から複合的に判断される。用言比率、中心補完単語比率、出現単語比率、第３単語重複率、同頻度数比率、の５つの乖離値の平均をとって数値化する。
【０１６７】
分析する力の評価要素となる分析力の評価は、体言率、テーマ設定力、テーマ補足力、行動表現力、主張構成力という観点から複合的に判断される。体言率、中心キーワード比率、中心１頻度単語比率、センテンス用言率、第３単語重複率、の５つの乖離値の平均をとって数値化する。
【０１６８】
分析する力の評価要素となる洞察力の評価は、語彙力、テーマ説明力、センテンス主張力、主張力、主張補足力という観点から複合的に判断される。除数詞名詞比率、中心補完単語比率、センテンス頻度平均値、第１単語頻度比率、第２単語頻度比率、の５つの乖離値の平均をとって数値化する。
【０１６９】
分析する力の評価要素となる問題発見力の評価は、体言率、テーマ補足力、単語構成力、行動表現力、主張構成力という観点から複合的に判断される。体言率、中心１頻度単語比率、出現単語比率、センテンス用言率、第３単語重複率、の５つの乖離値の平均をとって数値化する。
【０１７０】
考える力の評価要素となる創造力の評価は、語彙力、用言率、テーマ設定力、主張力、論理強制力、という観点から複合的に判断される。除数詞名詞率、用言比率、中心キーワード比率、第１単語頻度比率、単語回帰値、の５つの乖離値の平均をとって数値化する。
【０１７１】
表現する力の評価要素となる表現力の評価は、語彙力、テーマ補足力、単語構成力、センテンス構成力、主張力、という観点から複合的に判断される。そして、除数詞名詞率、中心１頻度単語比率、出現単語比率、センテンス比率、第１単語頻度比率、の５つの乖離値の平均をとって数値化する。
【０１７２】
表現する力の評価要素となる説得力の評価は、語彙力、テーマ設定力、テーマ補足力、主張力、論理強制力、という観点から複合的に判断される。除数詞名詞率、中心キーワード比率、中心１頻度単語比率、第１単語頻度比率、単語回帰値、の５つの乖離値の平均をとって数値化する。
【０１７３】
応用する力の評価要素となる応用力の評価は、語彙力、テーマ補足力、単語構成力、主張構成力、論理展開力という観点から複合的に判断される。除数詞名詞率、中心１頻度単語比率、出現単語比率、第３単語重複率、同頻度数比率、の５つの乖離値の平均をとって数値化する。
【０１７４】
応用する力の評価要素となる行動力の評価は、用言率、センテンス主張力、主張力、論理展開力、論理強制力という観点から複合的に判断される。用言比率、センテンス頻度平均値、第１単語頻度比率、同頻度数比率、単語回帰値、の５つの乖離値の平均をとって数値化する。
【０１７５】
応用する力の評価要素となる対応力の評価は、テーマ設定力、テーマ説明力、テーマ補足力、行動表現力、主張力という観点から複合的に判断される。中心キーワード比率、中心補完単語比率、中心１頻度単語比率、センテンス用言率、第１単語頻度比率、の５つの乖離値の平均をとって数値化する。
【０１７６】
ただし、評価要素となる力は上述した各力に限られるものではなく、例えば、思考力の評価は、語彙力、テーマ設定力、単語構成力、主張力、論理展開力という観点から複合的に判断することができる。この場合、除数詞名詞率、中心キーワード比率、出現単語比率、第１単語頻度比率、同頻度数比率、の５つの乖離値の平均をとって数値化する。
【０１７７】
そして、評価パラメータ記憶手段２０は、以上説明した乖離値や顕在能力評価値などの評価パラメータを記憶する手段であり、後述する職務適応性判定手段２２が判定する対象となる評価パラメータを記憶するとともに、情報出力手段２４が外部へ出力する出力情報としての評価パラメータを記憶する機能を果たす。図７に、評価パラメータ記憶手段２０のデータベース構造の一例を示す。個人識別情報と、その個人に対応する乖離値、顕在能力評価値を関連付けて記憶するデータ構造を採用するが、必ずしも、図７に示した形態に限られるものではなく、乖離値と顕在能力評価値を独立して記憶する構造を採用してもよい。
【０１７８】
他方、比較パラメータ記憶手段２１は、前述した評価パラメータ生成手段１９が生成した評価パラメータの比較基準となる比較パラメータを記憶する手段であり、詳細には、評価パラメータに応じて付与される複数の類型パターンのうち最も好適な職務に必要な職能を十分に備えるパターンを比較パラメータとして備えて、各個人の評価パラメータを評価する基準を設定する機能を果たす。比較パラメータ記憶手段２１のデータベース構造の一例を図８に示す。
【０１７９】
ここで、詳細は詳述するが、比較基準は、回答文の評価に際し、複数の優秀な人材を書き手として得られたパターン指標値を平均して得られた指標値から得られる基準指標値を用いる場合、あるいは、一人の優秀な人材を書き手として得られたパターン指標値から得られる基準指標値を用いる場合などがある。比較パラメータ記憶手段２１は、原則として、これらの基準指標値を事前に比較パラメータ記憶手段２１に記憶する。
【０１８０】
また、本実施形態においては、回答文も文章であり、厳密に小数点以下の細かい桁までの一致を判断しても、文章比較として適切な結果を得られないことに鑑み、日本語特有の表現の多様性に起因する数値のばらつき、いうなれば文章の表現の幅からくる「ゆらぎ」を加味して比較できるように乖離値の相対比較により得られるパターンを、比較・評価する基準としている。このため、必要とされる能力を備える場合に、その論述した文章が示す各乖離値の理想的なパターンを意味する指標値を基準指標値として位置づけており、さらには、乖離値をその意義に基づきグループ化した場合のパターンを示すパターン指標値を用いる構成としている。また、相対比較を取り入れることにより、表現技術の優劣に、他の能力が埋没しないように担保することもできる。
【０１８１】
具体的には、表現系に分類される体言率、用言率、付属語率からなる第１ブロック、テーマ系に分類される中心キーワード比率（テーマ設定力）、中心補完単語比率（テーマ説明力）、中心１頻度単語比率（テーマ補足力）からなる第２ブロック、単語構成系に分類される総頻度比率（文章整備力）、出現単語比率（単語構成力）からなる第３ブロック、センテンス系に分類されるセンテンス用言率（行動表現力）、センテンス頻度平均値（センテンス主張力）、センテンス比率（センテンス構成力）からなる第４ブロック、主張系に分類される第１単語頻度比率（主張力）、第２単語頻度比率（主張補足力）、第３単語重複率（主張構成力）、同頻度数比率（論理展開力）、単語回帰値（論理強制力）からなる第５ブロック、に分けて、ブロック毎のパターンを比較する手法を採用し、このブロックに応じたかたちの比較パラメータを比較パラメータ記憶手段２１に記憶する。
【０１８２】
ここで、上述した比較パラメータは、図８に示したように、各比較所属先毎に原則として異なるパターン指標値を持ち、独自のパターンを示すことになるのが一般的であるが、基準分析値を社説などのデータに基づく分析値から得たものに設定せずに、理想とする社員やモデルケースなどを直接基準分析値とした場合には、図８とは異なり、各ブロックの基準指標値は、ばらつきのない一様な状態を呈することになる。
【０１８３】
そして、本実施形態における職務適応性判定手段２２は、評価パラメータ生成手段１９が生成した評価パラメータと、比較パラメータ記憶手段２１が記憶する比較パラメータとを比較する手段であり、詳細には、評価パラメータおよび比較パラメータに応じた類型パターンを用いて、比較を行う手段として機能する。
【０１８４】
ここで、類型パターンとは、評価パラメータ生成手段１９が生成した複数の乖離値同士の相対関係を比較して得られるパターンであり、本実施形態では特に、ブロック毎に基準指標値を設定したブロック単位での相対関係を表すパターンを用いる。パターンの種類は、形ではなくパターン指標値を用いた数値により表す形態を採用し、このパターン指標値により、表現の多様性を考慮し、乖離値の相対関係を比較する。パターン指標値を生成するので、絶対的な数値比較よりも、より文章の持つあいまいさを加味した上で、文章のバランスについて的確な比較を行うことができる。
【０１８５】
詳細には、職務適応性判定手段２２は、一の乖離値を基準として他の乖離値を相対比較した結果に基づき、各個人の評価パラメータとしての乖離値群を用いて、乖離値相互の相対関係を表すパターン指標値を生成する手段として機能するものであり、相対比較の一手法として、減算処理を用いる。
【０１８６】
以下、減算処理を用いる具体的態様を説明する。職務適応性判定手段２２は、相対関係として、一の乖離値を基準として他の乖離値との差を用い、差の値に基づき各乖離値に対応するパターン指標値を生成するという機能を果たすが、差の値の分類パターンは、０以上と０未満の２種類、正の値、０、負の値の３種類、所定のしきい値により正の値と負の値を分け、それに０を加えた５種類とするなどのパターンを採り得る。種類数を増やすにつれて精密な比較が可能となるが、相対差の大きさ（パターンレベル）を加味したい場合には５種類以上を選択することが望ましいといえる。
【０１８７】
例えば、段階係数を５とした場合に５種類のパターン比較を行う場合には、しきい値を±１として、パターン指標値を＋２、＋１、０、−１、−２の５種類とする。各ブロックに振り分けられた第１番目の乖離値を基準として、相対比較しパターン化するのであるが、第１番目の乖離値の相対比較した指標値は０となる。そして、第２番目、第３番目と順次乖離値の相対比較を行うが、±１は基準値（第１番目の乖離値）よりその該当する乖離値との差が±１以内であるときに指標値として付与され、±２は基準値（第１番目の乖離値）より、その該当する乖離値との差が１より大きく差がある場合に指標値として付与される。
【０１８８】
したがって、パターン指標値の種類数が２種類の場合は、５の（２−１）乗のパターン指標値の組合せが存在し、パターン指標値の種類数が３種類の場合は、５の（３−１）乗のパターン指標値の組合せが存在し、上述したようにパターン指標値の種類数が５種類の場合は、５の（５−１）乗のパターン指標値の組合せが存在する。
【０１８９】
そして、この職務適応性判定手段２２は、生成したパターン指標値と、比較パラメータ記憶手段２１から読み出した基準指標値とを比較する機能を果たすが、前述した第１から第５までのブロックについて比較する場合には、全体では、２５×２５×５×２５×６２５＝４８８２５１２５のパターンの比較を行うことになる。ただし、比較の手法としては、必ずしもすべてのパターン指標値を比較する必要はなく、職務適応性を限定的に解釈し、適職性として判断する場合などには、第５ブロックの６２５通りのみを検証するようにしてもよい。
【０１９０】
このように、職務適応性判定手段２２は、パターンブロック単位の第１番目の乖離値を用いて、各乖離値をパターン指標値に変換し、相対比較を行うことにより、パターン認識に擬した比較判断を行い、回答文の書き手の能力と基準指標値が示す比較対象とを客観的に比較して、職務適応性判定結果記憶手段２３に記憶する。
【０１９１】
一の乖離値を基準として相対関係をみることにより、乖離値の相対関係がより明瞭となるパラメータの普遍化ができ、また、乖離値の絶対値の大小を比較せず、相対比較する手法を採用することにより、比較パターン数を少なくすることにより、マッチング処理の高速化が図れる。
【０１９２】
ここで、職務適応性判定手段２２は、職務適応性判定結果として、パターン指標値の数値を比較し、同じパターンとなっているか否かを比較判断した結果を取得するが、一致する場合には、比較パラメータ記憶手段２１に記憶した基準指標値が意味するところの必要とされる職能を備えることになり、一致しない場合には、その能力を備えていないということになる。
【０１９３】
なお、単に、一致するか否かを結果としてもよいが、一致しなかった部分と一致した部分とを分けた詳細な職務適応性判定結果とすることにより、その人物に備わった能力と欠けた能力とを客観的に示すことができる。そのような形態の職務適応性判定結果を記憶する構成を採用した場合には、単なる人材活用が図られているかを判断するための職能比較だけでなく、どこの能力を開発すると、最適な人材となるか、というコンピテンシー判断機能を兼ね備えた職務適応性判定手段２２としても機能する。
【０１９４】
また、職務適応性判定手段２２は、個人の評価パラメータと基準となる比較パラメータの比較に際し、個人の現在の所属と同じ所属について規定された比較パラメータを比較する態様だけでなく、前述した所属適合判定結果記憶手段１８が記憶する所属適合判定結果に基づき得られる、その個人の回答情報が示唆する現在の所属とは異なる所属についての比較パラメータと比較するべく、所属適合判定結果記憶手段１８が記憶する所属適合判定結果を参照して職務適応性を判断する機能を備えるようにしてもよい。
【０１９５】
以上説明した職務適応性判定手段２２が生成した職務適応性判定結果を記憶する手段が、職務適応性判定結果記憶手段２３であり、個人識別情報と関連付けて職務適応性判定結果を記憶する。具体的には、職務適応性判定結果記憶手段２３は、図９に示したようなデータベース構造を採用し、個人識別情報と関連付けて、上述した現在の所属についての比較パラメータとの適合性判定結果のみならず、好適所属についての比較パラメータとの適合性判定結果なども記憶する。図９において、パターン指標値の比較を行った場合の判定の内容は、すべての項目が一致したという結果を「○」で、一項目が一致しなかったという結果を「△」で、二項目以上が一致しなかったという結果を「×」で表されている。ただし、図９に示したデータ構造に限られるものではなく、比較対象となる所属毎に独立して所属適応性判定結果を記憶する構成としてもよい。
【０１９６】
そして、情報出力手段２４は、処理手段３３の制御下において、所属適合判定手段１７が判定した所属適合判定結果を所属適合判定結果記憶手段１８から読み出して出力する手段として、また、職務適応性判定手段２２が行った職務適応性判定結果を、職務適応性判定結果記憶手段２３から読み出して出力する手段として機能する。加えて、評価パラメータ生成手段１９が生成した乖離値や顕在能力評価値などの評価パラメータを評価パラメータ記憶手段２０から読み出して出力する手段として機能する。
【０１９７】
具体的には、本実施形態における情報出力手段２４は、記憶手段３１に設けられた所属適合判定結果記憶手段１８、職務適応性判定結果記憶手段２３、評価パラメータ記憶手段２０の所定の記憶領域に記憶された電子情報を、通信手段３２やネットワーク２を介してＬＡＮ３側へ出力する手段として機能する。また、出力に際し、情報出力手段２４は、出力先に応じた形態の出力情報の生成を必要に応じて行う。
【０１９８】
例えば、ＬＡＮ３に接続された端末装置４の表示手段３５に対して、所定の情報を出力する場合には、表示手段３５においてどのように表示されるかを規定する表示制御情報に乖離値または評価値を組み込んで出力する。また、ＬＡＮ３に接続された印刷手段５に対して、所定の情報を出力する場合には、印刷手段５がどのように情報を印刷するかを規定する印刷制御情報に乖離値または評価値を組み込んで出力する。
【０１９９】
なお、表示手段３５や印刷手段５に出力する際に、情報出力手段２４は、乖離値の中央値（評価基準軸値）を０から５に修正するなどの出力情報の調整処理を行ってもよい。これは、各乖離値により評価された表３に示した分析項目内容を視覚的に把握しやすいように、評価基準軸値を全体的に底上げして、すべての評価数値が原則として０以上になるように調整を行うものである。
【０２００】
なお、要求される出力形態に応じて、そのまま評価基準軸値を修正せずに乖離値を出力してもよいし、さらには、乖離値などに基づき、グラフ形式や表形式などとなるような出力情報を生成して出力してもよい。
【０２０１】
以上説明した各手段が有機的に結合し、機能することにより、多様な職務に応じたかたちで、任意の比較パラメータを基準とした多様な判定を実現でき、さらには、厳密な数値比較を行わずに職能比較が可能となり、ひいては、人間の能力というあいまいな部分を好適に比較できる人材活用支援システムを実現できる。
【０２０２】
特に、客観的な個人分析用回答情報に基づくパラメータを判断材料とする人材職能比較をすることができ、文を構成する単語のバランスという観点から個人の職能を客観的に判断できる人材活用支援システムを実現できる。
【０２０３】
［個人分析用回答情報分析処理の手順］
次に、人材活用支援システム１の個人分析用回答情報分析処理の手順につき、図１０に基づき説明する。図１０は、本実施形態にかかる個人分析用回答情報分析処理の主要手順を示すフローチャートである。以下に説明する個人分析用回答情報分析処理は、前述した回答単語などを用いて所属適合を評価する所属適合判定処理と、それに引き続き行われる職務適応性判定処理に大別される。
【０２０４】
［所属適合判定処理］
まず、所属適合判定処理について図１０に基づき説明する。ここで、本実施形態にかかる所属適合判定処理を行う人材活用支援システム１は、所属適合判定処理を行う際の前提として、上述したように、情報の受付手段、情報の記憶手段３１、および各手段の動作を制御する処理手段３３を備え、記憶手段３１は、回答文を単語に分解する際の基準となる基準単位単語を所定の分析区分に関連付けて記憶する分析用辞書記憶手段１３、各所属に必要とされる職務と関連付けられる識別単語を規定した職務識別単語記憶手段１６、回答単語の出現頻度の履歴を記憶する分析結果記憶手段１５、および、個人分析用評価パラメータの比較基準となる比較パラメータを記憶する比較パラメータ記憶手段２１を備える構成を採用する。なお、ここでいう受付手段は、通信手段３２などのインターフェースを想定している。
【０２０５】
そして、この人材活用支援システム１における所属適合判定処理および後述する職務適応性判定処理を含む個人分析用回答情報分析処理は、記憶手段３１の所定の領域に格納された人材活用支援プログラムにしたがって行われる。その人材活用支援プログラムは、処理手段３３に、現在の所属と所定の質問に対する回答文を含む個人分析用回答情報を受け付けるステップ（Ｓ１）と、当該回答文が有する回答単語を分析し、回答単語の出現頻度を当該回答単語に関連付けるステップ（Ｓ２）と、職務識別単語記憶手段１６を参照し、所定の出現頻度以上の回答単語に対応する識別単語に基づき得られた所属と現在の所属とを比較し、一致するかを判定するステップ（Ｓ３）と、当該所属を比較した判定結果を出力するステップ（Ｓ７）と、を実行させる。
【０２０６】
これらの各ステップを実行させることにより、第三者の恣意が入り込むことのない客観的な判断材料に基づく判定結果を得ることが可能となり、ひいては、企業の人材活用の一助となる資料を提供できる人材活用支援環境を提供するものである。
【０２０７】
より詳細には、図１０に示したように、まず、処理手段３３に、個人分析用回答情報を受け付けるステップ（Ｓ１）を実行させる。このとき、ＬＡＮ３側からネットワーク２および不図示のルータなどを介して個人分析用回答情報を受け付けることになるが、外部とのインタフェースである通信手段３２が受付手段となり、処理手段３３が受付手段を介して受け付けた個人分析用回答情報を、個人分析用回答情報記憶手段１２の所定の領域に格納する。
【０２０８】
そして、上述したように、処理手段３３に、受付手段を介して個人分析用回答情報を受け付け、受け付けた個人分析用回答情報を個人分析用回答情報記憶手段１２の所定の領域に格納することを実行させることにより、前述した個人分析用回答情報受付手段１１が実現され、機能する。
【０２０９】
続いて、図１０に示したように、処理手段３３に、分析用辞書記憶手段１３の体言辞書記憶領域、用言辞書記憶領域、接続詞辞書記憶領域、助詞辞書記憶領域、助動詞辞書記憶領域、および活用辞書記憶領域などを参照して、当該回答文が有する回答単語を分析し、回答単語の出現頻度を当該回答単語に関連付けるとともに、回答単語と関連付けられた出現頻度とを分析結果記憶手段１５に記憶するステップ（Ｓ２）を実行させる。
【０２１０】
具体的には、処理手段３３は、分析用辞書記憶手段１３の表記辞書記憶領域と削除記号辞書記憶領域とを参照し、記号変換や記号削除などを施して、単語への分解に適した回答文に加工した後、１センテンス単位で回答文の頭から所定の長さの文字列に分解し、分解された順番に記憶手段３１の所定の領域に一時的に格納する。
【０２１１】
そして、処理手段３３は、分析用辞書記憶手段１３が記憶する基準単位単語を参照して、回答単語を特定し、さらに、次の文字列の分析へと移行する。ここで、処理手段３３は、分解に際し活用辞書記憶領域を参照するが、この参照により、用言を終止形に置き換えるなどの処理を行う。
【０２１２】
この後、処理手段３３は、回答文から抽出した回答単語の数を単語の種類別にカウントし、カウントした値を出現頻度として回答単語に関連付けて分析結果記憶手段１５に記憶する。
【０２１３】
また、本実施形態では、図３に示したように、分析用辞書記憶手段１３が、基準単位単語の属性に応じた分析区分を基準単位単語に関連付けて記憶する構成を採用し、処理手段３３に、上述したような回答単語を特定し回答文を複数の回答単語に分解する処理をさせるとともに、後述する職務適応性判定処理で用いるべく、回答単語への分析区分の関連付け処理として、基準単位単語に付与されている分析区分をそのまま分解した回答単語に付与する処理を実行させる。
【０２１４】
これにより、その回答単語に付与した分析区分を、分解した回答単語とともに分析結果記憶手段１５に記憶できるので、回答文を回答単語に分解して出現頻度と関連付けるステップの実行に連動して、分析用辞書記憶手段１３を参照して回答単語と当該分析区分とを関連付けるステップを実行させることができる。なお、分析区分の付与は、表１および表２に示した分析区分コードを付与する方式を採用する。
【０２１５】
このように、処理手段３３に、分析用辞書を参照しながら回答文が有する回答単語を分析して特定・分解し、特定・分解した回答単語を分析結果記憶手段１５の所定の領域に格納することを実行させることにより、前述した回答情報分析手段１４が実現され、機能する。
【０２１６】
続いて、処理手段３３に、現在の所属と回答単語に基づき得られる好適所属とを比較・判定させるステップ（Ｓ３）を実行させる。具体的には、分析結果記憶手段１５に記憶された回答単語とその出現頻度とに基づき、所定の出現頻度以上の回答単語を抽出し、抽出した回答単語に対応する識別単語に基づき職務識別単語記憶手段１６を参照して関連付けられた所属を好適所属として取得する。ここで、好適所属の特定にあたっては、回答単語に対応する識別単語群と、所属に必要とされる職務と関連付けられる識別単語群との一致を見ることになるが、一致するか否かの判定は、前述したように、完全一致、もしくは所定の割合以上の部分一致をもって一致するものと判断し、その一致するものと判断した所属を、個人に適している好適所属として取得する。
【０２１７】
なお、回答文から抽出される回答単語のうち、前述したように約８０％ぐらいは出現頻度が１回（頻度数が１）であるので、フィルタリングの条件として設定される所定の出現頻度は、少なくとも２回以上とされる。ただし、ノイズ情報の除去という観点からは、３回以上とするのが好適である。
【０２１８】
加えて、処理手段３３に、単なる出現頻度だけでなく、分析結果記憶手段１５に記憶された出現頻度の履歴に基づき回答単語を抽出する場合には、履歴に基づき抽出された所定の時間サイクル以上で用いられる回答単語に対応する識別単語に基づき得られた所属を比較するステップを実行させる。
【０２１９】
具体的には、図４に示したような回答単語に関連付けられた情報受付日などの時間記録情報に基づき、判定に用いる回答単語を特定・抽出する。そして、処理手段３３は、抽出された所定の時間サイクル以上で用いられる回答単語に対応する識別単語に基づき、上述した手順を用いて好適所属を取得し、好適所属と現在の所属とを比較するステップを実行する。
【０２２０】
そして、処理手段３３は、得られた好適所属と現在の所属とを比較し、例えば、その所属に固有の所属コードなどに基づき、一致するかを判定し、その判定した比較結果を所属適合判定結果として所属適合判定結果記憶手段１８に記憶する（Ｓ３）。
【０２２１】
上述した所属適合判定処理の後、処理手段３３は、職務適応性判定処理を行うか否かの指示を受け付ける判断ステップを実行する（Ｓ４）。処理手段３３は、例えば、ＬＡＮ３、ネットワーク２などを介して、端末装置４が備える入力手段３４から入力された上記指示を受け付ける。ここで、個人の職能に基づく所属に必要とされる職務との適応性を評価する旨の指示を受け付けた場合には、職務適応性を判定する一連のステップ（Ｓ５、Ｓ６）へと移行する。一方、職務適応性判定処理を行わない旨の指示を受け付けた場合には、出力ステップ（Ｓ７）へと移行する。
【０２２２】
なお、この判断ステップ（Ｓ４）は、例えば、処理手段３３が生成した表示制御情報に基づき、端末装置４が備える表示手段３５に形成した質問画面（指示受付インターフェース）を介して、入力手段３４から指示命令を受け付けるなどの処理により実現される構成としてもよい。
【０２２３】
［職務適応性判定処理］
以下、職務適応性判定処理について説明する。ここで、前述したように、職務適応性判定処理は、記憶手段３１の所定の領域に格納された人材活用支援プログラムにしたがって行われるものであり、処理手段３３に、職務適応性を判定する一連のステップ（Ｓ５、Ｓ６）として、前記回答単語の出現頻度に基づく個人分析用評価パラメータを求めるステップと、生成した評価パラメータと、前記比較パラメータ記憶手段が記憶する比較パラメータとを比較するステップと、当該評価パラメータと比較パラメータとを比較した判定結果を出力するステップと、を実行させる。これらのステップの実行により、多様な職務に応じたかたちで、任意の比較パラメータを基準とした多様な判定を実現できる。
【０２２４】
より具体的には、評価パラメータの生成ステップ（Ｓ５）として、前述した乖離値の生成に用いるカウント値の取得、カウント値に基づく分析値の生成、および分析値に基づく乖離値の生成という段階的シーケンスと実行する。
【０２２５】
カウント値の取得シーケンスとして、処理手段３３は、分析結果記憶手段１５に格納されている回答文を分解して得られた複数の回答単語とその出現頻度を参照して、少なくとも回答単語にかかわる数を回答文に関連するカウント値として取得する。この際に取得、格納されるカウント値としては、前述した総頻度数や体言数、同頻度数などが挙げられる。
【０２２６】
また、処理手段３３は、個人分析用回答情報記憶手段１２に格納された回答文の全体の文字数や、回答文を構成するセンテンス数（自然文数）など、回答単語に基づくカウント値以外の分析値を生成するのに必要なカウント値の取得処理もあわせて実行し、上述した回答単語にかかわるカウント値とともに記憶手段３１の所定の記憶領域に記憶する。
【０２２７】
続いて、カウント値に基づき複数の分析値を生成するシーケンスとして、処理手段３３は、取得したカウント値を所定の記憶領域から読み出すとともに、前述した表３に示した分析値算出則を記憶手段３１に設けられた所定の分析値算出則記憶領域から読み出し、読み出した分析値算出則に基づきカウント値を用いて分析値を生成する。分析値の生成に際しては、複数の分析項目について複数種類の分析値を算出することになるので、結果として、処理手段３３は、カウント値に基づき複数の分析値を複数種類生成させることになる。
【０２２８】
より詳細には、処理手段３３は、分析結果記憶手段１５に記憶された分解された回答単語と、その単語と分析区分との関連付けを参照し、分析区分に属する単語をカウントしたカウント値に基づく複数の分析値を生成する。この際に生成される分析値として、センテンス用言率などが挙げられる。
【０２２９】
また、必要に応じて、処理手段３３が生成したカウント値から生成された二次的な数値もしくは他の分析値を用いて、処理手段３３は、前述した分析値算出則記憶領域などから読み出した分析値算出則にしたがって、分析値を生成する。この際に生成される分析値としては、体言率や用言率などが挙げられる。
【０２３０】
なお、キーワード認定に関する分析値の生成処理に際しては、再度、カウント値を取得するなどの処理を行う。具体的には、処理手段３３は、キーワード比率を生成するために、カウント値として得た同頻度数に基づき、単語の同頻度件数などを取得し、同頻度比率を取得する。このとき、同頻度比率が０．７以下の単語がキーワードと認定される。また、同じく、キーワードと認定された単語群の中に含まれる感想単語の数を取得するためにも、処理手段３３は、カウント処理を行う。この後、キーワード数などのカウント値に基づき、処理手段３３は、中心キーワード数などの分析値を生成する。
【０２３１】
そして、処理手段３３は、生成した分析値を記憶手段３１の所定の記憶領域に格納する。
【０２３２】
次に、基準分析値に対する分析値の乖離の程度を示す乖離値を生成するシーケンスとして、処理手段３３は、記憶手段３１の所定の記憶領域から基準分析値とともに、分析値の種類毎に複数の分析値を読み出し、読み出した分析値と基準分析値とに基づき、記憶手段３１に設けられた所定の乖離値算出則記憶領域に格納された乖離値算出則を用いて、基準分析値に対する乖離の程度を示す複数の乖離値を生成する。
【０２３３】
具体的には、処理手段３３に、前述した式（１）を記憶手段３１の所定の領域から読み出し、分析値の種類毎に対応する基準値から得られた補正係数などの各係数を記憶手段３１の所定の記憶領域から読み出し、読み出した係数と式（１）に基づいて各分析値について乖離値を生成するステップを実行させる。そして、処理手段３３は、生成した乖離値を評価パラメータ記憶手段２０に記憶する。
【０２３４】
なお、顕在能力評価値を生成する場合には、処理手段３３は、評価パラメータ記憶手段２０から乖離値を読み出し、さらには、前述したように、所定の領域に格納された評価値の算出則にしたがい、読み出した複数の乖離値に基づき顕在能力評価値を生成し、生成した顕在能力評価値を個人の評価パラメータとして評価パラメータ記憶手段２０に記憶する。
【０２３５】
以上が、評価パラメータを乖離値として場合の、分析区分に属する前記回答単語の出現頻度に基づき得られる数値を用いて評価パラメータを生成するステップおよび生成した評価パラメータを記憶するステップ（Ｓ５）の詳細である。
【０２３６】
このように、処理手段３３に、回答単語の出現頻度や回答文の文字数など、回答文に基づく数値を用いて個人分析用評価パラメータを求める処理を実行させることにより、前述した評価パラメータ生成手段１９がが実現され、機能する。
【０２３７】
続いて、処理手段３３に、生成した評価パラメータと、前記比較パラメータ記憶手段が記憶する比較パラメータとを比較するステップ（Ｓ６）を実行させるが、本実施形態では、比較パラメータ記憶手段２１から読み出した基準指標値に基づき、生成した複数の乖離値同士の相対関係を比較する手法を採用し、評価パラメータおよび比較パラメータに応じた類型パターンを用いて、比較を行うステップを実行させる
【０２３８】
具体的には、処理手段３３に、乖離値相互の相対関係を表すパターン指標値を、一の乖離値を基準として他の乖離値を相対比較した相対関係として、一の乖離値を基準として他の乖離値との差を算出し、その差の値に基づき各乖離値に対応するパターン指標値を生成するステップを実行させる。
【０２３９】
本実施形態でのパターン指標値生成においては、乖離値の厳密な差の数値結果を用いることなく、全体的な分析パターンの比較という思想を実現するために、乖離値の数値ではなく、乖離値同士の相対関係を表すパターン指標値に置き換える処理をおこなう。この処理により、より文章の持つあいまいさを加味した上で、文章のバランスについて的確な比較を行うことができる。
【０２４０】
したがって、パターン全体の比較を行う場合には、好適なパターン指標値として、−１、０、＋１という３種類のパターン指標値を用い、パターンの特性をより反映させた比較を行う場合には、図８に示したように、パターンレベルを考慮したパターン指標値として、前述したように乖離値の生成における段階係数を５にしたことから、±１をしきい値として用い、−２、−１、０、１、２という５種類のパターン指標値を用いる。
【０２４１】
ただし、前述したように、処理手段３３は、すべての乖離値について相対関係を取得する必要はなく、能力判定に必要な乖離値相互の関係をパターン指標値として生成・取得すればよい。
【０２４２】
また、パターン指標値を比較するステップとして、処理手段３３は、比較パラメータ記憶手段２１から、各所属の理想パターンとしての比較パラメータを示す基準指標値を読み出し、前述した回答文毎に生成された各個人の職能を推定できる顕在能力の各要素となる乖離値から得られるパターン指標値との比較を行う。
【０２４３】
すなわち、処理手段３３が、個人の評価パラメータから得られるパターン指標値を比較することにより、比較対象となる所属において望まれる能力をすべて備えるのか、それとも、その一部が足りないのか、という客観的な職務適用性判定結果を得ることができる。具体的には、本実施形態においては、図９に示したように、ブロックを構成する要素がすべて一致していると職務適応性が高く、一致していない要素がある場合には、配置転換や、研修による顕在能力の向上を図る必要がある、職務適応性が低いということになる、ブロック毎の判定という形態を採る。しかし、これに限られるものではなく、必要に応じて、図８に示したようなパターン指標値の各項目についての一致不一致を判定し、詳細に、どの能力が不足しているかを、職務適応性判定結果の追加情報として、提供する構成としてもよい。
【０２４４】
なお、処理手段３３は、上述した内容の職務適応性判定結果を職務適応性判定結果記憶手段２３に記憶する。
【０２４５】
このように、処理手段３３に、乖離値相互の相対関係を表すパターン指標値を、一の乖離値を基準として他の乖離値を相対比較した結果に基づき生成することを実行させるとともに、生成したパターン指標値と、比較パラメータ記憶手段２１から読み出した基準指標値とを比較することを実行させることにより、前述した職務適応性判定手段２２が実現され、機能する。
【０２４６】
［出力処理］
最後に、判定結果の出力ステップ（Ｓ７）について説明する。図１０に示したように、この出力ステップ（Ｓ７）へ移行してくる場合には、所属適合性判定結果の出力を目的とする場合のみならず、職務適応性判定結果の出力を目的とする場合がある。
【０２４７】
具体的には、処理手段３３に、好適所属と現在の所属とを比較して得られた所属適合判定結果を所属適合判定結果記憶手段１８から読み出し、読み出した判定結果を外部へ出力するステップ、もしくは、評価パラメータと比較パラメータとを比較して得られた職務適応性判定結果を職務適応性判定結果記憶手段２３から読み出し、読み出した判定結果を外部へ出力するステップを実行させる（Ｓ７）。
【０２４８】
ここで、処理手段３３が判定結果を出力する先である外部は、ＬＡＮ３を構成するサーバ装置６、端末装置４の表示手段３５、あるいは印刷手段５などが想定される。したがって、処理手段３３は、出力先に応じた情報形態、例えば、印刷制御情報や表示制御情報に判定結果を含める加工などを行った上で、出力要求に応じた出力を行う。なお、必要に応じて、評価パラメータ記憶手段２０に記憶された乖離値や顕在能力評価値を判定結果に付帯させて出力を行ってもよい。
【０２４９】
このように、処理手段３３に、所定の出力先へ要求された情報を出力させることにより、前述した情報出力手段２４が実現され、機能することになる。
【０２５０】
以上説明した個人分析用回答情報分析処理は、記憶手段３１の所定の記憶領域に格納された人材活用支援プログラムにしたがって実行され、分析対象となる個人分析用回答情報を分析、評価するものであり、言い換えれば、これらの処理が処理手段３３により実行されることにより、客観的な評価を行うことができる、数値化されたパラメータを用いて評価を可能とする人材活用支援システム１を実現するものである。
【０２５１】
【発明の効果】
本発明は、以上のように構成され、機能するので、請求項１に記載の発明では、回帰手法を応用して第三者の恣意が入り込むことのない客観的な判断材料に基づく判定結果を得ることが可能となり、ひいては、客観的に所属の適合性、配置転換必要性を判断するための資料を提供できるので、企業の人材活用の一助となる人材活用支援システムを実現できる。
【０２５２】
また、請求項２に記載の発明では、個人が頻繁に活用している職務に基づき判定することができ、より個人の特質に沿った判定が可能となる。
【０２５３】
また、請求項３に記載の発明では、多様な職務に応じたかたちで、任意の比較パラメータを基準とした多様な判定を実現できる。
【０２５４】
また、請求項４に記載の発明では、厳密な数値比較を行わずに職能比較が可能となり、ひいては、人間の能力というあいまいな部分を好適に比較できる人材活用支援システムを実現できる。
【０２５５】
また、請求項５に記載の発明では、客観的な個人分析用回答情報に基づくパラメータを判断材料とすることができ、ひいては、文を構成する単語のバランスという観点から個人の職能を客観的に判断できる人材活用支援システムを実現できる。
【０２５６】
また、請求項６に記載の発明では、第三者の恣意が入り込むことのない客観的な判断材料に基づく判定結果を得ることが可能となり、ひいては、企業の人材活用の一助となる資料を提供できる人材活用支援環境を提供できる。
【０２５７】
また、請求項７に記載の発明では、個人が頻繁に活用している職務に基づき判定することができ、より個人の特質に沿った判定が可能となる。
【０２５８】
また、請求項８に記載の発明では、多様な職務に応じたかたちで、任意の比較パラメータを基準とした多様な判定を実現できる。
【０２５９】
また、請求項９に記載の発明では、厳密な数値比較を行わずに職能比較が可能となり、ひいては、人間の能力というあいまいな部分を好適に比較できる人材活用支援環境を提供できる。
【０２６０】
また、請求項１０に記載の発明では、客観的な個人分析用回答情報に基づくパラメータを判断材料とすることができ、ひいては、文を構成する単語のバランスという観点から、個人の職能を客観的に判断できる人材活用支援環境を提供できる。
【図面の簡単な説明】
【図１】本発明に係る一実施形態の機能的構成を示すブロック図である。
【図２】本発明に係る一実施形態の概略的なシステム構成を示すシステム構成図である。
【図３】図１および図２に示した分析用辞書記憶手段のデータ構造の一例を示す模式的構造図である。
【図４】図１に示した分析結果記憶手段のデータ構造の一例を示すデータ構造図である。
【図５】図１に示した職務識別単語記憶手段のデータ構造の一例を示すデータ構造図である。
【図６】図１に示した所属適合判定記憶手段のデータ構造の一例を示すデータ構造図である。
【図７】図１に示した評価パラメータ記憶手段のデータ構造の一例を示すデータ構造図である。
【図８】図１に示した比較パラメータ記憶手段のデータ構造の一例を示すデータ構造図である。
【図９】図１に示した職務適応性判定結果記憶手段のデータ構造の一例を示すデータ構造図である。
【図１０】本発明にかかる情報処理手順の一例を示すフローチャートである。
【符号の説明】
１　　人材活用支援システム
２　　ネットワーク
３　　ＬＡＮ
４　　端末装置
５　　印刷手段
６　　サーバ装置
１１　個人分析用回答情報受付手段
１２　個人分析用回答情報記憶手段
１３　分析用辞書記憶手段
１４　回答情報分析手段
１５　分析結果記憶手段
１６　職務識別単語記憶手段
１７　所属適合判定手段
１８　所属適合判定結果記憶手段
１９　評価パラメータ生成手段
２０　評価パラメータ記憶手段
２１　比較パラメータ記憶手段
２２　職務適応性判定手段
２３　職務適応性判定結果記憶手段
２４　情報出力手段
３１　記憶手段
３２　通信手段
３３　処理手段
３４　入力手段
３５　表示手段
３６　記憶手段
３７　通信手段
３８　処理手段[0001]
TECHNICAL FIELD OF THE INVENTION
The present invention relates to a system and a program for supporting effective utilization of human resources.
[0002]
[Prior art]
In the past, in the evaluation of the abilities of individual jobs in companies, the subjective evaluation of HR personnel and senior management has been a major weight, referring to the results of careers, achievements, promotion tests and interviews etc. according to the type of job. Occupy. As a system for supporting these evaluations, there is a human resource finding system as disclosed in Japanese Patent Application Laid-Open No. 2002-41726, and a system that manages personal evaluations as electronic information, which can be browsed as needed, is generally used. It was a target.
[0003]
[Problems to be solved by the invention]
However, in a general information management system as described in Japanese Patent Application Laid-Open No. 2002-41726 described above, judgment of ability is performed based on subjective evaluation information obtained on the basis of the occupation in which the individual is engaged. You can't get out of the traditional way of doing things. The qualifications used as a material for objective judgment are also used to evaluate universal skills and knowledge, and not those that can evaluate the practical ability to perform the duties required in each occupation. No, after all, it's out of analogy. Therefore, even though these qualifications can be converted into electronic information as one element of individual evaluation, they cannot be used as the main objective criteria.
[0004]
In particular, in the current situation where business operations performed by companies are diversified and sophisticated, the content and level of duties required for each occupation are subdivided, and at the practical level, it is necessary to match the individual's profession to the actual In fact, the actual situation is that adjustment is made by OJT (ON-THE-JOB training) so as to forcibly adapt.
[0005]
Therefore, from the viewpoint of revitalizing existing human resources, companies must accurately grasp the individual's abilities as the ability to achieve their duties, and then objectively analyze what the existing human resources can do. Therefore, there is a need for a human resource utilization support system and a human resource utilization support program that clarify the correlation with the duties required by the affiliation and realize the true right person in the right job.
[0006]
[Object of the invention]
In view of such demands, the present invention provides an analysis environment for matching the functions provided by human resources with the duties required by companies, and realizes a human resource utilization support system and a human resource utilization support program that support the activation of human resources. The purpose is to do.
[0007]
[Means for Solving the Problems]
In order to achieve the above object, according to the first aspect of the present invention, there is provided a personal analysis response information receiving unit for receiving personal analysis response information including a current affiliation and a response to a predetermined question, and an answer included in the response. Answer information analyzing means for analyzing words and associating the appearance frequency of the answer words with the answer words, job identification word storage means defining identification words associated with the jobs required for each affiliation, and job identification word storage Affiliation determination means for comparing the affiliation obtained based on the identification word corresponding to the answer word having a predetermined appearance frequency or more with the current affiliation, and determining whether or not the affiliation matches; An information output unit that outputs the belonging match determination result generated by the match determination unit.
[0008]
According to the present invention, since the affiliation obtained based on the identification word corresponding to the answer word having the predetermined appearance frequency or more in the answer sentence to the predetermined question is compared with the current affiliation, the regression method is applied to the third. It is possible to obtain a judgment result based on objective judgment material without the arbitrariness of the person, and, by extension, to provide materials for objectively judging the suitability of the affiliation and the necessity of reassignment. A human resource utilization support system can be realized that helps the utilization of human resources.
[0009]
In the invention according to claim 2, in the human resource utilization support system according to claim 1, analysis result storage means for storing a history of the appearance frequency of the answer word is provided, and the affiliation determination means includes: A configuration is adopted in which affiliations obtained based on identification words corresponding to answer words used in a predetermined time cycle or more extracted on the basis thereof are compared.
[0010]
According to the present invention, since the affiliations obtained based on the identification words corresponding to the answer words used in a predetermined time cycle or more extracted based on the history of the appearance frequency of the answer words are compared, the individual frequently utilizes It is possible to make a determination based on the duties that are present, and it is possible to make a determination in accordance with the characteristics of the individual.
[0011]
According to a third aspect of the present invention, in the human resource utilization support system according to the first or second aspect, an evaluation parameter generating unit configured to generate an evaluation parameter for personal analysis based on an appearance frequency of the answer word; Comparison parameter storage means for storing a comparison parameter serving as a reference for comparison of evaluation parameters; job adaptability determination means for comparing the evaluation parameter generated by the evaluation parameter generation means with the comparison parameter stored by the comparison parameter storage means; , And the information output unit outputs the job adaptability determination result generated by the job adaptability determination unit.
[0012]
According to the present invention, since the evaluation parameter based on the appearance frequency of the answer word and the comparison parameter stored in the comparison parameter storage unit are compared, various judgments based on an arbitrary comparison parameter can be made according to various duties. Can be realized.
[0013]
Further, in the invention according to claim 4, the job adaptability determination means employs a configuration in which comparison is performed using a type pattern corresponding to the evaluation parameter and the comparison parameter.
[0014]
According to the present invention, comparisons are made using pattern patterns corresponding to the evaluation parameters and the comparison parameters, so that functional comparisons can be made without performing strict numerical comparisons. A human resource utilization support system that can be realized.
[0015]
According to a fifth aspect of the present invention, in the human resources utilization support system according to the third or fourth aspect, the storage means stores a reference unit word as a reference when the answer sentence is decomposed into words. An analysis dictionary storing means for storing the analysis information in association with the analysis section, wherein the answer information analysis means associates the answer word with the analysis section with reference to the analysis dictionary storage means, and The generation means employs a configuration in which the evaluation parameter is generated using a numerical value obtained based on the appearance frequency of the answer word belonging to the analysis section.
[0016]
According to the present invention, the analysis parameter is referred to the analysis dictionary stored in association with the predetermined analysis section, the answer word is associated with the analysis section, and the evaluation parameter is set using the numerical value obtained based on the appearance frequency of the answer word belonging to the analysis section. Since it is generated, a parameter based on objective personal analysis response information can be used as a judgment material, and furthermore, a human resource utilization support system capable of objectively judging an individual's profession in terms of balance of words constituting a sentence can be provided. realizable.
[0017]
The invention according to claim 6 includes an information receiving unit, an information storage unit, and a processing unit that controls the operation of each unit, and the storage unit is associated with a job required for each affiliation. Provided is a human resource utilization support program realized in a human resource utilization support system having a job identification word storage unit that defines an identification word.
[0018]
Then, the human resource utilization support program includes a step of receiving, by the processing unit, answer information for personal analysis including an answer sentence to a current affiliation and a predetermined question, analyzing the answer word included in the answer sentence, Associating an appearance frequency with the answer word, and referring to the job identification word storage means, comparing the affiliation obtained based on the identification word corresponding to the answer word having a predetermined appearance frequency or more with the current affiliation. Then, a step of determining whether or not they match, and a step of outputting a determination result obtained by comparing the affiliations are executed.
[0019]
According to the present invention, the affiliation obtained based on the identification word corresponding to the answer word having the predetermined appearance frequency or more in the answer sentence to the predetermined question is compared with the current affiliation, so that the arbitrariness of a third party may enter. Thus, it is possible to obtain a judgment result based on objective judgment materials without any problem, and thus, it is possible to provide a human resource utilization support environment capable of providing materials that help companies utilize human resources.
[0020]
According to a seventh aspect of the present invention, in the human resource utilization support program according to the sixth aspect, the storage means includes an analysis result storage means for storing a history of appearance frequencies of answer words, The processing means executes a step of comparing affiliations obtained based on identification words corresponding to answer words used in a predetermined time cycle or more extracted based on the history.
[0021]
According to the present invention, since the affiliations obtained based on the identification words corresponding to the answer words used in a predetermined time cycle or more extracted based on the history of the appearance frequency of the answer words are compared, the individual frequently utilizes It is possible to make a determination based on the duties that are present, and it is possible to make a determination in accordance with the characteristics of the individual.
[0022]
According to an eighth aspect of the present invention, in the human resource utilization support program according to the sixth or seventh aspect, the storage means stores a comparison parameter serving as a comparison reference of evaluation parameters for personal analysis. Means for obtaining an evaluation parameter for personal analysis based on the frequency of appearance of the answer word, and comparing the generated evaluation parameter with a comparison parameter stored in the comparison parameter storage means. And outputting a determination result obtained by comparing the evaluation parameter with the comparison parameter.
[0023]
According to the present invention, since the evaluation parameter based on the appearance frequency of the answer word and the comparison parameter stored in the comparison parameter storage unit are compared, various judgments based on an arbitrary comparison parameter can be made according to various duties. Can be realized.
[0024]
According to a ninth aspect of the present invention, in the human resource utilization support program according to the eighth aspect, the processing means executes a step of performing a comparison using a pattern pattern corresponding to the evaluation parameter and the comparison parameter. .
[0025]
According to the present invention, comparisons are made using pattern patterns corresponding to the evaluation parameters and the comparison parameters, so that functional comparisons can be made without performing strict numerical comparisons. Provide a human resource utilization support environment that can
[0026]
According to a tenth aspect of the present invention, in the human resource utilization support program according to the eighth or ninth aspect, the storage means stores a reference unit word as a reference when the answer sentence is decomposed into words. An analysis dictionary storage means for storing the analysis word in association with the analysis section, wherein the processing section refers to the analysis dictionary storage means to associate the answer word with the analysis section; And generating the evaluation parameter using a numerical value obtained based on the appearance frequency of the answer word belonging to.
[0027]
According to the present invention, the analysis parameter is referred to the analysis dictionary stored in association with the predetermined analysis section, the answer word is associated with the analysis section, and the evaluation parameter is set using the numerical value obtained based on the appearance frequency of the answer word belonging to the analysis section. Since it is generated, parameters based on objective personal analysis response information can be used as judgment materials, and by extension, a human resource utilization support environment that can objectively judge individual's profession from the viewpoint of balance of words constituting sentences Can be provided.
[0028]
BEST MODE FOR CARRYING OUT THE INVENTION
The present invention focuses on the affiliation of each individual who works for a company, the duties required for work in the affiliation, and the functions provided by each individual, and performs analysis. Here, the duties are duties assigned to each person, and refer to individual partial works when the type of job is disassembled and subdivided. The occupational abilities refer to occupational and occupational abilities, and the abilities required to perform business.
[0029]
At the time of the above-described analysis, personal analysis answer information including an answer sentence to a predetermined question is used. Specifically, the response information for personal analysis is obtained from a report corresponding to a predetermined question (entry request) related to the business. Naturally, a questionnaire-type labor questionnaire in which a dedicated question sentence is provided and a natural sentence is used to answer the questionnaire is most effective, but it is not limited to this. A daily report or the like provided with Tendently, the former is positioned as a survey based on the writer's individual consciousness under the subconscious consciousness, and the latter is positioned as a survey based on the writer's individual consciousness under the subconscious consciousness.
[0030]
What words are job-aware and what they represent is determined by correlations based on empirical and cumulative facts revealed through years of research and analysis by the inventor. Determined and prescribed. In addition, if a company has its own duties and duties, the correlation specified by that company is also used in the analysis.
[0031]
On the other hand, examples of the profession for performing business include management consciousness, business mind, marketing abilities, plan execution abilities, subordinate cultivation abilities, information gathering abilities, and planning abilities, which are positioned as abilities requirements in sales occupations. The present inventor focuses on competencies composed of comprehension, analysis, thinking, expression, and application as basic abilities to support this function. The target ability is defined as the manifest ability that affects the occupation.
[0032]
Therefore, based on the knowledge that the source of utilizing occupational skills lies in the use of manifest abilities, the ultimate goal is to utilize human resources and revitalize organizations based on the analysis and evaluation of these manifest abilities. To do.
[0033]
For this reason, in the present invention, as a human resource utilization support system and a human resource utilization support program, first, analysis of job matching is performed, and furthermore, by analyzing actual capabilities, a judgment material for determining a guideline for effective utilization of the professional capabilities is determined. I will provide a.
[0034]
Here, the inventor employs linguistic analysis of sentences as a technique for analyzing the manifest ability. The linguistic analysis of a sentence, which is the basis for evaluating the manifesting ability according to the present invention, mainly focuses on evaluating an answer sentence to a predetermined theme. This is based on the results of research that various abilities of a writer are used in writing a text, and as a result, the actual abilities of the writer can be evaluated by analyzing the text. Then, a language analysis method based on empirical and statistical facts obtained from the results of many years of research and analysis by the inventor is incorporated into the ability evaluation method, and a new evaluation sequence is realized.
[0035]
In detail, according to the research by the inventor, the texts that are judged to be excellent as texts discussing one theme in each genre, such as papers and editorials, are analyzed as 50 sentences, 100 sentences, 5000 sentences and included in the sentences. Focusing on the number of occurrences of a word, it was found that words with a high number of occurrences are keywords representing the center of the writer's point, and that many words are used only once.
[0036]
For example, if 100 words are used in a sentence, there are only about 5 keywords that appear frequently and are central to the essence of the point, and about 80 words are used only once. is there. There is a slight difference in numerical values when comparing each genre, but the ratio itself included in the text of this keyword is as long as it is a text that discusses one theme, even if it is a sentence of 600 characters in the same genre, As a result of analysis, it has been found that there is basically no change even in a sentence of 2,000 characters.
[0037]
Furthermore, referring to the number of appearances of a word, if all the words are used only once, as a polem, the sentence is merely a list of irrelevant words, and the point does not appear. On the other hand, if all words are the same, it will not be a sentence. In other words, in order to effectively convey one theme to the other party, as a rule of use for words, there must be a certain amount of words that are used repeatedly, and words that are used only once must surround the surrounding area. There is an image that there is. The analysis result that the value (word duplication rate) obtained by dividing the number of appearing words (including duplication) by the number of appearing words is 1.42 in the case of a dissertation and 2.01 in a novel. Has also been obtained.
[0038]
From the above-mentioned viewpoint, when the ratio of keywords used at a certain ratio or more is analyzed, for example, when the above-described 100 words are used, if the number of keywords exceeds 5 and is more, There is a growing tendency to argue, and there is a strong tendency that sufficient proof is not given in explaining each issue. On the other hand, when the number of keywords does not satisfy 5 and is even smaller, the content is not sufficiently aggregated into the matters to be claimed, and there is a strong tendency that the assembling for developing the thesis is insufficient.
[0039]
On the other hand, if the number of words that have been used only once exceeds 80%, the direction in which they are developed as arguments to the assertion becomes scattered, and the tendency to express a clear direction as a sentence becomes strong.
[0040]
As explained above, from the viewpoint of duplication of words, the finding that the superiority of a sentence, which is the result of comprehensive manifestation of abilities, is not an absolute amount but a balance, and can be judged based on the ideal ratio. The inventor has gained. Here, an excellent sentence is assumed to be a sentence that has a clear meaning, is reasonably proved, and has no imposition or ambiguity.
[0041]
As for the analysis from the viewpoint of balance, similar results are obtained from the viewpoint of the attributes of words such as parts of speech. For example, it can be easily understood by considering general parts of speech such as nouns, adjectives, verbs, and auxiliary verbs. Nouns are used as subjects and objects, but as a natural sentence, if there is no subject, the subject of the action is unknown and the meaning of the subject is unknown without the subject. . Also, if there are few adjectives, etc., it is not possible to sufficiently inform the third party of the detailed situation such as the situation of the problem and the environment surrounding it, and if there are too many adjectives, the information provided will overflow and the reader will want to convey it to the reader There is a high possibility that the image will be blurred. Then, in extreme terms, the absence of a verb means that there is no predicate as a natural sentence, and a sentence cannot be established unless it exists at least in proportion to the subject.
[0042]
By conducting analysis from the viewpoint of word attributes, the inventor was able to obtain the knowledge that whether or not words are used in a well-balanced manner from the viewpoint of attributes can be used as a criterion for evaluating the superiority of a sentence. . Note that, in the present invention, an analysis category different from the general part of speech of the national grammar is adopted as described later. However, this analysis category is important in terms of easy-to-understand sentence expression and sentence composition. Based on the viewpoint of evaluating sentences from the balance, it is necessary to extract the attributes of words to obtain analytical values that enable appropriate numerical evaluation. It is.
[0043]
Furthermore, when the above-mentioned keywords are studied in more detail, the words used twice or more in the sentence are about 20% of the whole word group as described above, and these are the main parts of the sentence. It consists of a group of words and a group of words that complements the group of words. Statistically, around 5% of the whole word group is a keyword group representing the center, and around 15% of the whole word group is a word group explaining this keyword group.
[0044]
In addition, in one sentence, the writer tends to unknowingly distinguish between important words and words that are not, and in addition, has no direct relation to the number of occurrences of frequently used words or sentiment. It was found that the ability of the writer is strongly reflected in the ratio of words present. Therefore, the words used in one sentence express the thoughts of the contained writer, and the expression of whether the words are used and conveyed to the reader as intended. Ability, persuasion ability, logical composition ability, etc. can also be seen.
[0045]
Looking back, if the important words used in the sentence correlate with a given job, it becomes clear that the writer's consciousness is potentially or overtly suited to a particular job. . Therefore, for example, when the answer is included in the answer about the job completed by the user, it can be determined that the user has the ability. By defining the correlation between the word and the job, the objective determination can be made.
[0046]
Embodiments will be described below for a human resource utilization support system and a human resource support program that incorporate evaluation based on linguistic analysis based on the results of the above analysis and research, but these embodiments are examples of the present invention. It is not intended to limit the scope of the invention. In addition, each unit will naturally include a unit that can be substituted by those skilled in the art, even if not otherwise specified in the description.
[0047]
[Configuration of human resource utilization support system]
FIG. 1 is a functional block diagram showing a system configuration of a human resource utilization support system 1 according to the present invention. As shown in FIG. 1, the human resource utilization support system 1 includes a personal analysis response information receiving unit 11 that receives response information for personal analysis including a current affiliation and a response to a predetermined question, and a response included in the response. Answer information analyzing means 14 for analyzing words and associating the appearance frequency of the answer words with the answer words; job identification word storage means 16 defining identification words associated with the jobs required for each affiliation; Referencing the storage means 16 and comparing the affiliation obtained based on the identification word corresponding to the answer word having the predetermined appearance frequency or higher with the current affiliation and determining whether or not the affiliation matches. Provide as a configuration.
[0048]
Further, as shown in FIG. 1, the human resource utilization support system 1 includes an analysis dictionary storage unit 13 that stores a reference unit word as a reference when decomposing an answer sentence into words in association with a predetermined analysis section, Analysis result storage means 15 for storing a history of the frequency of occurrence of words, evaluation parameter generation means 19 for generating evaluation parameters for personal analysis based on the frequency of appearance of answer words, and comparison parameters serving as comparison criteria for the evaluation parameters The job adaptability determining means 22 for comparing the evaluation parameters generated by the evaluation parameter generating means 19 with the comparison parameters stored in the comparative parameter storing means 21 and the affiliation matching determining means 17. An information output unit 24 that outputs the affiliation determination result and the job adaptability determination result generated by the job adaptability determination unit 22. , Comprising a.
[0049]
FIG. 2 is a schematic overall system configuration diagram for realizing a human resource utilization support environment including the human resource utilization support system 1 according to the present invention. As shown in FIG. 2, the entire system includes a human resource utilization support system 1 and a LAN 3 for transmitting information to the human resource utilization support system 1 via a network 2. Hereinafter, the human resource utilization support system 1, the network 2, and the LAN (Local Area Network) 3 will be described.
[0050]
As shown in FIG. 2, the human resource utilization support system 1 employs a configuration including an information storage unit 31, a communication unit 32 with an external device, and a processing unit 33 that controls the operation of each unit. Yes, specifically, a server device connected to the network 2 is assumed.
[0051]
Here, the storage unit 31 is a unit that provides a storage area for programs and data, and also provides a development area for the program. For example, a main storage device such as a RAM, a ROM, and an HDD, and an auxiliary storage device may be used. Here, as shown in FIG. 2, the storage unit 31 stores the personal analysis response information storage unit 12, the analysis dictionary storage unit 13, the analysis result storage unit 15, and the job identification word storage in a predetermined storage area. Means 16, affiliation determination result storage means 18, evaluation parameter storage means 20, comparison parameter storage means 21, and job adaptability determination result storage means 23. The outline of each storage means is as follows. It is as follows.
[0052]
The personal analysis response information storage unit 12 provides a storage area for storing electronic informationized response information to be analyzed. The analysis dictionary storage unit 13 provides a storage area for storing an analysis dictionary described later. The analysis result storage means 15 provides a storage area for storing an analysis result described later. The job identification word storage unit 16 provides a storage area for storing an affiliation, a job required for the job, and a word group associated with the job in association with each other. The affiliation determination result storage means 18 provides a storage area for storing affiliation determination results described below. The evaluation parameter storage means 20 provides a storage area for storing evaluation parameters of each individual obtained based on the answer information for individual analysis. The comparison parameter storage unit 21 provides a storage area for storing a comparison parameter serving as a comparison reference when determining an evaluation parameter of each individual. The job adaptability determination result storage means 23 provides a storage area for storing a job adaptability determination result described later.
[0053]
Although the number of the storage means 31 is one in FIG. 2 for ease of explanation, the present invention is not limited to this. Needless to say, a system capable of storing information in a computer may be used.
[0054]
The communication unit 32 is a unit for transmitting and receiving information to and from the outside. The communication unit 32 connects the human resource utilization support system 1 to the network 2 via a communication device such as a router (not shown), and communicates with the LAN 3. It is an information output interface such as a serial port or USB that enables information connection.
[0055]
The processing means 33 is an arithmetic processing means such as a CPU, and is a means for controlling the various devices described above. Although the number of processing means 33 is set to one in FIG. 2 for ease of explanation, the present invention is not limited to this, and a configuration in which control is performed in parallel by physically using a plurality of arithmetic processing means is also possible. Good. Further, the processing unit 33 may be configured to interpret and execute a program provided in an intermediate code format via a virtual machine or the like to realize processing.
[0056]
Next, a network using a dedicated line is used for a large company or the like in a case where a large company or the like operates in a company, and a network using the Internet or a dedicated line is used in a system in which a small company or the like has many clients. You. In the present embodiment, an in-house network using a dedicated line is adopted.
[0057]
The LAN 3 connects the plurality of terminal devices 4 and the printing unit 5 such as a printer to the server device 6 via a LAN cable, so that the server device 6 can share information and services between the terminal devices 4. It is a small network that can be managed and controlled. In the present embodiment, a daily report or the like may be used as the answer information for personal analysis. Therefore, a configuration in which the LAN 3 is provided for each affiliation is employed. When the network 2 is the Internet, a configuration in which the server device 6 functions as a proxy server can be adopted.
[0058]
The terminal device 4 connected to the LAN 3 controls information input means 34, information display means 35, information storage means 36, communication means 37 with other terminal devices 4 and the server device 6, and the operation of each means. Processing means 38 for performing the processing.
[0059]
Here, the input unit 34 is a unit for inputting information such as a keyboard and a scanner. The display unit 35 is a unit for displaying information such as a liquid crystal display and a CRT. The storage unit 36 is a unit for providing an area for expanding and storing information, like the storage unit 31 described above, and includes, for example, a memory and an HDD. The communication unit 37 is an information transmission / reception unit for receiving and transmitting information with another terminal device 4 or the server device 6 positioned as an external device. When configuring a LAN, a NIC (Network Interface) is used. Card). The processing unit 38 is an arithmetic processing unit such as a CPU that manages each unit, similar to the processing unit 33 described above, and interprets and executes a program provided in an intermediate code format via a virtual machine or the like to perform processing. The same applies to the configuration that realizes the following. Although the number of each unit is set to one in FIG. 2 for ease of explanation, the present invention is not limited to this.
[0060]
In FIG. 1, the response information receiving unit 11 for personal analysis, the response information analyzing unit 14, the affiliation determining unit 17, the evaluation parameter generating unit 19, the job adaptability determining unit 22, and the information output unit 24 are shown in FIG. In addition, this is a process realized by the processing means 33 of the human resource utilization support system 1 executing a predetermined program and exerting its function. These programs are usually stored in a predetermined storage area of the storage unit 31, read out by the processing unit 33 as needed, and executed in a program development area of the storage unit 31 such as a memory, and exhibit various functions.
[0061]
Hereinafter, each means constituting the human resource utilization support system 1 shown in FIG. 1 will be described in detail.
[0062]
The personal analysis response information receiving means 11 is a means for receiving the individual affiliation response information including the current affiliation of the individual and an answer sentence to a predetermined question related to work. The personal analysis answer information including the affiliation and the answer sentence is received in the form of electronic information. In the description of the present embodiment, in order to facilitate the understanding of the invention, the following description is based on the premise of a method of receiving the answer information for personal analysis in the dedicated labor survey form described above.
[0063]
Here, the current affiliation is not limited to the department to which the individual belongs. For example, a position may be included. In addition, the predetermined question is a question for allowing a person to answer a word related to a job related to the job, and in the present embodiment, in particular, a question for extracting a job utilized by the individual in the current job. Set the content. However, it is necessary to provide a multifaceted question for analysis and to bring out the whole picture of the individual. Therefore, it is preferable that the question is composed of a plurality of questions with different expressions. In addition, in order to generate an evaluation parameter to be described later, it is necessary to acquire an answer sentence composed of a natural sentence, although the end of the sentence may be a word end. For this reason, a question format that uses questions such as asking for ideas and explaining the reason is adopted.
[0064]
As shown in FIG. 2, the personal analysis response information receiving unit 11 receives personal analysis response information transmitted from the LAN 3 via the network 2. The response information may be received collectively, the response information for individual analysis may be intermittently received on an individual basis, or a combination of both may be received.
[0065]
The personal analysis response information receiving unit 11 also has a function of storing the received personal analysis response information in the personal analysis response information storage unit 12 under the control of the processing unit 33. In the case of storing the response information for personal analysis, a known information management technique is used, such as adopting a method of storing in association with personal identification information such as an ID code capable of identifying the individual.
[0066]
Here, the transmission of the personal analysis response information from the LAN 3 may be in the form of transmission in response to a transmission request from the human resource utilization support system 1 or spontaneous transmission from the LAN 3 side. The response information for personal analysis is transmitted under the management of. However, whether personal analysis information stored in storage means (not shown) provided in server device 6 or personal analysis information stored in storage means 36 provided in terminal device 4 is transmitted. Depends on the collection form of personal analysis information in the LAN 3. For example, when a report on a business such as a daily report is used instead of a labor survey form, as part of the daily report submission process from the terminal device 4, the daily report as personal analysis answer information under the control of the server device 6 Will be sent.
[0067]
The personal analysis response information storage unit 12 is a unit that stores the personal analysis response information received by the personal analysis response information receiving unit 11 in a predetermined storage area. A mode in which the current affiliation and the answer sentence for a plurality of questions are integrally stored in association with personal identification information, or a mode in which the current affiliation and a plurality of answer sentences are separately stored. Alternatively, the present affiliation and each answer sentence may be stored separately. Which storage mode is to be adopted depends on the analysis sequence of the answer information analyzing means 14 described later, and the personal analysis answer information received by the personal analysis answer information receiving means 11 is stored in the personal analysis answer information storing means 12. Set when memorizing.
[0068]
The analysis dictionary storage unit 13 is a unit that functions as an analysis dictionary that defines a reference unit word as a reference, which is referred to when the answer sentence received by the personal analysis answer information receiving unit 11 is decomposed into words. Further, the storage unit stores a reference unit word, which is a reference when decomposing the answer sentence into words, in association with an analysis section corresponding to the attribute of the reference unit word.
[0069]
Here, the attribute of the reference unit word generally refers to the property or characteristic of the word, and is determined from, for example, the classification according to the type or meaning of the part of speech. The analysis category is determined in consideration of attributes such as a type of part of speech and a use mode, in order to mean an evaluation parameter, which will be described later, as a result of analysis from a plurality of viewpoints based on a balance of use frequency of words.
[0070]
Specifically, as shown in FIG. 3, the analysis dictionary storage unit 13 includes a plurality of independent dictionaries, and includes a body language dictionary, a verbal dictionary, a conjunction dictionary, a particle dictionary, an auxiliary verb dictionary, and the like. For the main purpose of correcting natural sentences, phrases and word levels suitable for decomposition, it includes a utilization dictionary, a deleted symbol dictionary, a notation dictionary, a punctuation identification dictionary, an impression word dictionary storage area, and the like. Each dictionary is configured as a database and is stored in a predetermined storage area of the storage unit 31. The above-mentioned physical dictionary and verbal dictionary for storing the reference unit words store the reference unit words in association with the analysis categories. At this time, the analysis section assigned to the word is used and associated.
[0071]
Here, the analysis category to which the reference unit word stored in each dictionary storage area of the analysis dictionary storage unit 13 is associated is determined according to the analysis contents after the decomposition. Instead of using as is, an analysis section with its own arrangement is adopted.
[0072]
Tables 1 and 2 show specific examples of the analysis categories. As shown in Tables 1 and 2, in the present embodiment, a hierarchical structure is adopted for the analysis unit of the reference unit word stored in the analysis dictionary storage unit 13. In detail, Table 1 is a table showing a detailed hierarchical structure of a part where the first hierarchy is divided into a nominal, and Table 2 is a detail of a part where the first hierarchy is divided into a verb and an adjunct. 3 is a table showing a hierarchical structure.
[0073]
As shown in Tables 1 and 2, the reference unit word is divided into three categories, ie, a nominal, a declinable word, and an adjunct word, as the first hierarchy. Codes 01, 02, and 03 are assigned as first-layer section codes representing sections in the first layer, respectively.
[0074]
Further, the second hierarchy is positioned as a subdivision of the noun, verb, and adjunct of the first hierarchy. Specifically, the nouns of the first hierarchy are classified into common nouns, proper nouns, pronouns, numerics, conjunctions, adnominals, and verbs. Declarations in the first hierarchy are divided into verbs and adjectives. Adjuncts in the first hierarchy are divided into auxiliary verbs, particles, and collocations. Codes from 01 to 13 are assigned to each section of the second layer as a second layer section code representing a section in the second layer.
[0075]
In the third layer, as shown in Tables 1 and 2, one section in the second layer is further subdivided into lower sections according to attributes. In a typical example, common nouns are divided into common nouns, adult words, seasonal words, time periods, and other categories, and codes of 01, 02, 03, 04, and 09 are given as third-layer category codes. In addition, proper nouns are divided into personal names, company names, place names, country names, era names, historical terms, and other divisions. Codes of 01, 02, 03, 04, 05, 06, and 09 are used as the third hierarchical division code. Granted. The structure of the third hierarchy is described in Tables 1 and 2, and the description of the other classes and the classification codes will be omitted. Here, as a method of specifying an analysis section by a section code in the present embodiment, when indicating a third-layer analysis section, a third-layer section code is used in combination with at least the second-layer section code.
[Table 1]

[Table 2]

[0076]
Then, as shown in Tables 1 and 2, in the present embodiment, in addition to the above-described analysis section, an extraction section defined in view of the attribute of the reference unit word is defined. Here, the extraction section is a convenient section provided for specifying the words of the analysis section to be collectively extracted, and according to the predetermined section of the second layer section, a, b, c Three categories are specified. For example, when selecting words that have been decomposed as keywords, they are used as indices when extracting words corresponding to sections a and b.
[0077]
Here, the section a means a section in which a group of words for modifying or supplementing the contents are put together. The section b means a section in which a group of words specifically indicating the contents are put together. The section c means a section in which a group of words that does not make sense only by the word is put together.
[0078]
The hierarchical structure data of these analysis sections is stored in the analysis dictionary storage section 13 in a predetermined area of the storage section in consideration of the convenience when counting based on the analysis section is performed after the decomposition. It is stored as data that defines an analysis category associated with the reference unit word. The storage form adopts, for example, a database structure, and defines the association between the hierarchical division and the analysis division code according to the hierarchical structure of the analysis division of each dictionary, and extracts it as a search index for high-speed retrieval. The association with the category is also specified.
[0079]
Each dictionary for storing the reference unit words associated with the analysis categories described above and other dictionaries will be described in detail.
[0080]
The physical dictionary storage area stores reference unit words classified into common nouns, proper nouns, pronouns, numerics (mainly units, etc.), adverbs, adnominals, and inflections in general national grammar. Here, what is characteristic of the present invention is that, for connectives classified as a noun in general national grammar, a conjunctive dictionary is provided in an independent form, and collocations are registered in the noun dictionary. .
[0081]
In view of the fact that connective dictionaries are used frequently when actually analyzing the contents of sentence expressions, this configuration adopts a configuration that is provided separately and not included in the body language dictionary. In addition to the case of using words, it is assumed that a configuration for performing a different linguistic analysis is additionally provided. In addition, although the collocation word is a word expressing a group of ideas although two or more words are combined, it is stored in the body dictionary.
[0082]
The verbal dictionary storage area stores reference unit words classified as verbs and adjectives in general national grammar, and similarly, the particle dictionary storage area stores reference unit words classified as particles, and the auxiliary verb storage area Each reference unit word classified as an auxiliary verb is stored.
[0083]
In addition, the inflection dictionary storage area stores inflection patterns and the like as inflections of reference unit words classified into verbs and adjectives in general national grammar, and when the natural sentence is decomposed, the word is an adjective. It is used together with the lexical dictionary when determining whether or not it is.
[0084]
The period identification dictionary storage area stores a period (.) In order to recognize the end of a natural sentence. However, if necessary, the period identification dictionary storage area may further store a reading point (,), a question mark (?), An exclamation point (!), And the like.
[0085]
The deleted symbol dictionary storage area stores symbols that are not related to decomposition or analysis and that can be deleted without any problem. For example, symbols such as “「 ”,“ ● ”,“ ◎ ”,“ § ”, and“ * ”used for a heading and the like in a sentence can be given.
[0086]
The notation dictionary storage area stores conversion source information and conversion destination information in association with each other in order to convert symbols in natural sentences classified as special symbols in electronic documents and the like. For example, a percent sign (%) and the word “percent”, a dollar sign (＄) and the word “dollar”, a circle (￥) and “yen” or “en” are stored in association with each other.
[0087]
The thought word dictionary storage area does not use the sentence at the stage of decomposing the answer sentence into answer words, but stores impression words required when generating an analysis value such as a center keyword ratio and a center complement word ratio described later. In detail, the impression word dictionary storage area is provided with a special section for analysis called an impression word as a section unique to the present invention, and performs a process of generating an analysis value by subtracting the number of words corresponding to the analysis section. This is a dictionary storage area that stores a reference unit word corresponding to the impression word and serves as a reference dictionary when matching is determined.
[0088]
Here, the impression word corresponds to the subjective impression of the natural text creator, such as "I" or "I think" when expressed as "I think ...". It is a word used when describing. For example, words that correspond to impression nouns, such as "self", "self", "me", "me", "our company", and "we", are words that correspond to impression verbs such as "think", "can", "to", and "think". And "feel".
[0089]
Note that, in the present embodiment, as a configuration that generally matches the analysis category, such as the physical dictionary storage area, the verbal dictionary storage area, the conjunction dictionary storage area, and the like, it is possible to speed up reference when decomposing an answer sentence into answer words. As described above, the present invention is not limited to this. For example, if it is possible to realize an analysis procedure that is associated with an analysis section and referred to as a reference unit word when decomposing a natural sentence, the word is replaced with a database structure stored so as to form an independent word group. A mixed database structure may be adopted without being summarized based on the analysis section and the part of speech type.
[0090]
Further, as shown in FIG. 3, the single analysis dictionary storage unit 13 is not necessarily limited to a configuration in which a plurality of dictionaries are collectively stored. And a configuration in which a plurality of dictionary groups are stored in a distributed manner may be adopted. Further, the dictionary is not limited to a configuration including a plurality of dictionaries for each function described above. For example, a dictionary that performs a combined function of integrating a word dictionary and a utilization dictionary may be used as a component.
[0091]
Subsequently, the answer information analyzing means 14 is a means for analyzing the answer words included in the answer sentence, associating the appearance frequency of the answer words with the answer words, and acquiring information on the current affiliation of the individual from the answer information. And performs the function of converting the answer sentence into personal analysis information including information in word units.
[0092]
More specifically, the answer information analyzing means 14 reads the answer sentence from the answer information for personal analysis from the answer information storing means 12 for individual analysis, and reads the answer sentence based on the criteria stored in the dictionary storage means 13 for analysis. A process of referring to a unit word to determine whether an answer word included in an answer sentence matches a reference unit word, identifying an answer word included in the answer sentence with a match, and decomposing the answer sentence into answer word units I do. After that, the answer information analyzing means 14 obtains the duplicate number of the same word as the count value for the decomposed answer word, obtains the number of appearances of the answer word, and associates this number of appearances with the answer word as the appearance frequency. The answer information analyzing means 14 acquires the current affiliation of the individual to be analyzed from the respondent's personal information included in the answer information for personal analysis or the accompanying information of the answer sentence.
[0093]
In addition, the answer information analyzing means 14 is a means for storing the answer word obtained as a result of this analysis and the appearance frequency of the answer word in the analysis result storage means 15 in association with the personal identification information described above, and It also functions as a means for storing the affiliation of the user and the personal identification information in the analysis result storage means 15 in association with each other.
[0094]
Here, the decomposition of the answer sentence into answer words will be described in more detail. The answer information analyzing means 14 reads out the answer sentence to be analyzed stored in the personal analyzing answer information storing means 12 in sentence units while referring to the period identification dictionary storage area, and reads a predetermined number of characters of the read natural sentence. Character strings are extracted in order from the beginning of the natural sentence in units, and the respective dictionaries are referred to in a predetermined order, such as a physical dictionary storage area, a verbal dictionary storage area, and a conjunction dictionary storage area of the analysis dictionary storage unit 13, and The unit functions as a unit for decomposing into words by specifying a word which is a unit to be decomposed by comparing with the reference unit word stored in.
[0095]
Further, when decomposing a natural sentence into a plurality of words, the answer information analysis means 14 also functions as a means for associating the analysis section with the decomposed words. The analysis section associated with the reference unit word stored in the analysis dictionary storage section 13 is used as an analysis section to be associated with the decomposed word, thereby functioning as a means for providing an analysis section corresponding to the attribute of the decomposed word. I do.
[0096]
Note that the process of decomposing the answer sentence stored in the personal analysis answer information storage unit 12 into answer words, which is realized by the answer information analysis unit 14, is not limited to the above-described embodiment, and at least the word and the corresponding Anything that can be analyzed and stored in association with the analysis section is sufficient.
[0097]
As described above, the analysis result storage means 15 is a means for storing the answer words obtained by the answer information analyzing means 14 decomposing the answer sentence into word levels, and employs the database structure shown in FIG. It has a function of storing an answer word obtained by decomposing a natural sentence, an analysis section corresponding to the attribute of the answer word, and the above-mentioned appearance frequency in association with each other.
[0098]
In addition, the analysis result storage unit 15 has a function of storing the history of the appearance frequency of the answer word in association with the personal identification information described above. In detail, as shown in FIG. 4, when the response information for personal analysis is periodically received, such as day, month, year, etc., the appearance frequency of the answer word and the history of information related to the answer word can be stored. Adopt a structure to enable analysis including changes and transitions in human resources.
[0099]
The job identification word storage unit 16 is a storage unit that defines an identification word associated with the job required for each affiliation provided by the company. Specifically, the job identification word storage unit 16 has a database structure as shown in FIG. adopt. Here, the job identification word storage unit 16 shown in FIG. 5 shows an example in which a definition word representing a job required for belonging is set as an identification word as it is. For example, as shown in FIG. 5, common nouns such as accounting and accounting for a management department and sales and planning for a sales department are defined as identification words.
[0100]
However, looking at the duties of each affiliation, there are usually a plurality of duties required in the affiliation, and not only the duties unique to that affiliation but also the duties common to each affiliation, etc. The words are not limited to representative definitive words, but words that are common to a plurality of affiliations, such as negotiations, public relations, and negotiations, are identification words associated with affiliations that require them as duties.
[0101]
Further, in the present embodiment, the word that defines the job is used as the identification word that is directly associated with the job, but instead of using the word that directly represents the job, a word that suggests the job required for belonging is used. It may be indirectly used as an identification word representing a job. In such a case, the identification word is not limited to a noun but may also correspond to a verb.
[0102]
In addition to not only preliminarily defining and storing identification words based on recognition common to a plurality of companies, the processing unit 33 accepts company-specific words based on the corporate culture and company terms as identification words. By performing a process of storing the job identification word storage unit 16 and a process of adding the answer word extracted from the personal analysis answer information obtained from the veteran employee belonging to each job to the identification word, the identification word representing the job can be more accurately obtained. A prescribed job identification word storage means 16 can be provided.
[0103]
Subsequently, the affiliation matching determination unit 17 refers to the job identification word storage unit 16, compares the affiliation obtained based on the identification word corresponding to the answer word having the predetermined appearance frequency or higher with the current affiliation, and finds a match. Is a means for determining whether
[0104]
Specifically, the affiliation suitability determination means 17 determines the suitability of the affiliation of the individual by using a regression method based on the correlation between the affiliation and the word representing the job required for the affiliation. As an index, a function for acquiring the answer word described in the answer sentence by the individual and the current affiliation of the individual from the analysis result storage means 15 and the answer used for the analysis based on the appearance frequency associated with the answer word It has a word filtering function, and compares the filtered answer word group with the identification word group associated with the job required for the affiliation, and matches with a perfect match or a partial match of a predetermined ratio or more And a function to acquire the affiliation associated with the identification word group as a suitable affiliation, which is positioned as being suitable for the individual who has judged that it matches. Plus.
[0105]
The affiliation matching determination unit 17 not only uses the frequency of appearance of the answer word included in the one piece of answer information described above for filtering, but also uses the relevant information such as the answer word stored in the analysis result storage unit 15 and the frequency of appearance thereof. And a function of comparing suitable affiliations obtained based on identification words corresponding to answer words used in a predetermined time cycle or more, which are extracted based on the history of.
[0106]
Specifically, the affiliation matching determination means 17 determines not only the frequency of appearance in the description appearing in the answer sentence but also the frequency of appearance over time obtained from the viewpoint of the time cycle such as day, month, and year. Used as a filtering element. For example, in the case where job duties change depending on the season, as a condition, a time cycle of three consecutive months is used as an answer word used for determination of conformity, or an average of one year, an average of one month, an average of one week The preferred affiliation is acquired based on the identification word corresponding to the answer word having a certain frequency of occurrence within a predetermined period, for example, on condition of the time cycle as described above.
[0107]
In this way, the affiliation matching determination means 17 extracts answer words used in a predetermined time cycle or more based on the history of the appearance frequency of the answer words, and obtains the affiliation obtained based on the identification words corresponding to the extracted answer words. Is adopted, the judgment can be made based on the duties frequently used by the individual, and the judgment can be made more in line with the characteristics of the individual.
[0108]
In addition, the affiliation matching determination unit 17 compares the preferred affiliation determined to match with the current affiliation, determines whether or not they match, and stores the determined affiliation matching determination result in the affiliation matching determination result storage unit 18. It also functions as storage means.
[0109]
Here, the affiliation suitability determination result storage means 18 is a means for storing the affiliation suitability determination result determined by the affiliation suitability determination means 17 described above. Specifically, as shown in FIG. In this case, the personal identification information, the affiliation information, and the affiliation match determination result are stored in association with each other.
[0110]
As described above, the response information receiving unit for personal analysis 11, the response information storage unit for personal analysis 12, the analysis dictionary storage unit 13, the response information analysis unit 14, the analysis result storage unit 15, the job identification word storage unit 16, The determination means 17 and the affiliation match determination result storage means 18 are organically combined and function to support human resource utilization for determining the suitability of the current affiliation and the duties of the individual based on the received personal analysis response information. The system 1 is realized.
[0111]
By adopting such a configuration, it is possible to obtain a judgment result based on objective judgment material without any third party's arbitrariness, and thus, it is possible to provide materials that assist companies in utilizing human resources. A human resource utilization support system can be realized.
[0112]
Here, the duties currently performed by each individual are to be realized by the functions of the individual, and judging the suitability of the duties obtained from the answer words is, in turn, the functions of the individual. Is also determined. For example, in the case where the human resource utilization support system 1 provides a result that the business being performed while being in the business completely matches the management duties and conversely does not match the sales duties, The results will be used as a basis for making decisions regarding the utilization of human resources, such as by reassignment.
[0113]
Subsequently, the evaluation parameter generation unit 19, the evaluation parameter storage unit 20, the comparison parameter storage unit 21, the job adaptability determination unit 22, and the job adaptability determination result storage unit 23 related to the processing for determining the job adaptability will be described.
[0114]
First, the evaluation parameter generation unit 19 is a unit that generates an evaluation parameter for personal analysis based on the appearance frequency of an answer word. Here, as the evaluation parameter generated in the present embodiment, a divergence value indicating how much a numerical value such as a ratio indicating an individual's ability obtained from the number of answer words and the like deviates from a predetermined standard is And an actual ability evaluation value for evaluating the actual ability obtained based on the deviation value. Therefore, it can be said that each of these parameters is influenced by a deviation value obtained by objectively quantifying the balance of use of words in Japanese expressions.
[0115]
More specifically, the evaluation parameter generation unit 19 first refers to the association between the analysis unit and the reference unit word serving as a reference when the answer sentence stored in the analysis dictionary storage unit 13 is decomposed into words. The association between the assigned answer word and the analysis section, and the divergence value universalized using a numerical value obtained based on the number related to the answer word, such as the frequency of appearance of the answer word belonging to the analysis section and the number of types of the answer word. Function as means for generating an evaluation parameter. In addition, the evaluation parameter generation unit 19 also functions as a unit that generates an evaluation parameter that can evaluate the individual competency described above and that is an apparent ability evaluation value based on a plurality of generated divergence values.
[0116]
Hereinafter, the evaluation parameter generation function of the evaluation parameter generation means 19 will be described in detail. Schematically, the evaluation parameter generation unit 19 uses the count value related to the answer word stored in the analysis result storage unit to generate the evaluation parameter regardless of whether any of the functions described above is performed. A stepwise generation sequence is adopted in which a universalized ratio is used as an analysis value, and a divergence value indicating the degree of divergence of the analytic value with respect to a reference analysis value is generated as a parameter for determination. Therefore, for ease of understanding, the generation of the analysis value and the generation of the deviation value will be described separately. Although the details will be described later, the reference analysis value used to generate the divergence value is, in principle, the analysis value obtained from literature such as editorials and papers in order to incorporate the cultural background of Japan into the analysis. .
[0117]
The evaluation parameter generation unit 19 refers to the plurality of answer words stored in the analysis result storage unit 15 for each individual, acquires at least the number related to the decomposed answer words as a count value, and performs various analysis values based on the count value. Generate Here, “at least” is used not only for the number of words directly decomposed by the answer information analyzing means 14, but also for the personal analysis answer information storing means 12 as well as the analysis result storing means 15 as necessary. This is because the total number of characters of the target natural sentence, the number of sentences, the total number of characters of words corresponding to each analysis category, and the like are also counted, and the count value is obtained. Note that recounting based on the analysis value or the like generated by the evaluation parameter generation unit 19 is also performed as needed.
[0118]
Table 3 shows typical analytical values according to the present invention. Table 3 describes the analysis items that can evaluate the ability of the individual who wrote the answer based on the contents of the answer, the analysis values corresponding to the analysis items, and the rules for calculating the analysis values. .
[Table 3]

[0119]
For example, as shown in Table 3, as one element of the ability evaluation of the individual who described the answer sentence, the total frequency ratio can be used as the basic data of the evaluation index in the case of the writing ability. The calculation rule is to divide the total number of frequencies by the number of calculated characters. The details of each element used in the calculation rule will be described later.
[0120]
Here, the analysis values associated with the respective abilities described in the analysis item contents of Table 3 represent the respective abilities at the time of evaluating the contents of the text based on the experience obtained from the inventor's many years of research and analysis. It has been certified as suitable for numerical evaluation. Here, even if the numerical values obtained as the analysis values are absolutely compared, it is possible to determine the superiority or inferiority of another sentence regarding the ability. Note that the analysis values and the analysis value calculation rules shown in Table 3 are examples, and even when the calculation rules are defined from other viewpoints and the evaluation is performed using the analysis values obtained based on the specified calculation rules, Needless to say, the present invention can be suitably used.
[0121]
In the present embodiment, the configuration in which the analysis value calculation rule is stored in the analysis value calculation rule storage area as shown in Table 3 is stored in the storage unit 31 in the form of a program function or an arithmetic function of a module. And a method in which the evaluation parameter generation unit 19 generates an analysis value using a program function or a module stored in a predetermined storage area of the storage unit 31.
[0122]
However, as described above, the storage of the analysis value calculation rule is not limited to a form in which the analysis value calculation rule is stored in a subroutine so that it can be repeatedly used as a function. It may be a form to be performed.
[0123]
Hereinafter, the acquisition of the count value used for generating the analysis value itself will be described in detail. The evaluation parameter generation unit 19 counts the number of words while using the analysis section associated with the word as a search index, based on the decomposed word group stored in the analysis result storage unit 15 in the form described above. To get the count value.
[0124]
Here, as an example of the count value based on the word group stored in the analysis result storage means 15, the total number of words to which each section is assigned, such as a first layer section, a second layer section, and a third layer section, is classified. A count value counted for each case is listed. Of the data items obtained as the count value, the number of nouns obtained by counting the number of words corresponding to the noun of the first hierarchical division (first hierarchical division code 01), as well as the word (first hierarchical division The number of words obtained by counting the number of words corresponding to code 02), the number of attached words obtained by counting the number of words corresponding to attached words (first hierarchical division code 03), and the like correspond to this. I do.
[0125]
However, the count value obtained by the evaluation parameter generating means 19 is not limited to the total number of words corresponding to a specific analysis section, and the same word that is repeated is counted as one (one). (Hereinafter, referred to as the number of appearing words). Further, the number of words may be counted over the analysis section, instead of counting only words that fall within the frame of the analysis section.
[0126]
For example, the number of words having the same frequency (hereinafter referred to as the same frequency) obtained by counting the number of words having the same number of occurrences (hereinafter, the number of occurrences is referred to as the frequency number) in the text to be analyzed In addition, the number of words having a frequency of one (a word that appears only once in the sentence to be analyzed) (hereinafter, referred to as one frequency word) is used as a count value as an evaluation parameter generation unit. 19 may be obtained.
[0127]
Subsequently, the evaluation parameter generation unit 19 generates an analysis value using various count values. When generating the analysis value, an analysis value calculation rule storage area (not shown) provided in a predetermined area of the storage unit 31 is used. Is calculated with reference to the analysis value calculation rule as shown in Table 3 and stored in the table.
[0128]
Of the calculated analysis values, specific examples of the analysis values obtained by directly using the count value include the total frequency number, the number of divisor nouns, and the sentence verbal rate shown in Table 3. The total frequency refers to a sum obtained by counting the number of words in each analysis section corresponding to the extraction sections a and b among the analysis sections included in a plurality of natural sentences to be analyzed. The number of divisor nouns refers to the number of words obtained by subtracting the number of words corresponding to a numeral from the sum of the numbers of words corresponding to common nouns, proper nouns, and pronouns in the second hierarchical division. The sentence word rate is a numerical value obtained by dividing the number of words by the number of sentences.
[0129]
Here, in the present embodiment, the total frequency number is not limited to the number of all words, but is limited to the extraction categories a and b. The analysis is performed by excluding the number of words corresponding to the extraction category c from the empirical rule. This is because it has been found that by performing the above, a numerical analysis evaluation can be obtained in accordance with the human analysis evaluation when the expert determines that the sentence is a good sentence.
[0130]
On the other hand, as a specific example of an analysis value generated by using a secondary numerical value obtained by processing a count value, a total frequency ratio, a speech rate, or the like, which is represented by a ratio or a ratio, such as a calculated number of characters. Counts such as analysis values obtained by dividing the count value, etc., central keyword ratio, central complementary word ratio, central 1 frequency word ratio, 1st word frequency ratio, 2nd word frequency ratio, 3rd word duplication ratio An analysis value obtained by dividing a parameter calculated using a value by the number of characters to be calculated may be used.
[0131]
The number of characters calculated here means that the analysis value is greatly affected by physical or superficial factors such as the number of proper noun characters (presence or absence of foreign place names, etc.) and the target text itself. This is a parameter introduced so as not to exist, and is a concept characteristic of the present invention.
[0132]
The calculation rule of the calculated number of characters in the present embodiment is (calculated number of characters) = (total number of characters of a sentence to be analyzed) × (total frequency) / (number of characters of a word corresponding to the analysis section counted by the total frequency) Sum). Therefore, it was called the number of characters because it became the denominator when calculating the ratio.However, the analysis value was calculated using the total number of words, which is the number of words, and the number of frequencies obtained from the dimensionless coefficient obtained from the number of characters. This is to eliminate the influence of the character amount as much as possible.
[0133]
More specifically, when comparing a verbose and difficult-to-understand sentence with a concise and short sentence, in order to obtain an analysis value that is hardly influenced by the amount of characters, the number of words is reduced by the number of characters. Based on the above, the number of calculated characters is selected as a generalization parameter for realizing the generalization of the analysis value.
[0134]
For the central keyword ratio, the central complementary word ratio, the central one-frequency word ratio, and the like, the number of words recognized as keywords among the decomposed words is an important parameter.
[0135]
As mentioned earlier, a keyword is a word having a frequency ratio equal to or less than 0.7, which is obtained by dividing the number of same frequency cases by the number of patterns of the same frequency. Here, the number of same-frequency cases refers to the number of types of words having the same number of appearances for a specific number of occurrences. For example, if there are three types of words having four appearances, the number of the same frequency with the frequency of four is three. The number of patterns having the same frequency refers to the number of word groups recognized as having the same frequency. Specifically, the number of word groups having the same number of appearances is counted. For example, a word group with one appearance, a word group with two appearances, a word group with three appearances, a word group with four appearances, a word group with six appearances, and a word group with six appearances If there are 12 word groups, the number of patterns with the same frequency is 6. At this time, as for the word group having the frequency of 4, the number of the same frequency is 3 as described above, so that the same frequency ratio is 0.5, and the word is recognized as a keyword. For reference, analysis of one of the editorials considered to be excellent sentences shows that the frequency 12 is the same frequency, the frequency 6 is the same frequency 1, and the frequency 4 is the same. The number of the same frequency is 3, the number of the frequency 3 is the number of the same frequency 5, the number of the frequency 2 is the number 18 of the same frequency, the number 1 of the frequency is 178, and the number of the frequency is 12, up to 6, and the number of the frequency 4 Are recognized as keywords.
[0136]
For this keyword recognition, the evaluation parameter generation unit 19 determines whether the decomposed word or word group corresponds to the keyword by determining whether the same frequency ratio calculated based on each count value is 0.7 or more. It also has a function to judge. At this time, the evaluation parameter generation unit 19 counts the number of words corresponding to the keyword (hereinafter, referred to as the number of keywords) based on the determination result, and further, based on the count value, the evaluation parameter generation unit 19 calculates the analysis value. Generate.
[0137]
Specifically, the analysis value generated based on the number of words corresponding to the keywords described above is the number of complementary words, and is a numerical value obtained by subtracting the above-mentioned one-frequency word number from the number of keywords.
[0138]
Further, the evaluation parameter generation unit 19 has a function of determining whether or not an impression word exists in a keyword group determined to correspond to a keyword or a complementary word group determined to correspond to a complementary word. More specifically, the judgment is made by referring to a not-shown impression word dictionary storage area stored in the analysis dictionary storage unit 13 described above, and seeing a match with the reference unit word stored as the impression word. I do.
[0139]
The evaluation parameter generating means 19 generates a numerical value obtained by subtracting the number of impression words from the number of keywords as an analysis value, as the number of central keywords, based on the recognition of the impression words. In addition, a numerical value obtained by subtracting an impression word from the number of complementary words is generated as an analysis value as the number of central complementary words. Further, a numerical value obtained by subtracting an impression word from a word corresponding to the one-frequency word is generated as an analysis value as the number of central one-frequency words. Then, the evaluation parameter generating means 19 divides the number of central keywords, the number of central complementary words, and the number of central one-frequency words by the number of calculated characters, respectively, to obtain the central keyword ratio, central complementary word ratio, central one-frequency word ratio. Is generated and stored in a predetermined area of the storage unit 31 as an analysis value.
[0140]
However, as the keyword recognition method in the present embodiment, the above-mentioned frequency divided by the number of patterns of the same frequency is used as the same frequency ratio, but the concept of the keyword is, as described above, a certain number of times or more. It is a word used in the appearance frequency, and adopts the configuration using the above-mentioned same frequency ratio as an example of an index for performing this distinction. Therefore, the present invention is not limited to this as long as the frequency serving as a criterion for certifying a keyword can be determined. The value divided by the number of frequency cases may be used as the same frequency ratio.
[0141]
Other analysis values include a first word frequency ratio, a second word frequency ratio, and a third word duplication ratio. Here, for the first word, the second word, the third word, and the fourth word, four words having a high frequency (number of appearances) are extracted from the top based on the count value acquired by the evaluation parameter generation unit 19. Identify. Therefore, the frequency number of each word is treated as the first word number, the second word number, the third word number, the fourth word number, and the value obtained by dividing each by the number of calculated characters is the first word frequency ratio, The analysis values are a two-word frequency ratio, a third-word frequency ratio, and a fourth-word frequency ratio. The third word duplication rate is a value obtained by taking the sum of the number of types of the first, second, and third words and dividing the sum by three.
[0142]
Each analysis value is a numerical value representing the content of the text to be analyzed as shown in Table 3. Specifically, the total frequency ratio is a numerical value that indicates sentence maintenance, the divisor noun ratio is a numerical value that indicates vocabulary, the central keyword ratio is a numerical value that indicates the theme setting power, and the central complement is a numerical value that indicates the theme explanatory power. The word ratio is the central 1 frequency word ratio as a numerical value representing the theme supplemental power, the appearing word ratio is a numerical value representing the word composition power, the sentence ratio is a numerical value representing the sentence composing power, and the sentence frequency is a numerical value representing the sentence assertive power The average value indicates the first word frequency ratio as a numerical value indicating assertive power, the second word frequency ratio as a numerical value indicating assertive supplementary power, the third word duplication ratio as a numerical value indicating assertive constituent power, and the logical development power The same frequency number ratio is used as a numerical value, the word regression value is used as a numerical value representing logical coercive force, and the sentence verbal rate is used as a numerical value representing action expression power.
[0143]
Here, the sentence frequency average value is obtained by taking the sum of the word weight values of the words corresponding to the segment b of each sentence, and further summing the sum of the word weight values obtained for each sentence into a sentence. The value is divided by a number. The word weight value here is a frequency ratio of the word, and is obtained by dividing the number of appearances (frequency) by the number of calculated characters.
[0144]
Also, the regression value is a value that indicates the slope when the data distribution is represented by one straight line, but the original meaning of the word regression value is that the frequency ratio of all words is It refers to inclination. However, in the present embodiment, a configuration for approximately acquiring the inclination is adopted. Specifically, an ideal value is used as a parameter on the vertical axis, an analysis value of a text to be analyzed is taken as a parameter on the horizontal axis, and a distribution line having a 45-degree slope is assumed when the analysis value matches the ideal value. As the word frequency ratio, the value obtained by dividing the second word frequency ratio by the first word frequency ratio, the value obtained by dividing the third word frequency ratio by the first word frequency ratio, and the fourth word frequency ratio by the first word frequency ratio Using a total of three values divided by the frequency ratio, the slope is determined using these parameters. Note that the reason for dividing by the first word frequency ratio is that the slope for obtaining the regression value can be approximated to a straight line that increases monotonically.
[0145]
In detail, the vertical axis shows the ideal value calculated using the re-average value calculated from the value when the value falls within the range of the average value ± standard deviation, and the horizontal axis shows the value of each sentence to be analyzed. A real value is obtained, and a slope is calculated from the three pairs of data of the ideal value and the real value by the least square method. Here, a known program function or the like is used for the calculation by the least square method. In addition, the inclination of the word regression value tends to be redundant and ambiguous when the inclination is gentler than 45 degrees, and is conspicuous when the inclination is steeper than 45 degrees. Has been empirically obtained.
[0146]
As described above, the count value and the analysis value to be acquired and generated by the evaluation parameter generation unit 19 prior to the generation of the divergence value have been described. The count value and the analysis value may be stored in a predetermined storage area of the storage unit 31 as appropriate. It is stored in association with the information. Then, the evaluation parameter generation means 19 reads out from the predetermined storage area as needed and uses it for generating a deviation value.
[0147]
Next, generation of a deviation value will be described in detail. As described above, the evaluation parameter generation means 19 uses the reference analysis value as a median value used to represent the degree of the deviation in generating the deviation value, and uses the reference analysis value as a reference for the cultural background of Japan. It is an analysis value obtained from literature such as editorials and papers to incorporate into the analysis. Conversely, according to how to take the standard analysis value, it is possible to set a value as a standard of deviation, and in this embodiment, a culture obtained from an editorial or the like is used as a standard. May be based on a model case arbitrarily defined, or may be based on data obtained from an excellent employee who actually works. Note that the reference analysis value is stored in a predetermined storage area of the storage unit 31, and is read out by the evaluation parameter generation unit 19 when a deviation value is generated.
[0148]
Here, when setting the reference analysis values, sentences that are recognized as excellent in each genre, such as editorials, are selected as samples, and hundreds to thousands of samples are analyzed, and the average of the analysis values is used as the reference analysis value. I do. This is because considering the nature of analyzing sentences, it is desirable that the reference analysis value reflects the culture of Japanese, and from the viewpoint of balance mentioned above, it is converged as Japanese culture. The value is obtained and used as a reference analysis value. However, when the analysis values are obtained for each department using the excellent staff in each department as a group sample and the average value is used as the reference analysis value, the reference analysis value itself strongly reflects the corporate culture and the characteristics of the department. It becomes.
[0149]
However, as the average value, for example, the processing unit 33 calculates a standard deviation using a plurality of analysis values generated for each of a plurality of sentences, and filters and extracts the analysis value again using the calculated standard deviation. The average value of the analyzed analysis value group is preferable. More specifically, as filtering, an average value and a standard deviation are calculated for a plurality of analysis values using each analysis value generated for each sentence to be analyzed, and then (average value) ± (standard The analysis values that do not fall within the range of (deviation) are excluded, and the process of calculating the average value is performed again. The average value obtained again is the re-average value, and it can be said that it is more preferable to adopt a procedure in which this re-average value is positioned as a reference analysis value, that is, an ideal value, and is used as a numerical reference when analyzing a sentence. .
[0150]
In this way, the processing using the standard deviation is performed because the analysis target is a sentence, and the individual difference is likely to be large, so the sample with a large degree of variation is likely to be mixed with the analysis target. , The standard deviation is used to uniformly exclude the disturbance factors to obtain a highly objective evaluation criterion. When a reference including some disturbing factors may be used, an average value directly obtained from a plurality of analysis values may be used as a reference analysis value without performing the above-described numerical operation.
[0151]
Then, the evaluation parameter generation means 19 has a function of generating a divergence value indicating the degree of divergence of the analysis value with respect to the read reference analysis value based on the above-described reference analysis value and the generated analysis value. The calculation rule for calculating the deviation value is shown in the following equation (1). This calculation rule is stored in a divergence value calculation rule storage area (not shown) provided in a predetermined area of the storage unit 31 in the form of a program function or a module in the same manner as the analysis value calculation rule described above.
[0152]
In equation (1), A is called a correction coefficient, B is called a step coefficient, and C is called a variance coefficient. These coefficients are stored for each type of analysis value in a predetermined coefficient storage area of the storage unit 31 which stores coefficients for calculating divergence values.
(Equation 1)

[0153]
Below, the significance of each coefficient and the calculation procedure of each coefficient will be described for reference. First, since the correction coefficient A differs in the number of digits for each analysis value, the correction coefficient A is a coefficient for correcting the analysis value so that the analysis value is easily aligned and universalized when the analysis value is used as an index. In the equation (1), the correction coefficient A is determined such that a value obtained by multiplying the analysis value by the correction coefficient A becomes a median value around 1 in principle. Therefore, the correction coefficient A is obtained by a procedure of dividing 1 by the reference analysis value, and forms the basis of a process of generating a deviation value based on the reference analysis value. For example, a quotient obtained by dividing 1 by the reference analysis value of the total frequency ratio is the correction coefficient A of the total frequency ratio. Specifically, in the case of 0.7445, 1.3431 of the quotient obtained by dividing 1 by 0.7445 becomes the correction coefficient A.
[0154]
Further, in view of the property that the divergence value is based on the evaluation concept that the divergence value is not optimal when the perfect score is optimal but is close to the ideal value when the divergence value is optimal, when the degree of divergence is viewed as an analysis result, Is a coefficient for correcting the calculation result so that it can be easily grasped by the number of integer parts of the evaluation value. In the present embodiment, the step coefficient B is set to 5.
[0155]
Specifically, the divergence value used in the present invention indicates a degree of divergence from a reference value. Therefore, for example, considering the case where 0 is a reference, the degree of deviation is the same regardless of whether it is +5 or -5. In this case, it can be said that when the "moderate" portion represented by 0 is the best, the stage indicating the degree of deviation from the best value is set to 5, and depending on the content of the evaluation, 3 stages, 10 stages In some cases, a coefficient can be arbitrarily adopted instead of a fixed value, because there are cases where it is easier to intuitively understand or use the evaluation result. For this reason, a configuration is adopted in which the step coefficient B is not obtained from the calculation result by the processing unit 33 but is stored in a predetermined storage area in advance.
[0156]
Then, the variance coefficient C is for correcting the width indicating the degree of deviation of the deviation value. Simply adjusting the number of digits of the analysis value by multiplying by the correction coefficient A is not suitable for performing a relative comparison for each divergence value because the range of change of the numerical value is not universalized. What is used for this purpose is the dispersion coefficient C in the equation (1).
[0157]
In the present embodiment, the variance coefficient C is used as a standard deviation that can be calculated when a reference analysis value is obtained, and the divergence value is universalized. Specifically, for the data when the re-average value of the reference analysis value is calculated, the standard deviation calculated using each analysis value and the re-average value may be used, or the standard deviation calculated from the simple average value may be used. May be used.
[0158]
Here, regarding the significance of Expression (1), when the processing unit 33 generates the divergence value, in particular, in Expression (1), by subtracting 1 from the product of the correction coefficient A and the analysis value, By setting a numerical value indicating the degree of deviation based on 0 and multiplying by a step coefficient B, the degree of deviation is corrected so that it can be easily recognized when viewed as an analysis result, and divided by the variance coefficient C. Thus, the calculation result gives the significance of the ratio to the standard deviation, and the width of the maximum value and the minimum value of the divergence value is corrected to a scale that allows relative comparison, whereby the evaluation parameter generation unit 19 is realized. Generated.
[0159]
The method of evaluating using such a divergence value is based on an empirically obtained tendency that, in the case of a sentence that is easy to understand, the analysis value converges to a specific value. .
[0160]
Further, the setting of the variance of the divergence value by the divergence value step coefficient B is set based on the step evaluation according to the required evaluation variation, and may be set in advance. Alternatively, a configuration may be adopted in which the human resource utilization support system 1 receives the number of evaluations via the input unit 34 during the analysis.
[0161]
Further, the evaluation parameter generation unit 19 has a function of storing the generated divergence value in the evaluation parameter storage unit 20. Therefore, from the viewpoint of divergence value generation, the evaluation parameter generation unit 19 refers to a divergence value calculation rule storage area (not shown) provided in the storage unit 31 and a predetermined coefficient storage area, and generates the analysis value generated. Functioning as a means for generating a divergence value based on the divergence value calculation rule read from the divergence value calculation rule storage area and the respective coefficients read from the coefficient storage area. You can say that. Then, by generating a divergence value based on this coefficient, it functions as a means for generating a divergence value based on the reference analysis value, albeit indirectly.
[0162]
Hereinafter, a description will be given of an aspect in which the evaluation parameter generation unit 19 functions as a unit that generates a manifest ability evaluation value using the above-described divergence value. Examples of the manifestation ability evaluation value generated based on the plurality of divergence values generated by the evaluation parameter generation unit 19 include comprehension and grasping power as evaluation elements of understanding power, and analysis power and insight as evaluation elements of analysis power. Power and problem finding power, creativity as an evaluation factor of thinking power, expressive and persuasive power as evaluation elements of expression power, application, action and response as evaluation factors of applied power, Each is listed. In this way, the apparent ability as an evaluation factor is calculated using a plurality of divergence values from different viewpoints.
[0163]
As a method of calculating the apparent ability evaluation value realized by the processing unit 33, a method using an average value of a plurality of divergence values or a difference between the ideal value and the divergence value when compared with the ideal value of the divergence value is used. Any calculation method such as a method in which an absolute value is taken and how much less than the ideal value is based on the absolute value is represented by a deduction method in which the ideal value is a perfect score can be adopted. This calculation rule is also stored in a not-shown apparent ability evaluation value calculation rule storage area provided in a predetermined area of the storage unit 31.
[0164]
In other words, the evaluation parameter generation unit 19, in the same way as the generation procedure of the other units, specifically, generates the generated divergence value and the apparent ability evaluation value calculation rule read from the not-shown apparent ability evaluation value calculation rule storage area. , And functions as means for generating a manifest ability evaluation value.
[0165]
Specifically, in the present embodiment, the evaluation of the comprehension ability, which is an evaluation element of the comprehension ability, is determined in a complex manner from the viewpoints of the speech rate, the theme explanation ability, the word composition ability, the sentence composition ability, and the assertion supplementary ability. You. A numerical value is obtained by averaging five divergence values of the speech rate, the center complement word ratio, the appearance word ratio, the sentence ratio, and the second word frequency ratio.
[0166]
Comprehension of comprehension power, which is an evaluation factor of comprehension power, is determined in a complex manner from the viewpoint of verbal rate, theme explanation power, word composition power, assertion composition power, and logical development power. A numerical value is obtained by averaging five divergence values of the verbal ratio, the center complement word ratio, the appearance word ratio, the third word duplication ratio, and the same frequency ratio.
[0167]
The evaluation of the analytical power, which is an evaluation factor of the power of analysis, is determined in a complex manner from the viewpoints of speech rate, the ability to set the theme, the ability to supplement the theme, the ability to express actions, and the ability to make assertions. A numerical value is obtained by averaging the five divergence values of the speech rate, the central keyword ratio, the central one-frequency word ratio, the sentence verbal rate, and the third word duplication rate.
[0168]
The evaluation of insight, which is an evaluation factor of the ability to analyze, is determined in multiple terms from the viewpoints of vocabulary, theme explanation, sentence assertion, assertion, and assertiveness. A numerical value is obtained by taking an average of five divergence values of a divisor noun ratio, a center complement word ratio, a sentence frequency average value, a first word frequency ratio, and a second word frequency ratio.
[0169]
The evaluation of the problem-finding ability, which is an evaluation factor of the ability to analyze, is determined in a complex manner from the viewpoints of the speech rate, the theme supplementary ability, the word composition ability, the action expression ability, and the assertion composition ability. A numerical value is obtained by averaging five divergence values of the speech rate, the central one-frequency word rate, the appearance word rate, the sentence speech rate, and the third word duplication rate.
[0170]
The evaluation of creativity, which is an evaluation factor of thinking power, is determined in multiple terms from the viewpoints of vocabulary, utterance rate, theme setting power, assertive power, and logical coercive power. A numerical value is obtained by averaging the five divergence values of the divisor noun ratio, the verbal ratio, the central keyword ratio, the first word frequency ratio, and the word regression value.
[0171]
The evaluation of the expression power, which is an evaluation factor of the power of expression, is determined in a complex manner from the viewpoints of vocabulary, theme supplement, word composition, sentence composition, and assertiveness. Then, a numerical value is obtained by averaging the five divergence values of the divisor noun rate, the center 1 frequency word ratio, the appearance word ratio, the sentence ratio, and the first word frequency ratio.
[0172]
The evaluation of persuasive power, which is an evaluation factor of the power of expression, is determined in a complex manner from the viewpoints of vocabulary, theme setting, theme supplementary, assertive, and logical compelling. A numerical value is obtained by averaging five divergence values of the divisor noun ratio, the central keyword ratio, the central one-frequency word ratio, the first word frequency ratio, and the word regression value.
[0173]
The evaluation of applied power, which is an evaluation factor of applied power, is determined in a complex manner from the viewpoints of vocabulary, theme supplement, word composition, assertive composition, and logical development. A numerical value is obtained by averaging the five divergence values of the divisor noun rate, the center 1 frequency word rate, the appearance word rate, the third word duplication rate, and the same frequency count rate.
[0174]
The evaluation of behavioral ability, which is an evaluation factor of applied power, is determined in multiple terms from the viewpoints of verbal rate, sentence assertion, assertion, logical development, and logical coerciveness. A numerical value is obtained by averaging five divergence values of the verbal ratio, the sentence frequency average value, the first word frequency ratio, the same frequency number ratio, and the word regression value.
[0175]
The evaluation of the response ability, which is an evaluation factor of the applied power, is determined in a complex manner from the viewpoint of the ability to set the theme, the ability to explain the theme, the ability to supplement the theme, the ability to express behavior, and the ability to assert. A numerical value is obtained by averaging five divergence values of a central keyword ratio, a central complementary word ratio, a central one frequency word ratio, a sentence verbal rate, and a first word frequency ratio.
[0176]
However, the powers used as evaluation factors are not limited to the above-mentioned powers. For example, the evaluation of thinking power is complex in terms of vocabulary power, theme setting power, word composition power, assertive power, and logical development power. You can judge. In this case, a numerical value is obtained by averaging five divergence values of the divisor noun ratio, the central keyword ratio, the appearance word ratio, the first word frequency ratio, and the same frequency ratio.
[0177]
The evaluation parameter storage means 20 is a means for storing evaluation parameters such as the divergence value and the apparent ability evaluation value described above. The evaluation parameter storage means 20 stores evaluation parameters to be determined by the job adaptability determination means 22 described later. And a function of storing evaluation parameters as output information output from the information output means 24 to the outside. FIG. 7 shows an example of the database structure of the evaluation parameter storage means 20. A data structure is used in which the personal identification information is stored in association with the divergence value and the apparent ability evaluation value corresponding to the individual. However, the data structure is not necessarily limited to the form shown in FIG. A structure in which values are stored independently may be adopted.
[0178]
On the other hand, the comparison parameter storage unit 21 is a unit that stores a comparison parameter serving as a comparison reference of the evaluation parameter generated by the evaluation parameter generation unit 19 described above. More specifically, the comparison parameter storage unit 21 includes a plurality of types assigned according to the evaluation parameter. The comparison function is provided with a pattern having sufficient functions necessary for the most suitable job among the patterns, and performs a function of setting a criterion for evaluating an evaluation parameter of each individual. FIG. 8 shows an example of the database structure of the comparison parameter storage means 21.
[0179]
Here, the details will be described in detail, but the comparison criterion is a criterion index value obtained from an index value obtained by averaging pattern index values obtained using a plurality of excellent human resources as a writer when evaluating the answer sentence. In some cases, a reference index value obtained from a pattern index value obtained using one excellent human resource as a writer is used. The comparison parameter storage means 21 stores these reference index values in advance in the comparison parameter storage means 21 in principle.
[0180]
Also, in the present embodiment, the response sentence is also a sentence, and even if it is strictly determined that the match is made up to a small number of digits after the decimal point, an appropriate result is not obtained as a sentence comparison. The pattern obtained by the relative comparison of the divergence values is used as a standard for comparison and evaluation so that the comparison can be made in consideration of the variation of the numerical value caused by the diversity of, for example, the “fluctuation” coming from the width of the sentence expression. For this reason, when the required ability is provided, an index value indicating an ideal pattern of each divergence value indicated by the discussed text is positioned as a reference index value. The configuration is such that a pattern index value indicating a pattern in the case of grouping is used. In addition, by incorporating the relative comparison, it is possible to ensure that other abilities are not buried in the superiority of the expression technique.
[0181]
More specifically, the first block consisting of a speech rate, a verbal rate, and an adjunct rate classified into an expression system, a central keyword ratio (theme setting power) classified into a theme system, a central complementary word ratio (theme explanatory power) ), The second block consisting of the central one-frequency word ratio (theme supplemental power), the third block consisting of the total frequency ratio (sentence maintenance power) classified into the word composition system, and the appearing word ratio (the word composition power), the sentence system The fourth block consisting of sentence verbal rate (behavioral expression), average sentence frequency (sentence assertion), and sentence ratio (sentence composition ability), and first word frequency ratio (assertion) Power), the second word frequency ratio (assertion supplementary power), the third word duplication rate (argument composition power), the same frequency ratio (logical development power), and the fifth block consisting of word regression values (logical enforcement power). Divide, Tsu employs a method of comparing the pattern for each click, and stores a comparison parameter of the form corresponding to the block in comparison parameter storage unit 21.
[0182]
Here, as shown in FIG. 8, the above-mentioned comparison parameters generally have different pattern index values for each comparison affiliation in principle and generally indicate a unique pattern. In the case where an ideal employee or model case is directly used as a reference analysis value without setting the value to an analysis value based on data such as editorial data, unlike FIG. 8, the reference index of each block is different. The values will exhibit a uniform state with no variation.
[0183]
The job adaptability determination unit 22 in the present embodiment is a unit that compares the evaluation parameter generated by the evaluation parameter generation unit 19 with the comparison parameter stored in the comparison parameter storage unit 21. It also functions as a means for performing comparison using a type pattern corresponding to the comparison parameter.
[0184]
Here, the type pattern is a pattern obtained by comparing a relative relationship between a plurality of divergence values generated by the evaluation parameter generation unit 19, and in the present embodiment, in particular, a block in which a reference index value is set for each block A pattern representing the relative relationship in units is used. The type of the pattern adopts a form expressed by a numerical value using a pattern index value instead of a shape, and the relative relationship between the divergence values is compared with the pattern index value in consideration of the variety of expressions. Since the pattern index value is generated, it is possible to make a more accurate comparison of the sentence balance in consideration of the ambiguity of the sentence than in the absolute numerical comparison.
[0185]
More specifically, the job adaptability determining unit 22 uses the divergence value group as an evaluation parameter of each individual based on the result of the relative comparison of the other divergence values based on one divergence value, and It functions as a means for generating a pattern index value indicating a relationship, and a subtraction process is used as one method of relative comparison.
[0186]
Hereinafter, a specific mode using the subtraction processing will be described. The job adaptability determination unit 22 performs a function of generating a pattern index value corresponding to each divergence value based on the difference value, using a difference from another divergence value with respect to one divergence value as a relative relationship. However, there are two types of difference value classification patterns: 0 or more and less than 0, three types of positive values, 0, and negative values. Positive and negative values are divided according to a predetermined threshold value. For example, five patterns can be adopted. As the number of types is increased, precise comparison becomes possible. However, when it is desired to take the magnitude of the relative difference (pattern level) into consideration, it is desirable to select five or more types.
[0187]
For example, when five types of pattern comparisons are performed when the step coefficient is 5, the threshold value is set to ± 1, and the pattern index values are set to five types of +2, +1, 0, -1, and -2. Relative comparison is performed based on the first divergence value assigned to each block to form a pattern. The index value of the first divergence value obtained by the relative comparison is 0. Then, the relative values of the divergence values are sequentially compared with the second and third variance values, and ± 1 indicates that the difference between the reference value (first divergence value) and the corresponding divergence value is within ± 1. ± 2 is assigned as an index value when the difference from the corresponding divergence value is larger than 1 from the reference value (first divergence value).
[0188]
Therefore, when the number of types of pattern index values is two, there are combinations of 5 (2-1) powers of pattern index values, and when the number of types of pattern index values is three, 5 (3) In the case where there are combinations of -1) power pattern index values and there are five types of pattern index values as described above, there is a combination of 5 (5-1) power pattern index values.
[0189]
The job adaptability determination unit 22 has a function of comparing the generated pattern index value with the reference index value read from the comparison parameter storage unit 21. However, the job adaptability determination unit 22 compares the first to fifth blocks. In this case, a comparison of 25 × 25 × 5 × 25 × 625 = 48825125 is performed as a whole. However, as a comparison method, it is not necessary to compare all pattern index values. When interpreting the job adaptability in a limited manner and judging that the job is appropriate, only the 625 types in the fifth block are verified. You may make it.
[0190]
As described above, the job adaptability determining unit 22 converts each divergence value into a pattern index value using the first divergence value in a pattern block unit, and performs a relative comparison, thereby performing a comparison simulating pattern recognition. Judgment is made, the ability of the writer of the answer sentence is objectively compared with the comparison target indicated by the reference index value, and the result is stored in the job adaptability determination result storage means 23.
[0191]
By looking at the relative relationship based on one divergence value, it is possible to generalize parameters that make the relative relationship of divergence values clearer, and to perform a relative comparison without comparing the magnitude of the absolute value of the divergence value. By adopting the method, the number of comparison patterns can be reduced to speed up the matching process.
[0192]
Here, the job adaptability determining unit 22 obtains the result of comparing the numerical values of the pattern index values and comparing and determining whether or not the pattern is the same as the job adaptability determination result. Therefore, the reference index value stored in the comparison parameter storage means 21 has a required function, which means that if the values do not match, it means that the function is not provided.
[0193]
It should be noted that the result may simply be the result of the match or not.However, the detailed job adaptability determination result obtained by separating the unmatched portion and the matched portion from each other, lacks the ability of the person. Ability can be shown objectively. In the case of adopting a configuration that stores the results of job adaptability determination in such a form, it is not only a job function comparison for simply determining whether human resources are being utilized, , Also functions as the job adaptability determining means 22 having the competency determining function.
[0194]
In addition, the job adaptability determining means 22 compares the evaluation parameter of the individual with the comparison parameter serving as a reference when comparing the comparison parameter defined for the same affiliation with the current affiliation of the individual. The affiliation match determination result storage means 18 stores the comparison result obtained based on the affiliation match determination result stored in the determination result storage means 18 so as to compare with a comparison parameter about the affiliation different from the current affiliation suggested by the answer information of the individual. A function may be provided for judging job adaptability with reference to the affiliation determination result.
[0195]
The means for storing the job adaptability determination result generated by the job adaptability determination means 22 described above is the job adaptability determination result storage means 23, and stores the job adaptability determination result in association with personal identification information. Specifically, the job adaptability determination result storage means 23 employs a database structure as shown in FIG. 9 and associates the personal affiliation information with the above-described compatibility determination result of the current affiliation with the comparison parameter. In addition, it stores the determination result of suitability with the comparison parameter for the preferred affiliation. In FIG. 9, when the pattern index values are compared, the content of the judgment is as follows: the result that all items matched was indicated by “○”, the result that one item was not matched was indicated by “△”, and two items were judged. The result that the above did not match is represented by "x". However, the present invention is not limited to the data structure shown in FIG. 9, and a configuration may be employed in which the affiliation adaptability determination result is stored independently for each affiliation to be compared.
[0196]
Under the control of the processing unit 33, the information output unit 24 reads out and outputs the affiliation determination result determined by the affiliation determination unit 17 from the affiliation determination result storage unit 18; It functions as a means for reading and outputting the job adaptability determination result performed by the means 22 from the job adaptability determination result storage means 23. In addition, it functions as a means for reading and outputting the evaluation parameters such as the divergence value and the apparent ability evaluation value generated by the evaluation parameter generation means 19 from the evaluation parameter storage means 20.
[0197]
Specifically, the information output unit 24 in the present embodiment is stored in a predetermined storage area of the belonging matching determination result storage unit 18, the job adaptability determination result storage unit 23, and the evaluation parameter storage unit 20 provided in the storage unit 31. It functions as means for outputting the stored electronic information to the LAN 3 via the communication means 32 and the network 2. At the time of output, the information output means 24 generates output information in a form corresponding to the output destination as necessary.
[0198]
For example, when outputting predetermined information to the display means 35 of the terminal device 4 connected to the LAN 3, the display control information that defines how the information is displayed on the display means 35 includes a discrepancy value or an evaluation value. Include the value and output. When outputting predetermined information to the printing unit 5 connected to the LAN 3, the deviation value or the evaluation value is incorporated in the print control information that defines how the printing unit 5 prints the information. To output.
[0199]
When outputting to the display unit 35 and the printing unit 5, the information output unit 24 may perform output information adjustment processing such as correcting the median of deviation values (evaluation reference axis values) from 0 to 5. Good. This is because the evaluation reference axis values are raised as a whole so that the contents of the analysis items shown in Table 3 evaluated by each divergence value are easily grasped, and all the evaluation values are set to 0 or more in principle. This is to make adjustments.
[0200]
In addition, depending on the required output form, the divergence value may be output without modifying the evaluation reference axis value as it is, or further, based on the divergence value, a graph format or a table format may be used. Output information may be generated and output.
[0201]
The various means described above are organically combined and function, so that various judgments based on arbitrary comparison parameters can be realized in a variety of tasks, and strict numerical comparison is performed. This makes it possible to compare occupational functions without human resources, and to realize a human resource utilization support system that can appropriately compare ambiguous parts of human abilities.
[0202]
In particular, a human resource utilization support system that can compare human resource functions using parameters based on objective personal analysis response information as a source of judgment, and can objectively judge individual functions in terms of the balance of words constituting sentences. Can be realized.
[0203]
[Procedure for answer information analysis processing for personal analysis]
Next, the procedure of the personal analysis answer information analysis process of the human resource utilization support system 1 will be described with reference to FIG. FIG. 10 is a flowchart showing a main procedure of the personal analysis answer information analysis processing according to the present embodiment. The personal analysis answer information analysis processing described below is roughly classified into a affiliation suitability determination processing for evaluating affiliation suitability using the above-described answer words and the like, and a job adaptability determination processing performed subsequently.
[0204]
[Affiliation matching judgment processing]
First, the belonging matching determination process will be described with reference to FIG. Here, the human resources utilization support system 1 that performs the affiliation suitability determination process according to the present embodiment is based on the premise of performing the affiliation suitability determination process, as described above, the information receiving unit, the information storage unit 31, and the Processing means 33 for controlling the operation of the means, and the storage means 31 stores the reference dictionary word as a reference when decomposing the answer sentence into words in association with a predetermined analysis section and stores the analysis dictionary storage means 13, A job identification word storage unit 16 that defines an identification word associated with a job required for affiliation, an analysis result storage unit 15 that stores the history of the appearance frequency of answer words, and a comparison standard for evaluation parameters for individual analysis. A configuration including comparison parameter storage means 21 for storing comparison parameters is employed. Note that the accepting unit here is assumed to be an interface such as the communication unit 32.
[0205]
The personal information response information analysis processing including the affiliation determination processing and the job adaptability determination processing described later in the human resource utilization support system 1 is performed in accordance with the human resource utilization support program stored in a predetermined area of the storage unit 31. Is The human resource utilization support program includes a step (S1) of receiving response information for personal analysis including a current affiliation and an answer sentence to a predetermined question in the processing means 33, analyzing the answer word included in the answer sentence, Is associated with the answer word (S2), and referring to the job identification word storage means 16, the affiliation obtained based on the identification word corresponding to the answer word having the predetermined appearance frequency or higher and the current affiliation are determined. A step (S3) of comparing and judging whether they match, and a step (S7) of outputting a judgment result of comparing the affiliations are executed.
[0206]
By executing each of these steps, it is possible to obtain a judgment result based on objective judgment material without any third-party arbitrariness, and thus, it is possible to provide materials that help companies utilize human resources. It provides a human resource utilization support environment.
[0207]
More specifically, as shown in FIG. 10, first, the processing unit 33 is caused to execute a step (S1) of receiving response information for personal analysis. At this time, personal analysis answer information is received from the LAN 3 via the network 2 and a router (not shown). The communication means 32 as an interface with the outside serves as a receiving means, and the processing means 33 serves as a receiving means. The personal analysis response information received via the personal analysis response information storage unit 12 is stored in a predetermined area.
[0208]
Then, as described above, the processing unit 33 receives the personal analysis response information via the reception unit, and stores the received personal analysis response information in a predetermined area of the personal analysis response information storage unit 12. By executing this, the above-described individual analysis response information receiving unit 11 is realized and functions.
[0209]
Subsequently, as shown in FIG. 10, the processing unit 33 includes, in the analysis dictionary storage unit 13, a speech dictionary storage area, a verbal dictionary storage area, a conjunction dictionary storage area, a particle dictionary storage area, an auxiliary verb dictionary storage area, and With reference to the inflection dictionary storage area and the like, the answer words included in the answer sentence are analyzed, and the appearance frequency of the answer word is associated with the answer word, and the appearance frequency associated with the answer word is stored in the analysis result storage unit 15. The storing step (S2) is executed.
[0210]
Specifically, the processing unit 33 refers to the notation dictionary storage area and the deleted symbol dictionary storage area of the analysis dictionary storage unit 13, performs symbol conversion and symbol deletion, and performs an answer suitable for decomposition into words. After processing into a sentence, the sentence is decomposed into character strings of a predetermined length from the beginning of the answer sentence in units of one sentence, and temporarily stored in a predetermined area of the storage means 31 in the order of the decomposition.
[0211]
Then, the processing unit 33 specifies the answer word by referring to the reference unit word stored in the analysis dictionary storage unit 13, and further proceeds to the analysis of the next character string. Here, the processing unit 33 refers to the utilization dictionary storage area at the time of disassembly, and performs processing such as replacing the declinable word with the closed form by this reference.
[0212]
Thereafter, the processing unit 33 counts the number of answer words extracted from the answer sentence for each type of word, and stores the counted value in the analysis result storage unit 15 in association with the answer word as an appearance frequency.
[0213]
Further, in the present embodiment, as shown in FIG. 3, the analysis dictionary storage unit 13 employs a configuration in which the analysis section corresponding to the attribute of the reference unit word is stored in association with the reference unit word. In addition to performing the process of identifying the answer word as described above and decomposing the answer sentence into a plurality of answer words, the reference unit is used as a process of associating the analysis section with the answer word for use in the job adaptability determination process described later. A process of assigning the analysis section assigned to the word to the answer word decomposed as it is is executed.
[0214]
As a result, the analysis section assigned to the answer word can be stored in the analysis result storage unit 15 together with the decomposed answer word, so that the analysis is performed in conjunction with the execution of the step of decomposing the answer sentence into the answer word and associating it with the appearance frequency. A step of associating the answer word with the analysis section with reference to the dictionary storage unit 13 can be executed. It should be noted that a method of assigning the analysis category codes shown in Table 1 and Table 2 is used for assigning the analysis category.
[0215]
In this way, the processing unit 33 analyzes and specifies and decomposes the answer word included in the answer sentence while referring to the analysis dictionary, and stores the specified and decomposed answer word in a predetermined area of the analysis result storage unit 15. By executing the above, the above-described answer information analysis unit 14 is realized and functions.
[0216]
Subsequently, the processing unit 33 is caused to execute a step (S3) of comparing / determining the current affiliation with a suitable affiliation obtained based on the answer word. Specifically, based on the answer words stored in the analysis result storage means 15 and their appearance frequencies, answer words having a predetermined appearance frequency or higher are extracted, and the job identification words are extracted based on the identification words corresponding to the extracted answer words. The associated affiliation is acquired as a preferred affiliation with reference to the storage unit 16. Here, in specifying the suitable affiliation, the matching between the identification word group corresponding to the answer word and the identification word group associated with the job required for the affiliation is determined. As described above, it is determined that there is a match with a perfect match or a partial match of a predetermined ratio or more, and the affiliation determined to be a match is acquired as a suitable affiliation suitable for an individual.
[0219]
As described above, about 80% of the answer words extracted from the answer sentence have an appearance frequency of 1 (the frequency number is 1), so the predetermined appearance frequency set as the filtering condition is: At least two times. However, from the viewpoint of removing noise information, it is preferable that the number of times is three or more.
[0218]
In addition, when the processing unit 33 extracts an answer word based on the history of the appearance frequency stored in the analysis result storage unit 15 as well as the mere appearance frequency, a predetermined time cycle or more extracted based on the history is used. And a step of comparing the affiliations obtained based on the identification words corresponding to the answer words used in.
[0219]
Specifically, an answer word used for determination is specified and extracted based on time record information such as an information reception date associated with the answer word as shown in FIG. Then, the processing unit 33 acquires the preferred affiliation using the above-described procedure based on the extracted identification word corresponding to the answer word used in the predetermined time cycle or more, and compares the preferred affiliation with the current affiliation. Perform the steps.
[0220]
Then, the processing unit 33 compares the obtained preferred affiliation with the current affiliation, determines whether or not they match based on, for example, an affiliation code unique to the affiliation, and determines the determined comparison result as an affiliation matching determination. The result is stored in the belonging matching determination result storage means 18 (S3).
[0221]
After the affiliation suitability determination process described above, the processing unit 33 executes a determination step of receiving an instruction as to whether or not to perform the job adaptability determination process (S4). The processing unit 33 receives the instruction input from the input unit 34 provided in the terminal device 4 via, for example, the LAN 3 or the network 2. Here, when an instruction to evaluate the adaptability to the job required for the affiliation based on the individual's job function is received, the process proceeds to a series of steps (S5, S6) for determining the job adaptability. . On the other hand, when an instruction not to perform the job adaptability determination process is received, the process proceeds to the output step (S7).
[0222]
This determination step (S4) is performed, for example, from the input unit 34 via a question screen (instruction receiving interface) formed on the display unit 35 provided in the terminal device 4 based on the display control information generated by the processing unit 33. A configuration realized by processing such as receiving an instruction command may be adopted.
[0223]
[Job adaptability determination process]
Hereinafter, the job adaptability determination process will be described. Here, as described above, the job adaptability determination process is performed in accordance with the human resource utilization support program stored in a predetermined area of the storage unit 31. Determining (S5, S6) an evaluation parameter for personal analysis based on the frequency of appearance of the answer word; comparing the generated evaluation parameter with a comparison parameter stored in the comparison parameter storage means; Outputting a determination result obtained by comparing the evaluation parameter with the comparison parameter. By performing these steps, various judgments based on arbitrary comparison parameters can be realized in a manner corresponding to various duties.
[0224]
More specifically, as the evaluation parameter generation step (S5), a step-by-step process of acquiring a count value used for generating the above-described divergence value, generating an analysis value based on the count value, and generating a divergence value based on the analysis value is performed. Execute with sequence.
[0225]
As a count value acquisition sequence, the processing unit 33 refers to a plurality of answer words obtained by decomposing the answer sentence stored in the analysis result storage unit 15 and the appearance frequency thereof, and at least determines the number of the answer words. As a count value related to the answer sentence. The count value acquired and stored at this time includes the above-described total frequency number, the number of statements, and the same frequency number.
[0226]
In addition, the processing unit 33 performs analysis other than the count value based on the answer word, such as the total number of characters of the answer sentence stored in the answer information storage unit 12 for personal analysis and the number of sentences (number of natural sentences) constituting the answer sentence. Acquisition processing of a count value required to generate a value is also executed, and stored in a predetermined storage area of the storage means 31 together with the count value related to the above-mentioned answer word.
[0227]
Subsequently, as a sequence for generating a plurality of analysis values based on the count value, the processing unit 33 reads out the acquired count value from a predetermined storage area and stores the analysis value calculation rule shown in Table 3 described above in the storage unit 31. The analysis value is read from a predetermined analysis value calculation rule storage area provided in the storage device, and an analysis value is generated using the count value based on the read analysis value calculation rule. When generating analysis values, a plurality of types of analysis values are calculated for a plurality of analysis items. As a result, the processing unit 33 generates a plurality of types of analysis values based on the count value.
[0228]
More specifically, the processing unit 33 refers to the decomposed answer word stored in the analysis result storage unit 15 and the association between the word and the analysis section, and based on the count value obtained by counting the words belonging to the analysis section. Generate multiple analytical values. The analysis value generated at this time includes a sentence speech rate and the like.
[0229]
If necessary, the processing unit 33 reads out from the above-described analysis value calculation rule storage area using a secondary numerical value or another analysis value generated from the count value generated by the processing unit 33. An analysis value is generated according to the analysis value calculation rule. The analysis value generated at this time includes a speech rate and a verbal rate.
[0230]
In the process of generating the analysis value related to the keyword recognition, a process such as acquiring a count value is performed again. Specifically, the processing unit 33 acquires the same frequency number of words and the like based on the same frequency number obtained as the count value to generate the keyword ratio, and acquires the same frequency ratio. At this time, words having the same frequency ratio of 0.7 or less are recognized as keywords. Similarly, the processing unit 33 performs a count process in order to obtain the number of impression words included in the word group recognized as a keyword. Thereafter, based on the count value such as the number of keywords, the processing unit 33 generates an analysis value such as the number of central keywords.
[0231]
Then, the processing unit 33 stores the generated analysis value in a predetermined storage area of the storage unit 31.
[0232]
Next, as a sequence for generating a divergence value indicating the degree of divergence of the analysis value with respect to the reference analysis value, the processing unit 33 includes, from the predetermined storage area of the storage unit 31, a plurality of analysis values for each type of analysis value together with the reference analysis value. The analysis value is read, and based on the read analysis value and the reference analysis value, using a deviation value calculation rule stored in a predetermined deviation value calculation rule storage area provided in the storage unit 31, the deviation of the reference analysis value is calculated. Generate a plurality of divergence values indicating the degree.
[0233]
Specifically, the processing unit 33 reads the above-described equation (1) from a predetermined area of the storage unit 31 and stores each coefficient such as a correction coefficient obtained from a reference value corresponding to each type of analysis value in the storage unit 31. Then, a step of generating a divergence value for each analysis value based on the read coefficient and the equation (1) is executed. Then, the processing unit 33 stores the generated divergence value in the evaluation parameter storage unit 20.
[0234]
When generating the manifest ability evaluation value, the processing unit 33 reads the divergence value from the evaluation parameter storage unit 20, and further, as described above, complies with the calculation rule of the evaluation value stored in the predetermined area. Accordingly, based on the plurality of divergence values read out, an apparent ability evaluation value is generated, and the generated apparent ability evaluation value is stored in the evaluation parameter storage unit 20 as an individual evaluation parameter.
[0235]
The above is the details of the step of generating an evaluation parameter using a numerical value obtained based on the appearance frequency of the answer word belonging to the analysis category and the step of storing the generated evaluation parameter when the evaluation parameter is a deviation value (S5). It is.
[0236]
As described above, by causing the processing unit 33 to execute the process of obtaining the evaluation parameter for personal analysis using a numerical value based on the answer sentence such as the frequency of appearance of the answer word and the number of characters of the answer sentence, the evaluation parameter generating unit 19 described above is executed. Is realized and functions.
[0237]
Subsequently, the processing unit 33 is caused to execute a step (S6) of comparing the generated evaluation parameter with the comparison parameter stored in the comparison parameter storage unit. In the present embodiment, the processing unit 33 reads out the comparison parameter from the comparison parameter storage unit 21. Based on the reference index value, a method of comparing a relative relationship between a plurality of generated divergence values is employed, and a step of performing a comparison is performed using a type pattern corresponding to the evaluation parameter and the comparison parameter.
[0238]
Specifically, the processing unit 33 sets the pattern index value representing the relative relationship between the divergence values as a relative relationship obtained by relatively comparing the other divergence values with one divergence value as a reference, and using the one divergence value as a reference. Is calculated, and a step of generating a pattern index value corresponding to each divergence value based on the difference value is executed.
[0239]
In the generation of the pattern index value in the present embodiment, in order to realize the idea of comparing the entire analysis pattern without using the numerical result of the exact difference of the divergence value, the divergence value is used instead of the divergence value. A process of replacing with a pattern index value representing a relative relationship between the two is performed. By this processing, it is possible to make an accurate comparison of the sentence balance, taking into account the ambiguity of the sentence.
[0240]
Therefore, when comparing the entire pattern, three types of pattern index values of −1, 0, and +1 are used as suitable pattern index values, and when performing a comparison that further reflects the characteristics of the pattern, As shown in FIG. 8, as the pattern index value considering the pattern level, the step coefficient in the generation of the divergence value is set to 5 as described above. , 0, 1, 2 are used.
[0241]
However, as described above, the processing unit 33 does not need to acquire the relative relationship for all the divergence values, and may generate and acquire the mutual relationship between the divergence values required for the ability determination as the pattern index value.
[0242]
In addition, as a step of comparing the pattern index values, the processing unit 33 reads, from the comparison parameter storage unit 21, a reference index value indicating a comparison parameter as an ideal pattern to which each belongs, and generates a reference index value generated for each answer sentence described above. A comparison is made with a pattern index value obtained from a divergence value, which is each element of the manifest ability, which can estimate an individual's professional ability.
[0243]
That is, by comparing the pattern index values obtained from the evaluation parameters of the individual, the processing unit 33 objectively determines whether or not all the abilities desired in the affiliation to be compared are provided, or whether a part thereof is insufficient. Job applicability determination results can be obtained. More specifically, in the present embodiment, as shown in FIG. 9, the job adaptability is high when all the elements constituting the block match, and when there is an element that does not match, the relocation is performed. In addition, it is necessary to improve the manifestation ability through training, and the adaptability of the job is low, so that the determination is performed for each block. However, the present invention is not limited to this. If necessary, it is determined whether the pattern index values shown in FIG. It may be configured to be provided as additional information of the gender determination result.
[0244]
The processing unit 33 stores the job adaptability determination result having the above-described content in the job adaptability determination result storage unit 23.
[0245]
In this way, the processing unit 33 is caused to generate and generate a pattern index value representing the relative relationship between the divergence values based on the result of relative comparison of the other divergence values with reference to one divergence value. By comparing the pattern index value with the reference index value read from the comparison parameter storage means 21, the job adaptability determination means 22 described above is realized and functions.
[0246]
[Output processing]
Finally, the output step (S7) of the determination result will be described. As shown in FIG. 10, when the process proceeds to the output step (S7), the purpose is not only to output the membership suitability determination result but also to output the job adaptability determination result. There are cases.
[0247]
Specifically, the processing unit 33 reads the belonging match determination result obtained by comparing the preferred belonging with the current belonging from the belonging match determination result storage unit 18 and outputs the read determination result to the outside. Alternatively, a step of reading the job adaptability determination result obtained by comparing the evaluation parameter and the comparison parameter from the job adaptability determination result storage means 23 and outputting the read determination result to the outside is executed (S7).
[0248]
Here, the outside to which the processing unit 33 outputs the determination result is assumed to be the server unit 6 configuring the LAN 3, the display unit 35 of the terminal device 4, the printing unit 5, or the like. Therefore, the processing unit 33 performs an output according to the output request after performing an information form corresponding to the output destination, for example, processing for including the determination result in the print control information and the display control information. Note that, if necessary, the divergence value or the apparent ability evaluation value stored in the evaluation parameter storage unit 20 may be attached to the determination result and output.
[0249]
By causing the processing unit 33 to output the requested information to a predetermined output destination, the information output unit 24 described above is realized and functions.
[0250]
The personal analysis answer information analysis process described above is executed according to the human resource utilization support program stored in a predetermined storage area of the storage unit 31, and analyzes and evaluates the personal analysis answer information to be analyzed. In other words, by implementing these processes by the processing unit 33, the human resource utilization support system 1 that can perform objective evaluation and that can perform evaluation using numerical parameters is realized. It is.
[0251]
【The invention's effect】
Since the present invention is configured and functions as described above, according to the first aspect of the present invention, a regression technique is applied to determine a judgment result based on objective judgment material without any third-party arbitrariness. This makes it possible to obtain information for objectively determining the suitability of the affiliation of the affiliation and the necessity of the reassignment, so that it is possible to realize a human resource utilization support system which helps the utilization of the human resources of the company.
[0252]
According to the second aspect of the present invention, the determination can be made based on the duties frequently used by the individual, and the determination can be made in accordance with the characteristics of the individual.
[0253]
According to the third aspect of the present invention, various determinations based on arbitrary comparison parameters can be realized according to various duties.
[0254]
Further, according to the invention described in claim 4, it is possible to compare occupational functions without strict numerical comparison, and it is possible to realize a human resource utilization support system capable of suitably comparing ambiguous parts of human abilities.
[0255]
Further, in the invention according to claim 5, a parameter based on objective answer information for personal analysis can be used as a judgment material, and further, from the viewpoint of balance of words constituting a sentence, an objective function of an individual can be objectively determined. A human resource utilization support system that can judge can be realized.
[0256]
Further, according to the invention described in claim 6, it is possible to obtain a judgment result based on an objective judgment material without the arbitrariness of a third party, thereby providing a material which helps a company to utilize human resources. Provide a human resource utilization support environment that can
[0257]
According to the invention described in claim 7, the judgment can be made based on the duties frequently used by the individual, and the judgment can be made more in line with the characteristics of the individual.
[0258]
Further, in the invention according to claim 8, various judgments based on arbitrary comparison parameters can be realized according to various duties.
[0259]
Further, according to the ninth aspect of the present invention, it is possible to compare occupational functions without performing strict numerical comparison, and it is possible to provide a human resource utilization support environment capable of suitably comparing ambiguous parts of human abilities.
[0260]
According to the tenth aspect of the present invention, a parameter based on objective answer information for personal analysis can be used as a judgment material, and further, from the viewpoint of balance of words constituting a sentence, an objective function of an individual can be objectively determined. Can provide a human resource utilization support environment that can be determined in a timely manner
[Brief description of the drawings]
FIG. 1 is a block diagram showing a functional configuration of an embodiment according to the present invention.
FIG. 2 is a system configuration diagram showing a schematic system configuration of an embodiment according to the present invention.
FIG. 3 is a schematic structural diagram showing an example of a data structure of an analysis dictionary storage unit shown in FIGS. 1 and 2;
FIG. 4 is a data structure diagram showing an example of a data structure of an analysis result storage means shown in FIG. 1;
FIG. 5 is a data structure diagram illustrating an example of a data structure of a job identification word storage unit illustrated in FIG. 1;
FIG. 6 is a data structure diagram showing an example of a data structure of a belonging matching determination storage means shown in FIG. 1;
FIG. 7 is a data structure diagram showing an example of a data structure of an evaluation parameter storage means shown in FIG.
FIG. 8 is a data structure diagram illustrating an example of a data structure of a comparison parameter storage unit illustrated in FIG. 1;
9 is a data structure diagram illustrating an example of a data structure of a job adaptability determination result storage unit illustrated in FIG. 1. FIG.
FIG. 10 is a flowchart illustrating an example of an information processing procedure according to the present invention.
[Explanation of symbols]
1 Human resources utilization support system
2 Network
3 LAN
4 Terminal device
5 printing means
6 server device
11 Response information receiving means for personal analysis
12 Response information storage means for personal analysis
13 Analysis dictionary storage means
14 Response information analysis means
15 Analysis result storage means
16 Job identification word storage means
17 Membership determination means
18 Affiliation match determination result storage means
19 Evaluation parameter generation means
20 Evaluation parameter storage means
21 Comparison parameter storage means
22 Job adaptability determination means
23 Job adaptability judgment result storage means
24 Information output means
31 storage means
32 Communication means
33 processing means
34 input means
35 Display means
36 storage means
37 Communication means
38 Processing means

Claims (10)

Personal analysis answer information receiving means for receiving personal analysis answer information including the current affiliation and an answer sentence to a predetermined question,
An answer information analyzing means for analyzing an answer word included in the answer sentence and relating the appearance frequency of the answer word to the answer word;
Job identification word storage means defining an identification word associated with the job required for each affiliation;
Referencing the job identification word storage means, comparing the affiliation obtained based on the identification word corresponding to the answer word having a predetermined appearance frequency or more with the current affiliation, and determining whether or not the affiliation matches. Means,
Information output means for outputting the belonging suitability determination result generated by the belonging suitability determining means,
A human resource utilization support system characterized by comprising: In the human resource utilization support system according to claim 1,
An analysis result storage unit for storing a history of the frequency of appearance of the answer words;
The human resources utilization support system, wherein the affiliation determination unit compares affiliations obtained based on identification words corresponding to answer words used in a predetermined time cycle or more extracted based on the history. In the human resource utilization support system according to claim 1 or 2,
Evaluation parameter generation means for generating an evaluation parameter for personal analysis based on the frequency of appearance of the answer word,
Comparison parameter storage means for storing a comparison parameter serving as a comparison reference for the evaluation parameter;
The evaluation parameter generated by the evaluation parameter generation means, the job adaptability determination means for comparing the comparison parameters stored in the comparison parameter storage means,
The human resources utilization support system, wherein the information output unit outputs a job adaptability determination result generated by the job adaptability determination unit. 4. The human resource utilization support system according to claim 3, wherein the job adaptability determination unit performs comparison using a type pattern corresponding to the evaluation parameter and the comparison parameter. In the human resource utilization support system according to claim 3 or 4,
The storage means comprises an analysis dictionary storage means for storing a reference unit word as a reference when decomposing an answer sentence into words in association with a predetermined analysis section,
The answer information analysis unit refers to the analysis dictionary storage unit and associates the answer word with the analysis category,
A human resource utilization support system, wherein the evaluation parameter generation means generates the evaluation parameter using a numerical value obtained based on an appearance frequency of the answer word belonging to the analysis section. Information receiving means, information storing means, and processing means for controlling the operation of each means, wherein the storing means is a job identification word storing means defining an identification word associated with a job required for each affiliation. In the human resource utilization support system
In the processing means,
Accepting personal analysis answer information including the current affiliation and the answer sentence for the predetermined question;
Analyzing the answer words of the answer sentence and associating the frequency of appearance of the answer words with the answer word;
Referring to the job identification word storage means, comparing the affiliation obtained based on the identification word corresponding to the answer word having a predetermined appearance frequency or more and the current affiliation, and determining whether they match,
Outputting a determination result comparing the affiliations;
A human resource utilization support program characterized by running In the human resource utilization support program according to claim 6,
The storage means includes an analysis result storage means for storing a history of the appearance frequency of the answer word,
A human resource utilization support program characterized by causing the processing means to execute a step of comparing affiliations obtained based on identification words corresponding to answer words used in a predetermined time cycle or more extracted based on the history. In the human resource utilization support program according to claim 6 or 7,
The storage unit includes a comparison parameter storage unit that stores a comparison parameter serving as a comparison reference of the evaluation parameter for personal analysis,
In the processing means,
Obtaining an evaluation parameter for personal analysis based on the appearance frequency of the answer word;
Comparing the generated evaluation parameter with a comparison parameter stored by the comparison parameter storage means;
Outputting a determination result obtained by comparing the evaluation parameter with the comparison parameter;
A human resource utilization support program characterized by running In the human resource utilization support program according to claim 8,
A human resource utilization support program characterized by causing the processing means to execute a step of performing comparison using a type pattern corresponding to the evaluation parameter and the comparison parameter. In the human resource utilization support program according to claim 8 or 9,
The storage means comprises an analysis dictionary storage means for storing a reference unit word as a reference when decomposing an answer sentence into words in association with a predetermined analysis section,
In the processing means,
Associating the answer word with the analysis category with reference to the analysis dictionary storage means;
Generating the evaluation parameter using a numerical value obtained based on the appearance frequency of the answer word belonging to the analysis section;
A human resource utilization support program characterized by running

Images