JP4039282B2

JP4039282B2 - Natural language processing system, natural language processing method, and computer program

Info

Publication number: JP4039282B2
Application number: JP2003071544A
Authority: JP
Inventors: 智子大熊; 博増市; 宏樹吉村
Original assignee: Fuji Xerox Co Ltd; Fujifilm Business Innovation Corp
Current assignee: Fujifilm Business Innovation Corp
Priority date: 2003-03-17
Filing date: 2003-03-17
Publication date: 2008-01-30
Anticipated expiration: 2023-03-17
Also published as: JP2004280509A

Description

【０００１】
【発明の属する技術分野】
本発明は、人間が日常的なコミュニケーションに使用する自然言語を数学的に取り扱うための自然言語処理システム及び自然言語処理方法、並びにコンピュータ・プログラムに係り、特に、自然言語文の構文・意味解析を行なう自然言語処理システム及び自然言語処理方法、並びにコンピュータ・プログラムに関する。
【０００２】
さらに詳しくは、本発明は、単語の係り受け関係を正確に把握してより高精度な構文・意味解析処理を行なう自然言語処理システム及び自然言語処理方法、並びにコンピュータ・プログラムに係り、特に、数量詞との組み合わせで格助詞が用いられる場合の係り受け関係を正確に把握してより高精度の構文・意味解析処理を行なう自然言語処理システム及び自然言語処理方法、並びにコンピュータ・プログラムに関する。
【０００３】
【従来の技術】
日本語や英語など、人間が日常的なコミュニケーションに使用する言葉のことを「自然言語」と呼ぶ。多くの自然言語は、自然発生的な起源を持ち、人類、民族、社会の歴史とともに進化してきた。勿論、人は身振りや手振りなどによっても意思疎通を行なうことが可能であるが、自然言語により最も自然で且つ高度なコミュニケーションを実現することができる。
【０００４】
他方、情報技術の発展に伴い、コンピュータが人間社会に定着し、各種産業や日常生活の中に深く浸透している。いまやコンピュータ・データだけでなく、画像や音響などほとんどすべての情報コンテンツがコンピュータ上で取り扱われ、情報の編集・加工、蓄積、管理、伝達、共有など高度な処理を行なうことが可能となっている。
【０００５】
例えば、日本語や英語を始めと刷る各種の言語で記述される自然言語は、本来抽象的であいまい性が高い性質を持つが、文章を数学的に取り扱うことにより、コンピュータ処理を行なうことができる。この結果、機械翻訳や対話システム、検索システム、質問応答システムなど、自動化処理により自然言語に関するさまざまなアプリケーション／サービスが実現される。
【０００６】
かかる自然言語処理は一般に、形態素解析、構文解析、意味解析、文脈解析という各処理フェーズに区分される。
【０００７】
形態素解析では、文を意味的最小単位である形態素（morpheme）に分節して品詞の認定処理を行なう。構文解析では、文法規則などを基に句構造などの文の構造を解析する。文法規則が木構造であることから、構文解析結果は一般に個々の形態素が係り受け関係などを基にして接合された木構造となる。意味解析では、文中の語の語義（概念）や、語と語の間の意味関係などに基づいて、文が伝える意味を表現する意味構造を求めて、意味構造を合成する。また、文脈解析では、文の系列である文章（談話）を解析の基本単位とみなして、文間の意味的なまとまりを得て談話構造を構成する。
【０００８】
構文解析では、自然言語文を受け取り、文法規則に基づいて単語（文節）間の係り受け関係を決定する処理を行なう。構文解析結果は、依存構造と呼ばれる木構造（依存木）の形態で表現することができる。また、意味解析では、単語（文節）間の係り受け関係に基づいて、文中の格関係を決定する処理を行なうことができる。
【０００９】
したがって、構文解析において単語間の係り受け関係を正確に決定することは、力文の意味内容を正確に解釈する上で不可欠である。例えば、入力文に応答して動作する検索システムや質問応答システムにおいては、係り受け関係の決定は信頼性に大いに関わる。
【００１０】
ここで、名詞が格助詞を伴って文中に現れる場合、一般に、名詞は述語に係るとされている。これに対し、数量詞が格助詞を伴って現れる場合、その係り受け関係は明瞭であるとは限らない。
【００１１】
例えば、助数詞の意味を格納した単位意味辞書と名詞に対応する名詞単位辞書を設け、数詞表現に対してこれらの辞書を用いて数詞表現部分の修飾先と修飾関係を決定することができる（例えば、特許文献１を参照のこと）。
【００１２】
また、「一日に一個のりんご」など、数量詞は一般にニ格を受けることができるという性質がある（例えば、非特許文献１を参照のこと）。
【００１３】
また、結合価辞書と数量や程度を示す副詞についての再分類を付与した辞書を用いて、副詞の出現位置に応じて副詞の修飾先を決定することができる（例えば、特許文献２を参照のこと）。
【００１４】
数量詞が格助詞を伴って現れる場合、名詞と同様に述語に係ることが多い。ところが同じ文中にさらに別の数量詞が存在するときには、述語以外に係る可能性があり、解析エラーを発生し易い。数量詞が格助詞を伴って現れる場合であって、同じ文中にさらに別の数量詞が存在するときには、格助詞を伴った数量詞は、用言を修飾する場合、別の数量詞を修飾する場合、さらには曖昧性が生じる場合が考えられる。検索システムや質問応答システムに適用した場合、係り受け関係が誤った解析エラーにより、正確な回答を行なうことができなくなる。
【００１５】
【特許文献１】
特開平２−１１８８７９号公報
【特許文献２】
特開平９−６７７８号公報
【非特許文献１】
星野外著「日本語の表現」（圭文社１９９３，Ｐ９４）
【００１６】
【発明が解決しようとする課題】
本発明の目的は、単語の係り受け関係を正確に把握してより高精度な構文・意味解析処理を行なうことができる、優れた自然言語処理システム及び自然言語処理方法、並びにコンピュータ・プログラムを提供することにある。
【００１７】
本発明のさらなる目的は、数量詞との組み合わせで格助詞が用いられる場合の係り受け関係を正確に把握してより高精度の構文・意味解析処理を行なうことができる、優れた自然言語処理システム及び自然言語処理方法、並びにコンピュータ・プログラムを提供することにある。
【００１８】
【課題を解決するための手段及び作用】
本発明は、上記課題を参酌してなされたものであり、その第１の側面は、数量詞が格助詞を伴って現れる文についての意味解析を行なう自然言語処理システムであって、
同じ句の後方に別の数量詞が存在するかどうかを判断する数量詞判定手段と、同じ句の後方に別の数量詞が存在すると判定された場合に、それぞれの数量詞に含まれる助数詞が持つ意味役割を判定する意味役割判定手段と、
文中の各助数詞が持つ意味役割に応じて、格助詞を伴って現れる数量詞の係り受け関係を判定する意味解析手段と、
を具備することを特徴とする自然言語処理システムである。
【００１９】
ここで、前記意味役割判定手段は、数量詞を構成する助数詞とその意味役割の対応関係を記述した助数詞の対応表と、文中における数量詞の接続関係に応じて助数詞の意味役割を判定するためのルールを記述した助数詞の接続ルールとを備えることによって、それぞれの数量詞に含まれる助数詞が持つ意味役割を前記助数詞の対応表で引き、前記助数詞の接続ルールに従って各助数詞の意味役割を判定することができる。
【００２０】
前記の助数詞の対応表は、１つの助数詞につき複数の意味役割を規定することができる。助数詞の対応表では、助数詞としての「年」、「月」、「日」は、いずれも、「日時」及び「期間」などを意味役割として有していると規定されている。また、助数詞「時」は意味役割として「日時」のみを有していると規定されている。また、助数詞「度」は意味役割として「回数」及び「度数」を有していると規定されている。また、助数詞「回」は意味役割として「回数」のみを有していると規定されている。
【００２１】
また、前記の助数詞の接続ルールでは、助数詞の意味役割を判定するためのルールとして、例えば以下の事柄を規定している。
（１）２以上の意味役割「日時」を持つ数量詞が連続したら、数量詞の意味役割を「日時」に限定する。
（２）意味役割「日時」を持つ数量詞に接尾辞「間」が接続したら、その意味役割を「期間」に限定する。
（３）意味役割「度数」を持つ数量詞の数詞が「−」で始まっていたら、その意味役割を「度数」に限定する。
【００２２】
前記意味解析手段は、文中の各助数詞が持つ意味役割に応じて、格助詞を伴って現れる数量詞の係り受け関係を判定することができる。
【００２３】
例えば、格助詞を伴って現れる数量詞と、同じ句の後方に出現する数量詞の意味役割が一致しない場合には、格助詞を伴って現れる数量詞を文中の述語に掛けるものと判断することができる。
【００２４】
また、格助詞を伴って現れる数量詞と、同じ句の後方に出現する数量詞の意味役割が一致する場合には、格助詞を伴って現れる数量詞を後方に出現する数量詞に掛けるものと判断することができる。
【００２５】
また、少なくとも一方の数量詞が複数の意味役割を持つ場合には、格助詞を伴って現れる数量詞をさらに文中の述語に掛けるものと判断して、数量詞自体が持つあいまい性に対応することができる。
【００２６】
したがって、本発明に係る自然言語処理システムによれば、数量詞の後に格助詞が用いられている場合であっても、文中に数量詞が格助詞を伴って現れた場合、文中に別の数量詞が存在するときにはそれぞれの数量詞を構成する助数詞の属性に基づいて係り受け関係を判断して、構文意味・解析の処理精度を向上することができる。この結果、解析結果を例えば検索システムや質問応答システムに利用する際に、有用な結果を得ることができる。
【００２７】
また、本発明の第２の側面は、数量詞が格助詞を伴って現れる文についての意味解析を行なうための自然言語処理をコンピュータ・システム上で実行するようにコンピュータ可読形式で記述されたコンピュータ・プログラムであって、
同じ句の後方に別の数量詞が存在するかどうかを判断する数量詞判定ステップと、
同じ句の後方に別の数量詞が存在すると判定された場合に、それぞれの数量詞に含まれる助数詞が持つ意味役割を判定する意味役割判定ステップと、
文中の各助数詞が持つ意味役割に応じて、格助詞を伴って現れる数量詞の係り受け関係を判定する意味解析ステップと、
を具備することを特徴とするコンピュータ・プログラムである。
【００２８】
本発明の第２の側面に係るコンピュータ・プログラムは、コンピュータ・システム上で所定の処理を実現するようにコンピュータ可読形式で記述されたコンピュータ・プログラムを定義したものである。換言すれば、本発明の第２の側面に係るコンピュータ・プログラムをコンピュータ・システムにインストールすることによって、コンピュータ・システム上では協働的作用が発揮され、本発明の第１の側面に係る自然言語処理システムと同様の作用効果を得ることができる。
【００２９】
本発明のさらに他の目的、特徴や利点は、後述する本発明の実施形態や添付する図面に基づくより詳細な説明によって明らかになるであろう。
【００３０】
【発明の実施の形態】
以下、図面を参照しながら本発明の実施形態について詳解する。
【００３１】
本発明に係る自然言語処理システムは、数量詞との組み合わせで用いられる格助詞の係り受け関係を正確に把握してより高精度の構文・意味解析処理を行なうものである。
【００３２】
意味解析を行うための文法理論の代表的な例として、Lexical Functional Grammar （ＬＦＧ）を挙げることができる本発明は、例えばＬＦＧ文法理論に基づく構文・意味解析処理に組み込んで実装することができる。
【００３３】
ＬＦＧでは、ネイティブ・スピーカの言語知識すなわち文法を、コンピュータ処理や、コンピュータの処理動作に影響を及ぼすその他の非文法的な処理パラメータとは切り離したコンポーネントとして構成している。まず、自然言語処理システムの全体像について簡単に説明する。
【００３４】
図１には、ＬＦＧに基づく自然言語処理システム１の構成を模式的に示している。
【００３５】
形態素解析部２は、日本語など特定の言語に関する形態素ルール２Ａと形態素辞書２Ｂを持ち、入力文を意味的最小単位である形態素に分節して品詞の認定処理を行なう。例えば、「私の娘は英語を話します。」という文が入力された場合、形態素解析結果として、「私{Noun} の{up} 娘{Noun} は{up} 英語{Noun} を{up} 話す{Verb1}{tr} ます{jp} 。{pt}」が出力される。
【００３６】
このような形態素解析結果は、次いで、統語・意味解析部３に入力される。統語・意味解析部は、文法ルール３Ａや結合価辞書３Ｂなどの辞書を持ち、文法ルールなどに基づく句構造の解析や、文中の語の語義や語と語の間の意味関係などに基づいて文が伝える意味を表現する意味構造の解析を行なう（結合価辞書は動詞と主語などの文中の他の構成要素との関係を記述したものであり、述部とそれに係る語の意味関係を抽出することができる）。そして、構文解析した結果として、単語や形態素などからなる文章の句構造を木構造として表した"ｃ−ｓｔｒｕｃｔｕｒｅ（constituent structure）"と、主語、目的語などの格構造に基づいて入力文を疑問文、過去形、丁寧文など意味的・機能的に解析した結果として"ｆ−ｓｔｒｕｃｔｕｒｅ（functional structure）"を出力する。
【００３７】
図２及び図３には、入力文「私の娘は英語を話します。」を統語・意味解析部１により処理した結果として得られるｃ−ｓｔｒｕｃｔｕｒｅ及びｆ−ｓｔｒｕｃｔｕｒｅをそれぞれ示している。
【００３８】
ｃ−ｓｔｒｕｃｔｕｒｅは、文中の単語や句の構造を木構造形式で表したものであり、構文カテゴリーによって定義される。例えば音素列を生成するための音韻学的な解釈を、ｃ−ｓｔｒｕｃｔｕｒｅを基に行なうことができる。一方、ｆ−ｓｔｒｕｃｔｕｒｅは、文法的な機能を明確に表現したものであり、文法的な機能名、意味的形式、並びに特徴シンボルにより構成される。ｆ−ｓｔｒｕｃｔｕｒｅを参照することにより、主語（subject）、目的語（object）、補語（complement）、修飾語（adjunct）といった意味理解を得ることができる。ｆ−ｓｔｒｕｃｔｕｒｅは、ｃ−ｓｔｒｕｃｔｕｒｅの各節点に付随する素性の集合であり、図３に示すように属性−属性値のマトリックスの形で表現される。すなわち、［］で囲まれた中の左側は素性（属性）の名前であり、右側は素性の値（属性値）である。
【００３９】
なお、ＬＦＧの詳細に関しては、例えばR. M. Kaplan及びJ. Bresnan共著の論文"Lexical-Functional Grammar: A Formal System for Grammatical Representation"（The MIT Press, Cambridge (1982). Reprinted in Formal Issues in Lexical-Functional Grammar, pp. 29-130. CSLI publications, Stanford University(1995).）に記述されている。
【００４０】
次いで、本発明に係る自然言語処理による数量詞との組み合わせで用いられる格助詞の係り受け関係の判定処理について詳解する。
【００４１】
既に述べたように、数量詞が格助詞を伴って現れる場合であって、同じ文中にさらに別の数量詞が存在するときには、格助詞を伴った数量詞は、用言を修飾する場合、別の数量詞を修飾する場合、さらには曖昧性が生じる場合があり、係り受け関係の判定エラーを発生し易い。
【００４２】
そこで、本発明では、数量詞を構成する助数詞の属性すなわち意味役割をあらかじめ記述し、属性に応じた助数詞の接続ルールを規定しておき、数量詞が格助詞を伴って現れる場合であって、同じ文中にさらに別の数量詞が存在するときには、それぞれの数量詞に含まれる助数詞の属性に従って、格助詞を伴って現れる数量詞についての係り受け関係を判定し意味解析出力するようにした。なお、数量詞は、（数量接頭語）、（符号）、（数）、（助数詞）、（数量限定語）の組み合わせで構成されると定義することができる。
【００４３】
数量詞を構成する助数詞の属性すなわち意味役割を記述した助数詞の対応表の構成例を以下の表１に示している。
【００４４】
【表１】

【００４５】
上表に示す例では、助数詞としての「年」、「月」、「日」は、いずれも、「日時」及び「期間」などを意味役割として有していると規定されている。また、助数詞「時」は意味役割として「日時」のみを有していると規定されている。また、助数詞「度」は意味役割として「回数」及び「度数」を有していると規定されている。また、助数詞「回」は意味役割として「回数」のみを有していると規定されている。
【００４６】
また、助数詞の接続ルールは、文中における数量詞の接続関係に応じて助数詞の意味役割を判定するためのルールを記述している。本実施形態に係る自然言語処理システムでは、以下に示す３種類の助数詞の接続ルールを規定している。
【００４７】
（１）２以上の意味役割「日時」を持つ数量詞が連続したら、数量詞の意味役割を「日時」に限定する。
（２）意味役割「日時」を持つ数量詞に接尾辞「間」が接続したら、その意味役割を「期間」に限定する。
（３）意味役割「度数」を持つ数量詞の数詞が「−」で始まっていたら、その意味役割を「度数」に限定する。
【００４８】
図４には、本発明に係る自然言語処理の一実施形態である、数量詞との組み合わせで用いられる格助詞の係り受け関係を判定するための処理手順をフローチャートの形式で示している。
【００４９】
格助詞を接続した数量詞ａを含む文が当該システムに投入されると、まず、同じ句の右側に別の数量詞ｂが存在するかどうかを判定する（ステップＳ１）。
【００５０】
ここで、同じ句の右側に別の数量詞ｂが存在しない場合には、数量詞ａは、通常の名詞と同様に取り扱われ、述語に掛けて（ステップＳ７）、これを構文又は意味解析結果の出力として（ステップＳ６）、本処理ルーチン全体を終了する。
【００５１】
一方、同じ句の右側に別の数量詞ｂが存在すると判定された場合には、それぞれの数量詞に含まれる助数詞が持つ意味役割を決定する（ステップＳ２）。
【００５２】
上述したように、本実施形態では、数量詞を構成する助数詞の属性すなわち意味役割を記述した助数詞の対応表と、文中における数量詞の接続関係に応じて助数詞の意味役割を判定するためのルールを記述した助数詞の接続ルールが用意されているので、当該システムは、これら助数詞の対応表と助数詞の接続ルールを参照することによって、それぞれの数量詞に含まれる助数詞が持つ意味役割を容易に取得することができる。
【００５３】
そして、数量詞ａと数量詞ｂの意味役割が一致しているかどうかを判断する（ステップＳ３）。
【００５４】
ここで、数量詞ａと数量詞ｂの意味役割が一致しない場合には、数量詞ａは、通常の名詞と同様に取り扱われ、述語に掛けて（ステップＳ７）、これを意味解析結果の出力として（ステップＳ６）、本処理ルーチン全体を終了する。
【００５５】
一方、数量詞ａと数量詞ｂの意味役割が一致する場合には、数量詞ａを数量詞ｂに掛ける（ステップＳ４）。
【００５６】
そして、数量詞ａ又は数量詞ｂのうち少なくとも一方が複数の意味役割を持つかどうかをさらに判定する（ステップＳ５）。
【００５７】
ここで、数量詞ａ又は数量詞ｂのうち少なくとも一方が複数の意味役割を持つ場合には、格助詞を伴って出現する数量詞ａの意味役割が曖昧であるとして、数量詞ａを句の述語に掛けるという係り受け関係をさらに追加して（ステップＳ７）、２通りの構文又は意味解析結果を出力して（ステップＳ６）、本処理ルーチン全体を終了する。
【００５８】
以下では、図４に示した数量詞との組み合わせで用いられる格助詞の係り受け関係の判定処理手順に従った数量詞の処理結果の具体例について説明する。
【００５９】
例えば、「彼は５月３日に１回失恋している。」という例文の場合、格助詞を伴う数量詞として「５月３日」が抽出される。この数量詞は、２以上の意味役割「日時」を持つ数量詞が連続していることから、助数詞の接続ルールに従い、その意味役割を「日時」に限定することができる。また、同じ文中に別の数量詞「１回」が抽出されるが、この数量詞は、助数詞の対応表から意味役割として「回数」を持つことが分る。そして、これら同じ文中に出現する２つの数量詞の意味内容が相違することから、格助詞を伴う数量詞「５月３日」は、通常の名詞と同様に取り扱われ、用言「失恋している」を修飾していると判断される。
【００６０】
したがって、「彼はどのくらい失恋しているのですか」という質問文があった場合、この例文に基づく回答は「１回」となるべきである。図５には、この例文の構文解析木を示している。また、図６には、図５に示した構文解析木に基づいた意味解析結果（ｆ−ｓｔｒｕｃｔｕｒｅ）を示している。
【００６１】
また、「彼は３回に１回失恋している。」という例文の場合、格助詞を伴う数量詞として「３回」が抽出される。この数量詞は、助数詞の対応表から意味役割として「回数」を持つことが分る。また、同じ文中に別の数量詞「１回」が抽出されるが、この数量詞は、助数詞の対応表から意味役割として「回数」を持つことが分る。そして、これら同じ文中に出現する２つの数量詞の意味内容が同じであることから、格助詞を伴う数量詞「３回」を同じ文中で後続する数量詞文中の別の数量詞「１回」を修飾していると判断される。
【００６２】
したがって、「彼はどのくらい失恋しているのですか」という同様の質問文があった場合、この例文に基づく回答は「３回に１回」となるべきである。図７には、この例文の構文解析木を示している。また、図８には、図７に示した構文解析木に基づいた意味解析結果（ｆ−ｓｔｒｕｃｔｕｒｅ）を示している。
【００６３】
また、「彼は３日に１回失恋している。」という例文の場合、格助詞を伴う数量詞として「３日」が抽出される。この数量詞は、助数詞の対応表から、意味役割として「日時」以外に、「期間」や「頻度」を持つことが分る。また、同じ文中に別の数量詞「１回」が抽出されるが、この数量詞は、助数詞の対応表から意味役割として「回数」を持つことが分る。この場合、格助詞を伴う数量詞「３日」は複数の意味役割を持つが、助数詞の接続ルールを参照しても該当するルールが発見されないことから、１つの意味役割に特定することができない。
【００６４】
まず、格助詞を伴う数量詞「３日」と、同じ句の中で後続する別の数量詞「１回」がともに、意味役割として頻度を持つことから、格助詞を伴う数量詞「３日」を同じ文中で後続する数量詞文中の別の数量詞「１回」を修飾していると判断される。
【００６５】
また、格助詞を伴う数量詞「３日」が複数の意味役割を持つことから、用言「失恋している」を修飾しているとさらに判断される。
【００６６】
すなわち、この例文の場合、格助詞を伴う数量詞自体にあいまい性があるため、同じ質問文「彼はどれくらい失恋しているのですか」に対する回答は、「１回」となる場合と、「３日に１回」となる場合の両方が考えられる。図９には、この例文についての１つの構文解析結果を示している。また、図１０には、図９に示した構文解析木に基づいた意味解析結果（ｆ−ｓｔｒｕｃｔｕｒｅ）を示している。また、図１１には、この例文についての他の構文解析結果を示している。また、図１２には、図１１に示した構文解析木に基づいた意味解析結果（ｆ−ｓｔｒｕｃｔｕｒｅ）を示している。
【００６７】
［追補］
以上、特定の実施形態を参照しながら、本発明について詳解してきた。しかしながら、本発明の要旨を逸脱しない範囲で当業者が該実施形態の修正や代用を成し得ることは自明である。
【００６８】
本実施形態ではＬＦＧ文法理論に基づいて説明するが、勿論、他の文法ルールを備えた解析システムにおいても本発明を同様に適用することができる。
【００６９】
要するに、例示という形態で本発明を開示してきたのであり、本明細書の記載内容を限定的に解釈するべきではない。本発明の要旨を判断するためには、冒頭に記載した特許請求の範囲の欄を参酌すべきである。
【００７０】
【発明の効果】
以上詳記したように、本発明によれば、単語の係り受け関係を正確に把握してより高精度な構文・意味解析処理を行なうことができる、優れた自然言語処理システム及び自然言語処理方法、並びにコンピュータ・プログラムを提供することができる。
【００７１】
また、本発明によれば、数量詞との組み合わせで格助詞が用いられる場合の係り受け関係を正確に把握してより高精度の構文・意味解析処理を行なうことができる、優れた自然言語処理システム及び自然言語処理方法、並びにコンピュータ・プログラムを提供することができる。
【００７２】
本発明によれば、数量詞の後に格助詞が用いられている場合、文中に数量詞が格助詞を伴って現れた場合、文中に別の数量詞が存在するときにはそれぞれの数量詞を構成する助数詞の属性に基づいて係り受け関係を判断して、構文意味・解析の処理精度を向上することができる。この結果、解析結果を例えば検索システムや質問応答システムに利用する際に、有用な結果を得ることができる。
【図面の簡単な説明】
【図１】ＬＦＧに基づく自然言語処理システム１の構成を模式的に示した図である。
【図２】入力文「私の娘は英語を話します。」を統語・意味解析部１により処理した結果として得られるｃ−ｓｔｒｕｃｔｕｒｅを示した図である。
【図３】入力文「私の娘は英語を話します。」を統語・意味解析部１により処理した結果として得られるｆ−ｓｔｒｕｃｔｕｒｅを示した図である。
【図４】数量詞との組み合わせで用いられる格助詞の係り受け関係を判定するための処理手順を示したフローチャートである。
【図５】例文「彼は５月３日に１回失恋している。」についての構文解析木を示した図である。
【図６】図５に示した構文解析木に基づいた意味解析結果（ｆ−ｓｔｒｕｃｔｕｒｅ）を示した図である。
【図７】例文「彼は３回に１回失恋している。」についての構文解析木を示した図である。
【図８】図７に示した構文解析木に基づいた意味解析結果（ｆ−ｓｔｒｕｃｔｕｒｅ）を示した図である。
【図９】例文「彼は３日に１回失恋している。」についての構文解析木を示した図である。
【図１０】図９に示した構文解析木に基づいた意味解析結果（ｆ−ｓｔｒｕｃｔｕｒｅ）を示した図である。
【図１１】例文「彼は３日に１回失恋している。」についての他の構文解析木を示した図である。
【図１２】図１１に示した構文解析木に基づいた意味解析結果（ｆ−ｓｔｒｕｃｔｕｒｅ）を示した図である。
【符号の説明】
１…自然言語処理システム
２…形態素解析部
２Ａ…形態素ルール，２Ｂ…形態素辞書
３…統語・意味解析部
３Ａ…文法ルール，３Ｂ…結合価辞書[0001]
BACKGROUND OF THE INVENTION
The present invention relates to a natural language processing system, a natural language processing method, and a computer program for mathematically handling a natural language used by humans for daily communication, and in particular, to analyze syntax and semantics of a natural language sentence. The present invention relates to a natural language processing system, a natural language processing method, and a computer program.
[0002]
More particularly, the present invention relates to a natural language processing system and a natural language processing method, and a computer program for accurately grasping the dependency relationship of words and performing more accurate syntax / semantic analysis processing, and in particular, a quantifier. The present invention relates to a natural language processing system, a natural language processing method, and a computer program for accurately grasping a dependency relationship in the case where a case particle is used in combination with, and performing more accurate syntax / semantic analysis processing.
[0003]
[Prior art]
Words that humans use for everyday communication, such as Japanese and English, are called “natural languages”. Many natural languages have a naturally occurring origin and have evolved with the history of mankind, people and society. Of course, people can communicate with each other by gestures and hand gestures, but natural language can realize the most natural and advanced communication.
[0004]
On the other hand, with the development of information technology, computers have become established in human society and have deeply penetrated into various industries and daily life. Now, not only computer data, but almost all information content such as images and sounds are handled on the computer, making it possible to perform advanced processing such as editing / processing, storage, management, transmission and sharing of information. .
[0005]
For example, a natural language written in various languages such as Japanese and English has a nature that is inherently abstract and ambiguous, but it can perform computer processing by handling sentences mathematically. . As a result, various applications / services related to natural language are realized by automated processing such as machine translation, dialogue system, search system, and question answering system.
[0006]
Such natural language processing is generally divided into processing phases of morphological analysis, syntax analysis, semantic analysis, and context analysis.
[0007]
In morphological analysis, a sentence is segmented into morpheme, which is a semantic minimum unit, and part-of-speech recognition processing is performed. In syntax analysis, sentence structure such as phrase structure is analyzed based on grammatical rules. Since the grammatical rule is a tree structure, the parsing result generally has a tree structure in which individual morphemes are joined based on a dependency relationship. In semantic analysis, a semantic structure that expresses the meaning conveyed by a sentence is obtained based on the meaning (concept) of the words in the sentence and the semantic relationship between words, and the semantic structure is synthesized. In context analysis, a sentence series (discourse) is regarded as a basic unit of analysis, and a discourse structure is constructed by obtaining a semantic group between sentences.
[0008]
In the syntax analysis, a natural language sentence is received, and a dependency relationship between words (sentences) is determined based on grammatical rules. The parsing result can be expressed in the form of a tree structure (dependency tree) called a dependency structure. In the semantic analysis, it is possible to perform a process of determining a case relationship in a sentence based on a dependency relationship between words (sentences).
[0009]
Therefore, accurately determining the dependency relationship between words in syntax analysis is indispensable for accurately interpreting the semantic content of a power sentence. For example, in a search system or a question answering system that operates in response to an input sentence, determination of a dependency relationship greatly affects reliability.
[0010]
Here, when a noun appears in a sentence with a case particle, it is generally said that the noun relates to a predicate. On the other hand, when a quantifier appears with a case particle, the dependency relationship is not always clear.
[0011]
For example, a unit meaning dictionary storing the meaning of a classifier and a noun unit dictionary corresponding to a noun are provided, and the modification destination and the modification relationship of the part of the number expression can be determined using these dictionaries for the number expression (for example, , See Patent Document 1).
[0012]
In addition, quantifiers such as “one apple per day” generally have the property of being able to receive a second case (see, for example, Non-Patent Document 1).
[0013]
Further, by using the valency dictionary and quantity for adverbs of degree or reclassifying granted the dictionary, it is possible to determine the modified destination adverb according to the appearance position of the adverb (e.g., refer to Patent Document 2 thing).
[0014]
When quantifiers appear with case particles, they are often related to predicates as well as nouns. However, when another quantifier exists in the same sentence, there is a possibility that it relates to other than the predicate, and an analysis error is likely to occur. When a quantifier appears with a case particle and there are other quantifiers in the same sentence, the quantifier with a case particle modifies the noun, modifies another quantifier, and There may be ambiguity. When applied to a search system or a question answering system, an accurate answer cannot be made due to an analysis error in which the dependency relationship is incorrect.
[0015]
[Patent Document 1]
Japanese Patent Laid-Open No. 2-118879 [Patent Document 2]
Japanese Patent Laid-Open No. 9-6778 [Non-Patent Document 1]
Hoshinogai "Japanese Expression" (Koubunsha 1993, P94)
[0016]
[Problems to be solved by the invention]
An object of the present invention is to provide an excellent natural language processing system, natural language processing method, and computer program capable of accurately grasping the dependency relationship between words and performing more accurate syntax / semantic analysis processing There is to do.
[0017]
A further object of the present invention is to provide an excellent natural language processing system capable of accurately grasping the dependency relationship when a case particle is used in combination with a quantifier and performing more accurate syntax / semantic analysis processing, and It is to provide a natural language processing method and a computer program.
[0018]
[Means and Actions for Solving the Problems]
The present invention has been made in consideration of the above problems, and a first aspect thereof is a natural language processing system that performs semantic analysis on a sentence in which a quantifier appears with a case particle,
The quantifier judging means for judging whether another quantifier exists behind the same phrase, and when it is judged that another quantifier exists behind the same phrase, the semantic role of the classifier included in each quantifier A meaning role judging means for judging;
Semantic analysis means for determining the dependency relationship of quantifiers that appear with case particles according to the semantic role of each classifier in the sentence;
It is a natural language processing system characterized by comprising.
[0019]
Here, the semantic role determining means includes a correspondence table describing a correspondence relationship between a classifier constituting a quantifier and its semantic role, and a rule for determining a semantic role of the classifier according to the connection relationship of the quantifier in the sentence. The semantic role of the classifier included in each quantifier can be looked up in the correspondence table of the classifier, and the semantic role of each classifier can be determined according to the classifier connection rule. .
[0020]
The classifier correspondence table can define a plurality of semantic roles per classifier. In the correspondence table of classifiers, it is stipulated that “year”, “month”, and “day” as classifiers all have “date and time” and “period” as semantic roles. Further, the classifier “time” is defined as having only “date and time” as a semantic role. The classifier “degree” is defined to have “number of times” and “frequency” as semantic roles. Further, the classifier “times” is defined to have only “number of times” as a semantic role.
[0021]
Further, in the classifier connection rule, for example, the following matters are defined as rules for determining the semantic role of the classifier.
(1) If a quantifier having two or more semantic roles “date and time” continues, the semantic role of the quantifier is limited to “date and time”.
(2) When the suffix “ma” is connected to the quantifier having the semantic role “date and time”, the semantic role is limited to “period”.
(3) If the quantifier of the quantifier having the semantic role “frequency” starts with “−”, the semantic role is limited to “frequency”.
[0022]
The semantic analysis means can determine the dependency relationship of quantifiers that appear with case particles according to the semantic role of each classifier in the sentence.
[0023]
For example, if the quantifier appearing with a case particle does not match the semantic role of the quantifier appearing behind the same phrase, it can be determined that the quantifier appearing with the case particle is multiplied by the predicate in the sentence.
[0024]
In addition, if the quantifier that appears with a case particle matches the semantic role of the quantifier that appears later in the same phrase, it may be determined that the quantifier that appears with a case particle is multiplied by the quantifier that appears later. it can.
[0025]
Further, when at least one quantifier has a plurality of semantic roles, it can be determined that a quantifier appearing with a case particle is further multiplied by a predicate in the sentence, and the ambiguity of the quantifier itself can be dealt with.
[0026]
Therefore, according to the natural language processing system of the present invention, even when a case particle is used after a quantifier, when the quantifier appears with a case particle in the sentence, another quantifier exists in the sentence. When doing so, it is possible to determine the dependency relationship based on the attributes of the classifiers constituting the respective quantifiers, thereby improving the processing accuracy of the syntactic semantics / analysis. As a result, a useful result can be obtained when the analysis result is used in, for example, a search system or a question answering system.
[0027]
The second aspect of the present invention provides a computer program described in a computer-readable format so that natural language processing for performing semantic analysis on a sentence in which a quantifier appears with a case particle is executed on a computer system. A program,
A quantifier determination step of determining whether another quantifier exists behind the same phrase;
A semantic role determination step for determining a semantic role of a classifier included in each quantifier when it is determined that another quantifier is present behind the same phrase;
A semantic analysis step for determining a dependency relationship of a quantifier that appears with a case particle according to a semantic role of each classifier in the sentence;
A computer program characterized by comprising:
[0028]
The computer program according to the second aspect of the present invention defines a computer program described in a computer-readable format so as to realize predetermined processing on a computer system. In other words, by installing the computer program according to the second aspect of the present invention in the computer system, a cooperative action is exhibited on the computer system, and the natural language according to the first aspect of the present invention. The same effects as the processing system can be obtained.
[0029]
Other objects, features, and advantages of the present invention will become apparent from more detailed description based on embodiments of the present invention described later and the accompanying drawings.
[0030]
DETAILED DESCRIPTION OF THE INVENTION
Hereinafter, embodiments of the present invention will be described in detail with reference to the drawings.
[0031]
The natural language processing system according to the present invention accurately grasps the dependency relationship of a case particle used in combination with a quantifier, and performs more accurate syntax / semantic analysis processing.
[0032]
Lexical Functional Grammar (LFG) can be cited as a representative example of grammatical theory for performing semantic analysis, and the present invention can be implemented by being incorporated into syntax / semantic analysis processing based on LFG grammatical theory, for example.
[0033]
In LFG, linguistic knowledge, that is, grammar of native speakers is configured as a component separated from computer processing and other non-grammatical processing parameters that affect the processing operation of the computer. First, an overview of the natural language processing system will be briefly described.
[0034]
FIG. 1 schematically shows a configuration of a natural language processing system 1 based on LFG.
[0035]
The morpheme analysis unit 2 has a morpheme rule 2A and a morpheme dictionary 2B related to a specific language such as Japanese, and performs a part-of-speech recognition process by segmenting an input sentence into morphemes that are semantic minimum units. For example, if the sentence “My daughter speaks English” is entered, the result of the morphological analysis is that {up} daughter {Noun} of me {Noun} {up} English {Noun} } Speak {Verb1} {tr} Mas {jp}. {Pt} "is output.
[0036]
Such a morphological analysis result is then input to the syntactic / semantic analysis unit 3. The syntactic / semantic analysis unit has dictionaries such as grammar rules 3A and valence dictionaries 3B. Based on grammatical rule analysis, the meaning of words in a sentence, and the semantic relationship between words. Analyzes the semantic structure expressing the meaning conveyed by a sentence (a valence dictionary describes the relationship between verbs and other constituent elements in the sentence, such as the subject, and extracts the semantic relations between predicates and related words can do). As a result of the parsing, the input sentence is questioned based on “c-structure (constituent structure)” representing a phrase structure of a sentence composed of words, morphemes and the like as a tree structure, and a case structure such as a subject and an object. “F-structure (functional structure)” is output as a result of semantically and functionally analyzing sentences, past tense, polite sentences, and the like.
[0037]
FIGS. 2 and 3 respectively show c-structure and f-structure obtained as a result of processing the input sentence “My daughter speaks English” by the syntactic / semantic analysis unit 1.
[0038]
c-structure represents the structure of words and phrases in a sentence in a tree structure format, and is defined by a syntax category. For example, phonological interpretation for generating a phoneme string can be performed based on c-structure. On the other hand, f-structure clearly expresses a grammatical function, and includes a grammatical function name, a semantic form, and a feature symbol. By referring to the f-structure, it is possible to obtain an understanding of the meaning such as a subject, an object, a complement, and a modifier. The f-structure is a set of features attached to each node of the c-structure, and is expressed in the form of an attribute-attribute value matrix as shown in FIG. That is, the left side in [] is a feature (attribute) name, and the right side is a feature value (attribute value).
[0039]
For details on LFG, see, for example, the paper "Lexical-Functional Grammar: A Formal System for Grammatical Representation" by RM Kaplan and J. Bresnan (The MIT Press, Cambridge (1982). Reprinted in Formal Issues in Lexical-Functional Grammar. , pp. 29-130. CSLI publications, Stanford University (1995)).
[0040]
Next, the processing for determining the dependency relationship of a case particle used in combination with a quantifier by natural language processing according to the present invention will be described in detail.
[0041]
As already mentioned, if a quantifier appears with a case particle and there are other quantifiers in the same sentence, the quantifier with a case particle will use another quantifier if the qualifier is modified. In the case of modification, further ambiguity may occur, and a dependency relationship determination error is likely to occur.
[0042]
Therefore, in the present invention, the attribute of the classifier constituting the quantifier, that is, the semantic role, is described in advance, and the rule for connecting the classifier according to the attribute is defined, and the quantifier appears with a case particle, and the same sentence When there is another quantifier, the dependency relationship of the quantifier appearing with the case particle is determined according to the attribute of the classifier included in each quantifier, and the semantic analysis is output. A quantifier can be defined as a combination of (quantity prefix), (symbol), (number), (numerator), and (quantity limiter).
[0043]
Table 1 below shows an example of the structure of the correspondence table of the classifiers describing the attributes of the classifiers constituting the quantifiers, that is, the semantic roles.
[0044]
[Table 1]

[0045]
In the example shown in the above table, “year”, “month”, and “day” as classifiers are all defined to have “date and time” and “period” as semantic roles. Further, the classifier “time” is defined as having only “date and time” as a semantic role. The classifier “degree” is defined to have “number of times” and “frequency” as semantic roles. Further, the classifier “times” is defined to have only “number of times” as a semantic role.
[0046]
The classifier connection rule describes a rule for determining the semantic role of the classifier in accordance with the connection relationship of the quantifiers in the sentence. In the natural language processing system according to this embodiment, the following three classifier connection rules are defined.
[0047]
(1) If a quantifier having two or more semantic roles “date and time” continues, the semantic role of the quantifier is limited to “date and time”.
(2) When the suffix “ma” is connected to the quantifier having the semantic role “date and time”, the semantic role is limited to “period”.
(3) If the quantifier of the quantifier having the semantic role “frequency” starts with “−”, the semantic role is limited to “frequency”.
[0048]
FIG. 4 shows a processing procedure for determining the dependency relationship of case particles used in combination with a quantifier, which is an embodiment of natural language processing according to the present invention, in the form of a flowchart.
[0049]
When a sentence including a quantifier a connected to a case particle is input to the system, it is first determined whether another quantifier b exists on the right side of the same phrase (step S1).
[0050]
Here, when another quantifier b does not exist on the right side of the same phrase, the quantifier a is handled in the same manner as a normal noun, and is multiplied by a predicate (step S7), which is output as a result of syntax or semantic analysis. (Step S6), the entire processing routine is terminated.
[0051]
On the other hand, when it is determined that another quantifier b exists on the right side of the same phrase, the semantic role of the classifier included in each quantifier is determined (step S2).
[0052]
As described above, in this embodiment, a classifier correspondence table describing attributes of a classifier constituting a quantifier, that is, a semantic role, and a rule for determining the semantic role of the classifier according to the connection relationship of the quantifier in the sentence are described. Since the connection rules of the classifiers are prepared, the system can easily acquire the semantic role of the classifier included in each quantifier by referring to the correspondence table of the classifiers and the connection rules of the classifiers. it can.
[0053]
Then, it is determined whether the semantic roles of the quantifier a and the quantifier b are the same (step S3).
[0054]
Here, when the semantic roles of the quantifier a and the quantifier b do not match, the quantifier a is handled in the same manner as a normal noun, and is applied to a predicate (step S7), which is output as a semantic analysis result (step S7). S6) The entire processing routine is terminated.
[0055]
On the other hand, when the semantic roles of the quantifier a and the quantifier b match, the quantifier a is multiplied by the quantifier b (step S4).
[0056]
Then, it is further determined whether at least one of the quantifier a or the quantifier b has a plurality of semantic roles (step S5).
[0057]
Here, when at least one of the quantifier a or the quantifier b has a plurality of semantic roles, the semantic role of the quantifier a that appears with a case particle is ambiguous, and the quantifier a is multiplied by the phrase predicate. A dependency relationship is further added (step S7), two kinds of syntax or semantic analysis results are output (step S6), and the entire processing routine is terminated.
[0058]
Hereinafter, a specific example of the processing result of the quantifier according to the determination processing procedure of the dependency relationship of the case particles used in combination with the quantifier illustrated in FIG. 4 will be described.
[0059]
For example, in the case of the example sentence “He is broken once on May 3”, “May 3” is extracted as a quantifier with a case particle. Since this quantifier has two or more quantifiers having a semantic role “date and time” in succession, the semantic role can be limited to “date and time” according to the connection rule of the classifier. Further, another quantifier “one time” is extracted in the same sentence, and it is understood that this quantifier has “number of times” as a semantic role from the correspondence table of the classifier. And since the meaning content of the two quantifiers appearing in the same sentence is different, the quantifier “May 3” with a case particle is treated in the same way as an ordinary noun, and the predicate “I am in love”. Is determined to be qualified.
[0060]
Therefore, if there is a question “How much is he in love”, the answer based on this example sentence should be “once”. FIG. 5 shows a parse tree of this example sentence. FIG. 6 shows a semantic analysis result (f-structure) based on the syntax analysis tree shown in FIG.
[0061]
Also, in the case of the example sentence “He is broken once every three times”, “three times” is extracted as a quantifier with a case particle. It can be seen that this quantifier has “number of times” as a semantic role from the correspondence table of the classifier. Further, another quantifier “one time” is extracted in the same sentence, and it is understood that this quantifier has “number of times” as a semantic role from the correspondence table of the classifier. And since the two quantifiers appearing in the same sentence have the same meaning, the quantifier “3 times” with a case particle is modified with another quantifier “1” in the quantifier sentence that follows in the same sentence. It is judged that
[0062]
Thus, if there is a similar question, “How much is he in love”, the answer based on this example sentence should be “once every three times”. FIG. 7 shows a parse tree of this example sentence. FIG. 8 shows a semantic analysis result (f-structure) based on the syntax analysis tree shown in FIG.
[0063]
Further, in the case of the example sentence “He is broken once every 3 days”, “3 days” is extracted as a quantifier with a case particle. It can be seen from the classifier correspondence table that this quantifier has “period” and “frequency” in addition to “date and time” as a semantic role. Further, another quantifier “one time” is extracted in the same sentence, and it is understood that this quantifier has “number of times” as a semantic role from the correspondence table of the classifier. In this case, the quantifier “3 days” with a case particle has a plurality of semantic roles. However, even if the connection rule of the classifier is referred to, the corresponding rule is not found, so it cannot be specified as one semantic role.
[0064]
First, the quantifier “3 days” with a case particle is the same as the quantifier “3 days” with a case particle because both the quantifier “3 days” with a case particle and another quantifier “1 time” that follows in the same phrase have frequency. It is determined that another quantifier “one time” in the quantifier sentence that follows in the sentence is modified.
[0065]
Further, since the quantifier “3 days” with a case particle has a plurality of semantic roles, it is further determined that the predicate “I am in love” has been modified.
[0066]
That is, in the case of this example sentence, the quantifier with a case particle itself is ambiguous, so the answer to the same question sentence “How much is he broken?” Both cases of “once a day” are conceivable. FIG. 9 shows one syntax analysis result for this example sentence. FIG. 10 shows a semantic analysis result (f-structure) based on the syntax analysis tree shown in FIG. FIG. 11 shows another syntax analysis result for this example sentence. FIG. 12 shows a semantic analysis result (f-structure) based on the syntax analysis tree shown in FIG.
[0067]
[Supplement]
The present invention has been described in detail above with reference to specific embodiments. However, it is obvious that those skilled in the art can make modifications and substitutions of the embodiment without departing from the gist of the present invention.
[0068]
Although the present embodiment will be described based on LFG grammar theory, of course, the present invention can be similarly applied to an analysis system having other grammar rules.
[0069]
In short, the present invention has been disclosed in the form of exemplification, and the description of the present specification should not be interpreted in a limited manner. In order to determine the gist of the present invention, the claims section described at the beginning should be considered.
[0070]
【The invention's effect】
As described above in detail, according to the present invention, an excellent natural language processing system and natural language processing method capable of accurately grasping the dependency relationship between words and performing more accurate syntax / semantic analysis processing As well as computer programs.
[0071]
In addition, according to the present invention, an excellent natural language processing system capable of accurately grasping the dependency relationship when a case particle is used in combination with a quantifier and performing more accurate syntax / semantic analysis processing And a natural language processing method and a computer program can be provided.
[0072]
According to the present invention, when a case particle is used after a quantifier, when a quantifier appears with a case particle in a sentence, when another quantifier is present in the sentence, the attribute of the classifier constituting each quantifier Based on this, the dependency relationship can be determined, and the processing accuracy of the syntax meaning / analysis can be improved. As a result, a useful result can be obtained when the analysis result is used in, for example, a search system or a question answering system.
[Brief description of the drawings]
FIG. 1 is a diagram schematically showing a configuration of a natural language processing system 1 based on LFG.
FIG. 2 is a diagram showing c-structure obtained as a result of processing an input sentence “My daughter speaks English” by the syntactic / semantic analysis unit 1;
FIG. 3 is a diagram showing f-structure obtained as a result of processing an input sentence “My daughter speaks English” by the syntactic / semantic analysis unit 1;
FIG. 4 is a flowchart showing a processing procedure for determining a dependency relationship of a case particle used in combination with a quantifier.
FIG. 5 is a diagram showing a parse tree for an example sentence “He is broken once on May 3.” FIG.
6 is a diagram showing a semantic analysis result (f-structure) based on the parse tree shown in FIG. 5. FIG.
FIG. 7 is a diagram showing a parse tree for an example sentence “He is broken once every three times”.
FIG. 8 is a diagram showing a semantic analysis result (f-structure) based on the syntax analysis tree shown in FIG. 7;
FIG. 9 is a diagram showing a parse tree for an example sentence “He is broken once every three days”.
FIG. 10 is a diagram showing a semantic analysis result (f-structure) based on the syntax analysis tree shown in FIG. 9;
FIG. 11 is a diagram showing another parse tree for the example sentence “He is broken once every three days”.
12 is a diagram showing a semantic analysis result (f-structure) based on the syntax analysis tree shown in FIG. 11. FIG.
[Explanation of symbols]
DESCRIPTION OF SYMBOLS 1 ... Natural language processing system 2 ... Morphological analysis part 2A ... Morphological rule, 2B ... Morphological dictionary 3 ... Syntactic / semantic analysis part 3A ... Grammar rule, 3B ... Valency dictionary

Claims

A natural language processing system that performs semantic analysis of sentences in which quantifiers appear with case particles,
Quantifier determining means for determining whether another quantifier exists behind the same phrase;
A semantic role determination means for determining a semantic role of a classifier included in each quantifier when it is determined that another quantifier exists behind the same phrase;
Semantic analysis means for determining the dependency relationship of quantifiers that appear with case particles according to the semantic role of each classifier in the sentence;
With
The semantic role determination means includes a correspondence table describing a correspondence relationship between a classifier constituting a quantifier and one or more semantic roles of the classifier, and one semantic role of the classifier according to the connection relationship of the quantifier in the sentence. And a connection rule of the classifier describing the rule for specifying the classifier, and subtracting the semantic role of the classifier included in each quantifier existing in the same phrase in the correspondence table of the classifier, and the classifier has a plurality of semantic roles. Specify one semantic role of each classifier according to the connection rules of the classifiers,
When the semantic role of the quantifier appearing behind the same phrase as the quantifier appearing with the case particle does not match the semantic analysis means, the quantifier appearing with the case particle is multiplied by the predicate in the sentence, When the semantic role of a quantifier appearing behind the same phrase as the appearing quantifier matches, the quantifier appearing with a case particle is multiplied by the quantifier appearing behind, but it is judged that there is no other quantifier after the same phrase. The quantifier that appears with a case particle is multiplied by the predicate in the sentence,
Natural language processing system characterized by that.

The semantic role determination means is a connection rule of a classifier describing a rule for determining a semantic role of a classifier.
(1) If a quantifier having two or more semantic roles “date and time” continues, limit the semantic role of the quantifier to “date and time”.
(2) If the suffix “ma” is connected to a quantifier with the semantic role “date and time”, the semantic role is limited to “period”.
(3) If the quantifier of a quantifier having the semantic role “frequency” starts with “−”, the semantic role is limited to “frequency”.
Natural language processing system according to claim 1, characterized in that it comprises a.

The semantic analysis means is a case where a quantifier appearing behind the same phrase as a quantifier appearing with a case particle coincides, and when at least one quantifier has a plurality of semantic roles, Multiply the quantifier that appears along with the predicate in the sentence,
The natural language processing system according to claim 1.

A natural language processing system that performs semantic analysis on an input sentence in which two or more quantifiers appear,
A correspondence table describing a correspondence relationship between a classifier constituting a quantifier and one or more semantic roles of the classifier, and a rule for specifying one semantic role of the classifier according to the connection relationship of the quantifier in the sentence And when it is determined that there is another quantifier behind the same phrase, the semantic role of the classifier included in each quantifier is subtracted from the correspondence table of the classifier. Semantic role determination means for specifying one semantic role of each classifier according to the connection rule of the classifier when having a plurality of semantic roles;
Semantic analysis means for semantic analysis of the input sentence according to the semantic role of each classifier;
With
The semantic analysis means includes a quantifier that appears behind the same phrase as a quantifier that appears with a case particle. When the semantic roles do not match, the quantifier that appears with the case particle is multiplied by the predicate in the sentence, and when the semantic role of the quantifier that appears after the same phrase as the quantifier that appears with the case particle matches, Multiply the quantifier that appears with the quantifier that appears later, but if it is determined that there is no other quantifier after the same phrase, multiply the quantifier that appears with the case particle by the predicate in the sentence.
Natural language processing system characterized by that.

The classifier connection rule is a rule for determining the semantic role of a classifier.
(1) If a quantifier having two or more semantic roles “date and time” continues, limit the semantic role of the quantifier to “date and time”.
(2) If the suffix “ma” is connected to a quantifier with the semantic role “date and time”, the semantic role is limited to “period”.
(3) If the quantifier of a quantifier having the semantic role “frequency” starts with “−”, the semantic role is limited to “frequency”.
The natural language processing system according to claim 4, wherein:

A natural language processing method for performing semantic analysis on a sentence in which a quantifier appears with a case particle on a natural language processing system constructed on a computer,
A quantifier determining means provided in the computer, a quantifier determining step of determining whether another quantifier exists behind the same phrase;
When the semantic role determination means provided in the computer determines that another quantifier is present behind the same phrase, a semantic role determination step of determining a semantic role of a classifier included in each quantifier;
The semantic analysis means provided in the computer, according to the semantic role of each classifier in the sentence, a semantic analysis step of determining the dependency relationship of the quantifier that appears with a case particle;
With
In the semantic role determination step, a semantics correspondence table describing a correspondence relationship between a classifier constituting a quantifier and one or more semantic roles of the classifier, and one semantic role of the classifier according to the connection relationship of the quantifier in the sentence. And a connection rule of the classifier describing the rule for specifying the classifier, and subtracting the semantic role of the classifier included in each quantifier existing in the same phrase in the correspondence table of the classifier, and the classifier has a plurality of semantic roles. Specify one semantic role of each classifier according to the connection rules of the classifiers,
In the semantic analysis step, when the semantic role of the quantifier appearing behind the same phrase as the quantifier appearing with the case particle does not match, the quantifier appearing with the case particle is multiplied by the predicate in the sentence, When the semantic role of the quantifier appearing behind the same phrase as the appearing quantifier matches, the quantifier appearing with the case particle is multiplied by the quantifier appearing behind, but it is judged that there is no other quantifier after the same phrase. The quantifier that appears with a case particle is multiplied by the predicate in the sentence,
A natural language processing method comprising:

A natural language processing method for performing semantic analysis on an input sentence in which two or more quantifiers appear on a natural language processing system constructed on a computer,
When the semantic role determination means provided in the computer determines that another quantifier exists behind the same phrase, the classifier and the classifier that constitute the quantifier have the semantic role of the classifier included in each quantifier. In accordance with the connection rule of the classifier describing the rule for specifying one semantic role of the classifier according to the connection relationship of the quantifier in the sentence, as well as drawing in the correspondence table of the classifier describing the correspondence of one or more semantic roles A semantic role determination step for determining the semantic role of the classifier;
Semantic analysis means provided in the computer, a semantic analysis step for semantic analysis of the input sentence according to the semantic role of each classifier,
With
In the semantic role determination step, when it is determined that another quantifier exists behind the same phrase, the semantic role of the classifier included in each quantifier is looked up in the correspondence table of the classifier, and the classifier has a plurality of semantic roles. Specify one semantic role of each classifier according to the connection rule of the classifier,
In the semantic analysis step, when the semantic role of the quantifier appearing behind the same phrase as the quantifier appearing with the case particle does not match, the quantifier appearing with the case particle is multiplied by the predicate in the sentence, When the semantic role of the quantifier appearing behind the same phrase as the appearing quantifier matches, the quantifier appearing with the case particle is multiplied by the quantifier appearing behind, but it is judged that there is no other quantifier after the same phrase. The quantifier that appears with a case particle is multiplied by the predicate in the sentence,
A natural language processing method characterized by that.

A computer program written in a computer-readable format to execute natural language processing on a computer for semantic analysis of a sentence in which a quantifier appears with a case particle, the computer comprising:
Quantifier determining means for determining whether another quantifier exists behind the same phrase;
A semantic role determination means for determining a semantic role of a classifier included in each quantifier when it is determined that another quantifier exists behind the same phrase;
Semantic analysis means for determining the dependency relationship of quantifiers that appear with case particles according to the semantic role of each classifier in the sentence;
A computer program for functioning as
The semantic role determination means includes a correspondence table describing a correspondence relationship between a classifier constituting a quantifier and one or more semantic roles of the classifier, and one semantic role of the classifier according to the connection relationship of the quantifier in the sentence. And a connection rule of the classifier describing the rule for specifying the classifier, and subtracting the semantic role of the classifier included in each quantifier existing in the same phrase in the correspondence table of the classifier, and the classifier has a plurality of semantic roles. Specify one semantic role of each classifier according to the connection rules of the classifiers,
When the semantic role of the quantifier appearing behind the same phrase as the quantifier appearing with the case particle does not match the semantic analysis means, the quantifier appearing with the case particle is multiplied by the predicate in the sentence, When the semantic role of the quantifier appearing behind the same phrase as the appearing quantifier matches, the quantifier appearing with the case particle is multiplied by the quantifier appearing behind, but it is judged that there is no other quantifier after the same phrase. The quantifier that appears with a case particle is multiplied by the predicate in the sentence,
A computer program characterized by the above.

A computer program written in a computer-readable format so as to execute natural language processing for performing semantic analysis on an input sentence in which two or more quantifiers appear on a computer, the computer comprising:
When it is determined that another quantifier exists behind the same phrase, the semantic role of the classifier included in each quantifier is described, and the correspondence between the classifier constituting the quantifier and one or more semantic roles of the classifier is described. Semantic Role Judgment for Determining the Semantic Role of Each Classifier in accordance with the Connection Rule of a Classifier Describes a Rule for Specifying One Semantic Role of a Classifier in accordance with the Connection Relationship of Quantifiers in a Sentence Means,
Let it function as a semantic analysis means to analyze the input sentence according to the semantic role of each classifier,
The semantic role determination means, when it is determined that another quantifier exists behind the same phrase, subtracts the semantic role of the classifier included in each quantifier in the correspondence table of the classifier, and the classifier has a plurality of semantic roles. Specify one semantic role of each classifier according to the connection rule of the classifier,
When the semantic role of the quantifier appearing behind the same phrase as the quantifier appearing with the case particle does not match the semantic analysis means, the quantifier appearing with the case particle is multiplied by the predicate in the sentence, When the semantic role of the quantifier appearing behind the same phrase as the appearing quantifier matches, the quantifier appearing with the case particle is multiplied by the quantifier appearing behind, but it is judged that there is no other quantifier after the same phrase. The quantifier that appears with a case particle is multiplied by the predicate in the sentence,
A computer program characterized by the above.