WO2020054465A1 - Problem solution assistance device and method therefor - Google Patents

Problem solution assistance device and method therefor Download PDF

Info

Publication number
WO2020054465A1
WO2020054465A1 PCT/JP2019/034221 JP2019034221W WO2020054465A1 WO 2020054465 A1 WO2020054465 A1 WO 2020054465A1 JP 2019034221 W JP2019034221 W JP 2019034221W WO 2020054465 A1 WO2020054465 A1 WO 2020054465A1
Authority
WO
WIPO (PCT)
Prior art keywords
word
score
occurrence
document
dictionary
Prior art date
Application number
PCT/JP2019/034221
Other languages
French (fr)
Japanese (ja)
Inventor
高明 森谷
裕之 冨士井
学 西尾
直規 立石
吉田　敦
Original Assignee
日本電信電話株式会社
Priority date (The priority date is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the date listed.)
Filing date
Publication date
Application filed by 日本電信電話株式会社 filed Critical 日本電信電話株式会社
Publication of WO2020054465A1 publication Critical patent/WO2020054465A1/en

Links

Images

Classifications

    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06F—ELECTRIC DIGITAL DATA PROCESSING
    • G06F16/00—Information retrieval; Database structures therefor; File system structures therefor

Definitions

  • the present invention relates to a problem solving support device and a method for classifying a document input by a user and supporting creation thereof.
  • Patent Document 1 As a technique for classifying information input by a user, for example, a technique disclosed in Patent Document 1 is known.
  • the technology disclosed in Patent Literature 1 is for supporting a user to find a desired service from a large number of Web services.
  • the method classifies services in consideration of features related to intents that express the intentions of service providers by natural language expressions, and assists users in finding services.
  • AI Artificial Intelligence
  • AI can be used to solve business problems.
  • problems that can be solved by conventional system development such as batch processing and intersystem cooperation
  • problems that must be solved by AI such as machine learning and mathematical optimization.
  • the problem to be solved is first expressed in, for example, a document, and the document is often written by confusing the above two problems.
  • the problem to be solved is first expressed in, for example, a document, and the document is often written by confusing the above two problems.
  • Patent Document 1 cannot support creation of a document created by a user, for example, representing a problem to be solved. That is, there is a problem that an apparatus and a method for supporting creation of a document representing a problem to be solved do not yet exist.
  • the present invention has been made in view of this problem, and an object of the present invention is to provide a problem solving support device and a method thereof that can support creation of a document representing a problem to be solved by a user.
  • a problem solving support device is a problem solving support device that supports creation of a document representing a problem to be solved, and refers to a recommended word display unit on which a recommended word to be included in the document is displayed.
  • a word co-occurrence extraction unit that extracts a noun and a verb from each of the morphologically analyzed phrases and extracts the word co-occurrence if the word co-occurrence including the extracted word is stored in the dictionary;
  • a score determining unit that adds together the AI score and the system score of the co-occurrence of the written word, and determines whether the combined AI score or the system score is greater, and each word of the word co-occurrence extracted by referring to the dictionary
  • a recommended word extracting unit that extracts the recommended word that is the word with the highest score of the type determined to be large by the score determining unit and that is not included in the phrase among the words linked to Make a summary.
  • the problem solving support method is a problem solving supporting method executed by a problem solving support device that supports creation of a document representing a problem to be solved, wherein the problem should be solved by AI.
  • a word co-occurrence which is a combination of two different words with an AI score and a system score, in which the sum of both is a constant or a problem to be solved in system development, is a constant, is stored.
  • a document input step for inputting the document created by a user with reference to a recommended word display unit, which includes a dictionary and displaying recommended words to be included in the document, and morphologically analyzing the document, A noun and a verb are extracted from each clause obtained by morphologically analyzing the document, and the word co-occurrence including the extracted word is stored in the dictionary.
  • a word co-occurrence extraction step of extracting the word co-occurrence and a score determining step of summing the AI score and the system score of the extracted word co-occurrence and determining which of the summed AI score and the system score is larger
  • a score determining step of summing the AI score and the system score of the extracted word co-occurrence and determining which of the summed AI score and the system score is larger
  • the word having the highest score of the type determined to be large in the score determination step and the phrase And extracting a recommended word that does not include the recommended word.
  • FIG. 2 is a flowchart showing a processing procedure of the problem solving support device shown in FIG.
  • FIG. 2 is a diagram schematically illustrating an example of sentences and words displayed on a document input unit and a recommended word display unit of the problem solving support device illustrated in FIG. 1.
  • FIG. 2 is a diagram illustrating an example of a word extracted from a phrase and a word co-occurrence in a word co-occurrence extraction unit of the problem solving support device illustrated in FIG.
  • FIG. 2 is a diagram illustrating word co-occurrence extracted by a word co-occurrence extraction unit of the problem solving support device illustrated in FIG. 1.
  • FIG. 2 is a diagram schematically illustrating processing of a recommended word extracting unit of the problem solving support device illustrated in FIG. 1. It is a block diagram showing the example of functional composition of the problem solving support device concerning a 2nd embodiment of the present invention. 9 is a flowchart showing a processing procedure of the problem solving support device shown in FIG. It is a figure showing an example of newly generated word co-occurrence (unregistered word co-occurrence).
  • FIG. 1 is a block diagram illustrating a functional configuration example of the problem solving support device according to the first embodiment of the present invention.
  • the problem solving support device 100 shown in FIG. 1 supports creation of a document representing a problem to be solved which is input by a user.
  • the problem solving support device 100 includes a document input unit 10, a morphological analysis unit 20, a dictionary 30, a word co-occurrence extraction unit 40, a score determination unit 50, a recommended word extraction unit 60, and a recommended word display unit 70.
  • Each functional component of the problem solving support device 100 is realized by, for example, a computer including a ROM, a RAM, a CPU, and the like. When each functional component is implemented by a computer, the processing content of the function that each functional component should have is described by a program.
  • the problem solving support device 100 may be configured by a client server system.
  • the document input unit 10 is a client constituted by one personal computer (hereinafter, PC), and the other functional components are servers.
  • the client and the server may be connected via a network, and a plurality of clients may be connected to the server via the network.
  • FIG. 2 is a flowchart showing a processing procedure of the problem solving support device 100. Hereinafter, the operation will be described with reference to FIGS. 1 and 2 and other drawings.
  • the document input unit 10 inputs a document created by the user while referring to the recommended word display unit 70 on which a recommended word to be included in the document is displayed (step S1).
  • the document input unit 10 includes, for example, one PC, and a user inputs a document using a keyboard (not shown).
  • the recommended word display unit 70 includes a display panel (not shown) of the PC, and a document input by a user is also displayed on the display panel.
  • FIG. 3 is a diagram schematically showing the document input unit 10 and the recommended word display unit 70.
  • FIG. 3 (a) shows a state in which a document representing a problem to be solved which is input by a user is input, for example, "classify characteristics of traffic growth from past traffic data, data on the number of users, etc.” (Step S1).
  • the recommended word display section 70 in this state does not display anything.
  • the morphological analysis unit 20 morphologically analyzes the input document and outputs the morpheme of the document (step S3).
  • Morphological analysis is to decompose a document into the smallest meaningful units (words) and determine parts of speech.
  • the morphological analysis unit 20 can be configured by a well-known morphological analysis engine (processing device).
  • the word co-occurrence extraction unit 40 extracts a noun and a verb from a phrase obtained by morphologically analyzing a document. Since the above example sentence "from past traffic data, user count data, etc.” does not include a verb, no word is extracted from this part. Since the next phrase includes a noun and a verb, for example, four words of the noun “traffic”, the verb “extension”, the noun “feature”, and the verb “classify” are extracted.
  • "classify” is composed of two morphemes, a noun “classify” and a verb "shi".
  • the noun immediately before the verb is treated as a verb in combination. Since the morpheme immediately before the verb "elongation" is the particle "no" in this example, only "elongation" is treated as a verb.
  • the word co-occurrence extraction unit 40 generates a word co-occurrence that is a combination of all two words of the extracted word. Assuming that there are u extracted words, the number n of word co-occurrences is expressed by the following equation.
  • FIG. 4 shows the generated n word co-occurrences P (n).
  • the word co-occurrence extraction unit 40 extracts the word co-occurrence Q (m) stored in the dictionary 30 from the generated word co-occurrence P (n) (step S4). That is, if the word co-occurrence P (i) generated from the input document is stored in the dictionary 30, the word co-occurrence extraction unit 40 converts the stored word co-occurrence P (i) into the word co-occurrence P (i). Copy to Q (j).
  • FIG. 5 is a diagram showing a word co-occurrence Q (j) obtained by copying the word co-occurrence P (i) stored in the dictionary 30.
  • the dictionary 30 indicates whether the solved problem input by the user is a problem to be solved by AI or a problem to be solved by system development.
  • the sum of both the AI score and the system score is 1.0.
  • Word co-occurrence which is a set of assigned words, is stored. In the present embodiment, an example will be described in which the sum is 1.0, but the sum may be a determined constant. For example, the sum may be a constant such as 10 or 100.
  • the AI score of the word co-occurrence P (1) of “traffic” and “extension” is 0.8, and the system score is 0.2. Further, the AI score of the word co-occurrence Q (2) of “feature” and “classify” is 0.9, and the system score is 0.1.
  • the dictionary 30 is created in advance.
  • the score determination unit 50 adds together the AI score and the system score of the word co-occurrence Q (j) extracted by the word co-occurrence extraction unit 40 (Step S5), and determines which of the combined AI score and the system score is larger.
  • Judge steps S6, S7). If the document input by the user is “past traffic data,... Omitting,... Categorized” and the recommendation execution button 11 is pressed, the sum of the AI scores is 1.7 and the sum of the system score is 0.3. It is.
  • step S7 it is determined that the AI score is larger than the system score (AI score> system score) (YES in step S7). If the combined AI score and the system score are equal (NO in step S6), the process returns to step S1 for continuously inputting a document. In this case, a message urging the user to continue to input a document may be displayed on the display panel.
  • the score determination unit 50 may output the determined result and the determined document to the outside. By attaching the judgment result to the document, a recognition gap does not occur between the persons in charge of solving the problem.
  • the recommended word extraction unit 60 decomposes the word co-occurrence Q (j) into words. Then, for each word, the linked word is searched in the dictionary 30, and the words are arranged in the descending order of the score.
  • FIG. 6 is a diagram schematically showing words linked in the dictionary 30.
  • “Traffic” is associated with, for example, “system”, “growth”, and “prediction”. The linking between words is performed by a person in advance.
  • FIG. 7 is a diagram schematically illustrating a state in which the word co-occurrence Q (j) is decomposed into words, and the linked word co-occurrences are arranged in the order of higher scores.
  • traffic is M1.
  • the word M11 having the highest AI score linked to M1 shown in FIG. The AI score is 0.8
  • the second M12 is "prediction (AI score for co-occurrence with M1 is 0.7)
  • the third M13 is "system (AI score for co-occurrence with M1 is 0.2)” .
  • step S7 If it is determined that the system score is higher than the AI score (system score> AI score) (NO in step S7), the recommended words are similarly extracted from the arrangement in the descending order of the system score. (Step S9).
  • the extracted recommended words are displayed on the recommended word display unit 70.
  • the user creates a subsequent document while referring to the recommended word display unit 70 and inputs the document to the document input unit 10.
  • FIG. 3B is a diagram showing a state in which recommended words are displayed on the recommended word display unit 70.
  • FIG. 3C is a diagram illustrating an example of a subsequent sentence input by the user while referring to the recommended words. As shown in FIG. 3C, by creating a subsequent document using the recommended word, the document is guided to an expression (content) representing a type of task having a high score.
  • the problem solving support device 100 is a problem solving support device that supports creation of a document representing a problem to be solved, and is a recommended word in which a recommended word to be included in the document is displayed.
  • a document input unit 10 for inputting a document created by a user while referring to the display unit 70, a morphological analysis unit 20 for morphologically analyzing the above-mentioned document and outputting a morpheme of the document, and a problem solved by AI.
  • a word co-occurrence which is a combination of two different words to which an AI score and a system score are added, which represents the degree of whether the task is a task to be solved or a task to be solved in system development, is a constant.
  • the extraction unit 40, the AI score and the system score of the extracted word co-occurrence are added together, and the score determination unit 50 that determines which of the combined AI score and the system score is larger, and the dictionary 30 are referred to.
  • Word extraction unit that extracts a word that has the highest score of the type determined to be large by the score determination unit 50 and that is not included in a phrase among the words associated with each word of the co-occurred words. 60.
  • FIG. 8 is a block diagram illustrating a functional configuration example of the problem solving support device according to the second embodiment of the present invention.
  • the problem solving support device 200 shown in FIG. 8 differs from the problem solving supporting device 100 (FIG. 1) in that a word co-occurrence extraction unit 240 and an unregistered word co-occurrence recording unit 280 are provided.
  • FIG. 9 is a flowchart showing a processing procedure of the problem solving support device 200.
  • the flowchart shown in FIG. 9 shows the processing procedure after step S4 of the flowchart shown in FIG.
  • the word co-occurrence extraction unit 240 also extracts nouns and verbs that are not stored in the dictionary 30 from phrases obtained by morphological analysis of a document input by the user (YES in step S40).
  • the constant that is the sum of the AI score and the system score is set to 1.0.
  • the score determination unit 50 adds together the AI score and the system score of the extracted word co-occurrence, including the newly recorded word co-occurrence (unregistered word co-occurrence) (step S5).
  • Steps S5 to S9 are the processing steps described in FIG.
  • the unregistered word co-occurrence recording unit 280 re-records the newly recorded word co-occurrence with the AI score of +0.1 and the system score of -0.1 if the AI score is determined to be high after extracting the recommended word. (Step S43). If it is determined that the system score is high, the AI score of the newly recorded word co-occurrence is set to -0.1, and the system score is set to +0.1 and re-recorded (step S45).
  • FIG. 10 is a diagram showing an example of a newly recorded word co-occurrence (unregistered word co-occurrence).
  • FIG. 10A shows an example of the word co-occurrence recorded in step S41, and
  • FIG. Shows an example of the word co-occurrence re-recorded in step S43.
  • the AI score and system score of the newly recorded word co-occurrence are sequentially updated and automatically adjusted to an appropriate score value.
  • the score of the word co-occurrence stored in the dictionary 30 in advance is unchanged.
  • the problem solving support device 200 includes the word co-occurrence extraction unit 240 and the unregistered word co-occurrence recording unit 280, and the word co-occurrence extraction unit 240 Nouns and verbs that are included and not stored in the dictionary 30 are extracted, and the unregistered word co-occurrence recording unit 280 extracts the words that are not stored in the dictionary extracted by the word co-occurrence extraction unit 240 from the above phrases.
  • An unregistered word co-occurrence which is a set of other words taken out, is generated, and the score of one type determined to be large by the score determination unit 50 is increased, and the score of the other type is reduced and recorded in the dictionary 30. I do.
  • the problem solving support devices 100 and 200 when the problem is represented by a document for the purpose of solving a business problem, for example, the problem is a system development problem. Or the problem to be solved by AI can be described separately. As a result, there is no perception gap between persons in charge of task resolution.
  • the dictionary 30 includes machine learning and regression. Prediction, chat pot, optimal ratio, text mining, data mining, identification / classification, simulation, inference, reinforcement learning, clustering, recommendation, neighborhood search, statistical causal inference, knowledge expression / search / inference, probability inference, state estimation, Word co-occurrence (words) related to sequence labeling, feature extraction, etc. may be stored.
  • the present invention is not limited to the above embodiment, and can be modified within the scope of the gist.

Landscapes

  • Engineering & Computer Science (AREA)
  • Theoretical Computer Science (AREA)
  • Data Mining & Analysis (AREA)
  • Databases & Information Systems (AREA)
  • Physics & Mathematics (AREA)
  • General Engineering & Computer Science (AREA)
  • General Physics & Mathematics (AREA)
  • Machine Translation (AREA)
  • Information Retrieval, Db Structures And Fs Structures Therefor (AREA)

Abstract

The present invention assists creation of a document describing a problem to be solved. The present invention is provided with: a document input unit 10 that receives an input of a sentence created by a user referring to a recommended word display unit 70 on which a recommended word desired to be included in the document is displayed; a morphological analysis unit 20 that outputs morphemes in the sentence; a dictionary 30 that stores word co-occurrences which are word sets to each of which a score indicating a degree of whether a problem should be solved by AI or by system development, has been given; a word co-occurrence extraction unit 40 that extracts, from a paragraph of the sentence, word co-occurrences which are stored in the dictionary 30; a score determination unit 50 that sums the AI scores and the system scores of the extracted word co-occurrences, and determines which summed score is greater; and a recommended word extraction unit 60 that extracts, by referring to the dictionary 30, a word having the highest score of the type determined to have the greater score, among words associated with the words of the extracted word co-occurrences, and outputs the extracted word to the recommended word display unit 70.

Description

課題解決支援装置とその方法Problem solving support apparatus and method
　本発明は、利用者が入力する文書を分類してその作成を支援する課題解決支援装置とその方法に関する。 {Circle over (1)} The present invention relates to a problem solving support device and a method for classifying a document input by a user and supporting creation thereof.
　利用者が入力する情報を分類する技術としては、例えば特許文献１に開示されたものが知られている。特許文献１に開示された技術は、数多いＷｅｂサービスの中から、自分が望むサービスを発見できるように支援するものである。 As a technique for classifying information input by a user, for example, a technique disclosed in Patent Document 1 is known. The technology disclosed in Patent Literature 1 is for supporting a user to find a desired service from a large number of Web services.
　その方法は、サービス提供者の意図を自然言語表現により表したインテントに関する特徴を考慮してサービスを分類し、利用者のサービスの発見を支援するようにしたものである。 (4) The method classifies services in consideration of features related to intents that express the intentions of service providers by natural language expressions, and assists users in finding services.
特開２０１４－２１５６３３号公報JP 2014-215633 A
　人の活動を支援する分野では、昨今ＡＩ（Artificial Intelligence）が注目されている。ＡＩは、業務上の課題を解決する目的で使用することができる。しかし、業務上の課題は、大きく分けてバッチ処理やシステム間連携などの従来のシステム開発で解決できる課題と、機械学習及び数理最適化等のＡＩで解決すべき課題の二つがある。 In the field of supporting human activities, AI (Artificial Intelligence) has recently been receiving attention. AI can be used to solve business problems. However, there are two main tasks to be solved, which are problems that can be solved by conventional system development such as batch processing and intersystem cooperation, and problems that must be solved by AI such as machine learning and mathematical optimization.
　業務の効率化を図るのに、先ず解決したい課題を例えば文書で表すことになるが、その文書は上記の二つの課題を混同して書かれる場合が多い。その結果、課題解決を任された担当者間で認識のギャップが生じ、課題解決に時間を要するという問題がある。 課題 In order to improve work efficiency, the problem to be solved is first expressed in, for example, a document, and the document is often written by confusing the above two problems. As a result, there is a gap in recognition between the persons in charge of solving the problem, and there is a problem that it takes time to solve the problem.
　そこで、文書の作成中に、解決すべき課題が、ＡＩとシステム開発のどちらの範疇に有るのか分かれば適切な文書作成を支援することができる。しかしながら、特許文献１の技術では、利用者が作成する例えば解決したい課題を表す文書の作成を支援することはできない。つまり、解決したい課題を表す文書の作成を支援する装置及び方法は未だ存在しないという課題がある。 Therefore, if it is known during the creation of a document whether the problem to be solved falls into the category of AI or system development, appropriate document creation can be supported. However, the technique of Patent Document 1 cannot support creation of a document created by a user, for example, representing a problem to be solved. That is, there is a problem that an apparatus and a method for supporting creation of a document representing a problem to be solved do not yet exist.
　本発明は、この課題に鑑みてなされたものであり、利用者が作成する解決したい課題を表す文書の作成を支援することができる課題解決支援装置とその方法を提供することを目的とする。 The present invention has been made in view of this problem, and an object of the present invention is to provide a problem solving support device and a method thereof that can support creation of a document representing a problem to be solved by a user.
　本発明の一態様に係る課題解決支援装置は、解決したい課題を表す文書の作成を支援する課題解決支援装置であって、前記文書に含めたい推薦単語が表示される推薦単語表示部を参照しながら利用者が作成する前記文書が入力される文書入力部と、前記文書を形態素解析し、該文書の形態素を出力する形態素解析部と、前記課題がＡＩで解決すべき課題であるか又はシステム開発で解決すべき課題であるかの程度を表す両者の和が定数となるＡＩスコアとシステムスコアが付与された2つの異なる単語の組みである単語共起が記憶された辞書と、前記文書を形態素解析したそれぞれの文節から名詞と動詞を取り出し、該取り出した単語を含む前記単語共起が前記辞書に記憶されていれば該単語共起を抽出する単語共起抽出部と、抽出された前記単語共起のＡＩスコアとシステムスコアをそれぞれ合算し、合算したＡＩスコアとシステムスコアのどちらが大きいかを判定するスコア判定部と、前記辞書を参照し、抽出された前記単語共起の各単語に紐付けられた単語の中で、前記スコア判定部で大きいと判定された種別のスコアが最も高い単語で且つ前記文節に含まれない前記推薦単語を抽出する推薦単語抽出部とを備えることを要旨とする。 A problem solving support device according to an aspect of the present invention is a problem solving support device that supports creation of a document representing a problem to be solved, and refers to a recommended word display unit on which a recommended word to be included in the document is displayed. A document input unit into which the document created by the user is input, a morphological analysis unit that morphologically analyzes the document and outputs a morpheme of the document, and a system that determines whether the problem is to be solved by AI. A dictionary in which a word co-occurrence, which is a set of two different words to which an AI score and a system score are added, in which the sum of both representing a degree to be solved by the development is a constant, and the document, A word co-occurrence extraction unit that extracts a noun and a verb from each of the morphologically analyzed phrases and extracts the word co-occurrence if the word co-occurrence including the extracted word is stored in the dictionary; A score determining unit that adds together the AI score and the system score of the co-occurrence of the written word, and determines whether the combined AI score or the system score is greater, and each word of the word co-occurrence extracted by referring to the dictionary A recommended word extracting unit that extracts the recommended word that is the word with the highest score of the type determined to be large by the score determining unit and that is not included in the phrase among the words linked to Make a summary.
　また、本実施形態の一態様に係る課題解決支援方法は、解決したい課題を表す文書の作成を支援する課題解決支援装置が実行する課題解決支援方法であって、前記課題がＡＩで解決すべき課題であるか又はシステム開発で解決すべき課題であるかの程度を表す両者の和が定数となるＡＩスコアとシステムスコアが付与された2つの異なる単語の組みである単語共起が記憶された辞書を備え、前記文書に含めたい推薦単語が表示される推薦単語表示部を参照しながら利用者が作成する前記文書が入力される文書入力ステップと、前記文書を形態素解析し、該文書の形態素を出力する形態素解析ステップと、前記文書を形態素解析したそれぞれの文節から名詞と動詞を取り出し、該取り出した単語を含む前記単語共起が前記辞書に記憶されていれば該単語共起を抽出する単語共起抽出ステップと、抽出された前記単語共起のＡＩスコアとシステムスコアをそれぞれ合算し、合算したＡＩスコアとシステムスコアのどちらが大きいかを判定するスコア判定ステップと、前記辞書を参照し、抽出された前記単語共起の各単語に紐付けられた単語の中で、前記スコア判定ステップで大きいと判定された種別のスコアが最も高い単語で且つ前記文節に含まれない前記推薦単語を抽出する推薦単語抽出ステップとを行うことを要旨とする。 The problem solving support method according to an aspect of the present embodiment is a problem solving supporting method executed by a problem solving support device that supports creation of a document representing a problem to be solved, wherein the problem should be solved by AI. A word co-occurrence, which is a combination of two different words with an AI score and a system score, in which the sum of both is a constant or a problem to be solved in system development, is a constant, is stored. A document input step for inputting the document created by a user with reference to a recommended word display unit, which includes a dictionary and displaying recommended words to be included in the document, and morphologically analyzing the document, A noun and a verb are extracted from each clause obtained by morphologically analyzing the document, and the word co-occurrence including the extracted word is stored in the dictionary. For example, a word co-occurrence extraction step of extracting the word co-occurrence, and a score determining step of summing the AI score and the system score of the extracted word co-occurrence and determining which of the summed AI score and the system score is larger With reference to the dictionary, among the words linked to each of the extracted words of the word co-occurrence, the word having the highest score of the type determined to be large in the score determination step and the phrase And extracting a recommended word that does not include the recommended word.
　本発明によれば、利用者が作成する解決したい課題を表す文書の作成を支援することができる。 According to the present invention, it is possible to support creation of a document that represents a problem to be solved by a user.
本発明の第１実施形態に係る課題解決支援装置の機能構成例を示すブロック図である。It is a block diagram showing the example of functional composition of the problem solving support device concerning a 1st embodiment of the present invention. 図１に示す課題解決支援装置の処理手順を示すフローチャートである。2 is a flowchart showing a processing procedure of the problem solving support device shown in FIG. 図１に示す課題解決支援装置の文書入力部と推薦単語表示部に表示された文章及び単語の例を模式的に示す図である。FIG. 2 is a diagram schematically illustrating an example of sentences and words displayed on a document input unit and a recommended word display unit of the problem solving support device illustrated in FIG. 1. 図１に示す課題解決支援装置の単語共起抽出部において、文節から抽出した単語と単語共起の例を示す図である。FIG. 2 is a diagram illustrating an example of a word extracted from a phrase and a word co-occurrence in a word co-occurrence extraction unit of the problem solving support device illustrated in FIG. 図１に示す課題解決支援装置の単語共起抽出部で抽出された単語共起を示す図である。FIG. 2 is a diagram illustrating word co-occurrence extracted by a word co-occurrence extraction unit of the problem solving support device illustrated in FIG. 1. 辞書で紐付けられている単語と、単語共起に付与されたスコアを模式的に示す図である。It is a figure which shows typically the word linked with the dictionary and the score provided to the word co-occurrence. 図１に示す課題解決支援装置の推薦単語抽出部の処理を模式的に示す図である。FIG. 2 is a diagram schematically illustrating processing of a recommended word extracting unit of the problem solving support device illustrated in FIG. 1. 本発明の第２実施形態に係る課題解決支援装置の機能構成例を示すブロック図である。It is a block diagram showing the example of functional composition of the problem solving support device concerning a 2nd embodiment of the present invention. 図８に示す課題解決支援装置の処理手順を示すフローチャートである。9 is a flowchart showing a processing procedure of the problem solving support device shown in FIG. 新たに生成された単語共起（未登録単語共起）の例を示す図である。It is a figure showing an example of newly generated word co-occurrence (unregistered word co-occurrence).
　以下、本発明の実施形態について図面を用いて説明する。複数の図面中同一のものには同じ参照符号を付し、説明は繰り返さない。 Hereinafter, embodiments of the present invention will be described with reference to the drawings. The same components in the multiple drawings have the same reference characters allotted, and description thereof will not be repeated.
 〔第１実施形態〕
　図１は、本発明の第１実施形態に係る課題解決支援装置の機能構成例を示すブロック図である。図１に示す課題解決支援装置１００は、利用者が入力する解決したい課題を表す文書の作成を支援するものである。 [First Embodiment]
FIG. 1 is a block diagram illustrating a functional configuration example of the problem solving support device according to the first embodiment of the present invention. The problem solving support device 100 shown in FIG. 1 supports creation of a document representing a problem to be solved which is input by a user.
　課題解決支援装置１００は、文書入力部１０、形態素解析部２０、辞書３０、単語共起抽出部４０、スコア判定部５０、推薦単語抽出部６０、及び推薦単語表示部７０を備える。課題解決支援装置１００の各機能構成部は、例えば、ＲＯＭ、ＲＡＭ、ＣＰＵ等からなるコンピュータで実現される。各機能構成部をコンピュータによって実現する場合、各機能構成部が有すべき機能の処理内容はプログラムによって記述される。 The problem solving support device 100 includes a document input unit 10, a morphological analysis unit 20, a dictionary 30, a word co-occurrence extraction unit 40, a score determination unit 50, a recommended word extraction unit 60, and a recommended word display unit 70. Each functional component of the problem solving support device 100 is realized by, for example, a computer including a ROM, a RAM, a CPU, and the like. When each functional component is implemented by a computer, the processing content of the function that each functional component should have is described by a program.
　なお、課題解決支援装置１００は、クライアントサーバシステムで構成しても良い。その場合、例えば、文書入力部１０は一つのパーソナルコンピュータ（以降、ＰＣ））で構成されるクライアントとなり、それ以外の機能構成部がサーバとなる。クライアントとサーバは、ネットワークで接続され、ネットワークを介して複数のクライアントがサーバに接続されるようにしても良い。 The problem solving support device 100 may be configured by a client server system. In this case, for example, the document input unit 10 is a client constituted by one personal computer (hereinafter, PC), and the other functional components are servers. The client and the server may be connected via a network, and a plurality of clients may be connected to the server via the network.
　図２は、課題解決支援装置１００の処理手順を示すフローチャートである。ここからは、図１、図２、及び他の図も参照してその動作を説明する。 FIG. 2 is a flowchart showing a processing procedure of the problem solving support device 100. Hereinafter, the operation will be described with reference to FIGS. 1 and 2 and other drawings.
　文書入力部１０は、文書に含めたい推薦単語が表示される推薦単語表示部７０を参照しながら利用者が作成する文書が入力される（ステップＳ１）。文書入力部１０は、例えば一つのＰＣで構成され、利用者は文書をキーボード（図示せず）で入力する。推薦単語表示部７０は、そのＰＣの表示パネル（図示せず）で構成され、利用者が入力する文書もその表示パネルに表示される。 The document input unit 10 inputs a document created by the user while referring to the recommended word display unit 70 on which a recommended word to be included in the document is displayed (step S1). The document input unit 10 includes, for example, one PC, and a user inputs a document using a keyboard (not shown). The recommended word display unit 70 includes a display panel (not shown) of the PC, and a document input by a user is also displayed on the display panel.
　図３は、文書入力部１０と推薦単語表示部７０を模式的に示す図である。図３（ａ）は、利用者が入力する解決したい課題を表す文書を構成する例えば「過去のトラフィックデータ、ユーザ数データ等から、トラフィックの伸びの特徴を分類し、」の文章を入力した状態を示している（ステップＳ１）。この状態の推薦単語表示部７０は何も表示していない。 FIG. 3 is a diagram schematically showing the document input unit 10 and the recommended word display unit 70. FIG. 3 (a) shows a state in which a document representing a problem to be solved which is input by a user is input, for example, "classify characteristics of traffic growth from past traffic data, data on the number of users, etc." (Step S1). The recommended word display section 70 in this state does not display anything.
　利用者が、上記の文書を入力した段階で、例えば表示パネルに表示された推薦実行ボタン１１を押下する（ステップＳ２のＹＥＳ）。そうすると形態素解析部２０は、入力済みの文書を形態素解析し、該文書の形態素を出力する（ステップＳ３）。 (4) At the stage where the user has input the above document, for example, the recommendation execution button 11 displayed on the display panel is pressed (YES in step S2). Then, the morphological analysis unit 20 morphologically analyzes the input document and outputs the morpheme of the document (step S3).
　形態素解析は、文書を意味のある最小の単位（単語）に分解し、品詞などを判別することである。形態素解析部２０は、周知の形態素解析エンジン（処理装置）で構成することができる。 Morphological analysis is to decompose a document into the smallest meaningful units (words) and determine parts of speech. The morphological analysis unit 20 can be configured by a well-known morphological analysis engine (processing device).
　単語共起抽出部４０は、文書を形態素解析した文節から名詞と動詞を取り出す。上記の例文の「過去のトラフィックデータ、ユーザ数データ等から、」までは、動詞を含まないのでこの部分からは単語が取り出されない。次の文節は名詞と動詞を含むので、例えば名詞の「トラフィック」、動詞の「伸び」、名詞の「特徴」、動詞の「分類し」の4つの単語が取り出される。ここで、「分類し」は、名詞の「分類」と動詞の「し」の2つの形態素から構成される。本実施形態では、動詞の直前の名詞は、組み合わせて動詞として扱う。なお、動詞の「伸び」の直前の形態素は、この例では助詞の「の」であるので「伸び」のみを動詞として扱う。 The word co-occurrence extraction unit 40 extracts a noun and a verb from a phrase obtained by morphologically analyzing a document. Since the above example sentence "from past traffic data, user count data, etc." does not include a verb, no word is extracted from this part. Since the next phrase includes a noun and a verb, for example, four words of the noun “traffic”, the verb “extension”, the noun “feature”, and the verb “classify” are extracted. Here, "classify" is composed of two morphemes, a noun "classify" and a verb "shi". In the present embodiment, the noun immediately before the verb is treated as a verb in combination. Since the morpheme immediately before the verb "elongation" is the particle "no" in this example, only "elongation" is treated as a verb.
　次に単語共起抽出部４０は、取り出した単語の全ての2単語の組み合わせである単語共起を生成する。抜き出された単語がｕ個あったとすると、単語共起の数ｎは次式で表される。 Next, the word co-occurrence extraction unit 40 generates a word co-occurrence that is a combination of all two words of the extracted word. Assuming that there are u extracted words, the number n of word co-occurrences is expressed by the following equation.
Figure JPOXMLDOC01-appb-M000001
  
　図４は、生成されたｎ個の単語共起Ｐ（ｎ）を示す。図４に示すように単語共起Ｐは、Ｐ（１）＝{Ｗ１，Ｗ２}，…Ｐ（ｎ）＝{Ｗｕ－１，Ｗｕ}と表せる。入力済みの文書から、上記の4つの単語が取り出された場合は、Ｐ（１）＝{トラフィック，伸び}、Ｐ（２）＝{伸び，特徴}、Ｐ（３）＝{トラフィック，特徴}、Ｐ（４）＝{トラフィック，分類し}、Ｐ（５）＝{伸び，分類し}、Ｐ（６）＝{特徴，分類し}の単語共起が生成される。
Figure JPOXMLDOC01-appb-M000001
FIG. 4 shows the generated n word co-occurrences P (n). As shown in FIG. 4, the word co-occurrence P can be expressed as P (1) = {W1, W2},... P (n) = {W u−1 , W u }. When the above four words are extracted from the input document, P (1) = {traffic, feature}, P (2) = {stretch, feature}, P (3) = {traffic, feature} , P (4) = {traffic, classify}, P (5) = {stretch, classify}, and P (6) = {feature, classify}.
　次に単語共起抽出部４０は、生成した単語共起Ｐ（ｎ）の中から辞書３０に記憶されている単語共起Ｑ（ｍ）を抽出する（ステップＳ４）。つまり、単語共起抽出部４０は、入力済みの文書から生成した単語共起Ｐ（ｉ）が辞書３０に記憶されていれば、その記憶されている単語共起Ｐ（ｉ）を単語共起Ｑ（ｊ）にコピーする。 Next, the word co-occurrence extraction unit 40 extracts the word co-occurrence Q (m) stored in the dictionary 30 from the generated word co-occurrence P (n) (step S4). That is, if the word co-occurrence P (i) generated from the input document is stored in the dictionary 30, the word co-occurrence extraction unit 40 converts the stored word co-occurrence P (i) into the word co-occurrence P (i). Copy to Q (j).
　図５は、辞書３０に記憶されている単語共起Ｐ（ｉ）をコピーした単語共起Ｑ（ｊ）を示す図である。例えば、単語共起Ｐ（１）＝{トラフィック，伸び}は、単語共起Ｑ（１）＝{トラフィック，伸び}にコピーされる。 FIG. 5 is a diagram showing a word co-occurrence Q (j) obtained by copying the word co-occurrence P (i) stored in the dictionary 30. For example, word co-occurrence P (1) = {traffic, growth} is copied to word co-occurrence Q (1) = {traffic, growth}.
　辞書３０は、利用者が入力する解決した課題が、ＡＩで解決すべき課題であるか又はシステム開発で解決すべき課題であるかの程度を表す両者の和が1.0のＡＩスコアとシステムスコアが付与された単語の組みである単語共起が記憶されている。なお、本実施形態では、和が1.0になる例で説明するが、和は決められた定数であれば良い。例えば、和は10又は100などの定数にしても良い。この例では、「トラフィック」、「伸び」の単語共起Ｐ（１）のＡＩスコアは0.8、システムスコアは0.2である。また、「特徴」、「分類し」の単語共起Ｑ（２）のＡＩスコアは0.9、システムスコアは0.1である。辞書３０は予め作成しておく。 The dictionary 30 indicates whether the solved problem input by the user is a problem to be solved by AI or a problem to be solved by system development. The sum of both the AI score and the system score is 1.0. Word co-occurrence, which is a set of assigned words, is stored. In the present embodiment, an example will be described in which the sum is 1.0, but the sum may be a determined constant. For example, the sum may be a constant such as 10 or 100. In this example, the AI score of the word co-occurrence P (1) of “traffic” and “extension” is 0.8, and the system score is 0.2. Further, the AI score of the word co-occurrence Q (2) of “feature” and “classify” is 0.9, and the system score is 0.1. The dictionary 30 is created in advance.
　スコア判定部５０は、単語共起抽出部４０で抽出された単語共起Ｑ（ｊ）のＡＩスコアとシステムスコアをそれぞれ合算し（ステップＳ５）、合算したＡＩスコアとシステムスコアのどちらが大きいかを判定する（ステップＳ６、Ｓ７）。利用者が入力した文書が「過去のトラフィックデータ、…途中省略…分類し」の場合に、推薦実行ボタン１１が押下されると、ＡＩスコアの合算は1.7であり、シスシテムスコアの合算は0.3である。 The score determination unit 50 adds together the AI score and the system score of the word co-occurrence Q (j) extracted by the word co-occurrence extraction unit 40 (Step S5), and determines which of the combined AI score and the system score is larger. Judge (steps S6, S7). If the document input by the user is “past traffic data,... Omitting,... Categorized” and the recommendation execution button 11 is pressed, the sum of the AI scores is 1.7 and the sum of the system score is 0.3. It is.
　この例では、ＡＩスコアの方がシステムスコアよりも大きい（ＡＩスコア＞システムスコア）と判定される（ステップＳ７のＹＥＳ）。合算したＡＩスコアとシステムスコアが等しい場合（ステップＳ６のＮＯ）は、引き続き文書を入力するステップＳ１に戻る。この場合は、利用者に引き続き文書を入力するように促すメッセージを、表示パネルに表示するようにしても良い。 In this example, it is determined that the AI score is larger than the system score (AI score> system score) (YES in step S7). If the combined AI score and the system score are equal (NO in step S6), the process returns to step S1 for continuously inputting a document. In this case, a message urging the user to continue to input a document may be displayed on the display panel.
　なお、スコア判定部５０は、判定した判定結果と、判定した文書を外部に出力するようにしても良い。文書に判定結果を添付することで、課題解決を任された担当者間で認識のギャップを生じさせない。 The score determination unit 50 may output the determined result and the determined document to the outside. By attaching the judgment result to the document, a recognition gap does not occur between the persons in charge of solving the problem.
　推薦単語抽出部６０は、スコア判定部５０の判定が終了すると、単語共起Ｑ（ｊ）を単語に分解する。そして、各単語について、辞書３０において紐付けられている単語を検索し、スコアの高い順番に配列する。 (4) When the judgment by the score judgment unit 50 is completed, the recommended word extraction unit 60 decomposes the word co-occurrence Q (j) into words. Then, for each word, the linked word is searched in the dictionary 30, and the words are arranged in the descending order of the score.
　図６は、辞書３０で紐付けられている単語を模式的に示す図である。「トラフィック」は、例えば、「システム」、「伸び」、及び「予測」に紐付けられている。単語間の紐付けは、予め人が行う。 FIG. 6 is a diagram schematically showing words linked in the dictionary 30. “Traffic” is associated with, for example, “system”, “growth”, and “prediction”. The linking between words is performed by a person in advance.
　図７は、単語共起Ｑ（ｊ）を単語に分解し、紐付けられている単語共起をスコアの高い順番に配列した様子を模式的に示す図である。「トラフィック」をＭ１と仮定する。この例では、ＡＩスコアの方がシステムスコアよりも大きいと判定されているので、図７に示すＭ１に紐付けられた最もＡＩスコアの高い単語であるＭ１１は「伸び（Ｍ１との共起に対するＡＩスコアは0.8）」、2番目のＭ１２は「予測（Ｍ１との共起に対するＡＩスコアは0.7）」、3番目のＭ１３は「システム（Ｍ１との共起に対するＡＩスコアは0.2）」となる。 FIG. 7 is a diagram schematically illustrating a state in which the word co-occurrence Q (j) is decomposed into words, and the linked word co-occurrences are arranged in the order of higher scores. Assume "traffic" is M1. In this example, since the AI score is determined to be higher than the system score, the word M11 having the highest AI score linked to M1 shown in FIG. The AI score is 0.8), the second M12 is "prediction (AI score for co-occurrence with M1 is 0.7)", and the third M13 is "system (AI score for co-occurrence with M1 is 0.2)" .
　他の単語Ｍｉについても、Ｍ１と紐付いた単語（Ｍ１１、Ｍ１２、Ｍ１３…）の共起を、同様にＡＩスコアの大きい順番に配列する。そして、各共起に紐付けられた単語で、且つ形態素解析した文節に含まれない単語を推薦単語として抽出する。上記の例の場合は、文節に含まれていない「予測」が抽出される。また、例えばｉ＝２の単語Ｍ２が「伸び」であるとするならば、Ｍ２と共起し紐付けられている単語のうち、ここまでの文節に含まれていない単語Ｍ２１として「原因」が抽出される（ステップＳ８）。 共 Similarly, for other words Mi, co-occurrence of words (M11, M12, M13,...) Linked to M1 is similarly arranged in the descending order of the AI score. Then, words that are linked to each co-occurrence and are not included in the morphologically analyzed phrase are extracted as recommended words. In the case of the above example, “prediction” not included in the phrase is extracted. For example, if the word M2 of i = 2 is “extended”, among the words co-occurring and linked with M2, the “cause” is the word M21 that is not included in the phrase so far. It is extracted (step S8).
　システムスコアの方がＡＩスコアよりも大きい（システムスコア＞ＡＩスコア）と判定された場合（ステップＳ７のＮＯ））は、システムスコアの大きい順番に配列された中から、同様に推薦単語が抽出される（ステップＳ９）。 If it is determined that the system score is higher than the AI score (system score> AI score) (NO in step S7), the recommended words are similarly extracted from the arrangement in the descending order of the system score. (Step S9).
　抽出された推薦単語は、推薦単語表示部７０で表示される。利用者は、推薦単語表示部７０を参照しながら、続きの文書を作成し、文書入力部１０に入力する。 The extracted recommended words are displayed on the recommended word display unit 70. The user creates a subsequent document while referring to the recommended word display unit 70 and inputs the document to the document input unit 10.
　図３（ｂ）は、推薦単語表示部７０に推薦単語が表示された様子を示す図である。図３（ｃ）は、推薦単語を参照しながら利用者が入力した続きの文章の例を示す図である。図３（ｃ）に示すように、推薦単語を用いて続きの文書を作成することで、文書はスコアの高い種別の課題を表す表現（内容）に誘導されることになる。 FIG. 3B is a diagram showing a state in which recommended words are displayed on the recommended word display unit 70. FIG. 3C is a diagram illustrating an example of a subsequent sentence input by the user while referring to the recommended words. As shown in FIG. 3C, by creating a subsequent document using the recommended word, the document is guided to an expression (content) representing a type of task having a high score.
　以上述べたように本実施形態に係る課題解決支援装置１００は、解決したい課題を表す文書の作成を支援する課題解決支援装置であって、上記の文書に含めたい推薦単語が表示される推薦単語表示部７０を参照しながら利用者が作成する文書が入力される文書入力部１０と、上記の文書を形態素解析し、該文書の形態素を出力する形態素解析部２０と、課題がＡＩで解決すべき課題であるか又はシステム開発で解決すべき課題であるかの程度を表す両者の和が定数となるＡＩスコアとシステムスコアが付与された2つの異なる単語の組みである単語共起が記憶された辞書３０と、文書を形態素解析したそれぞれの文節から名詞と動詞を取り出し、該取り出した単語を含む単語共起が辞書３０に記憶されていれば該単語共起を抽出する単語共起抽出部４０と、抽出された単語共起のＡＩスコアとシステムスコアをそれぞれ合算し、合算したＡＩスコアとシステムスコアのどちらが大きいかを判定するスコア判定部５０と、辞書３０を参照し、抽出された単語共起の各単語に紐付けられた単語の中で、スコア判定部５０で大きいと判定された種別のスコアが最も高い単語で且つ文節に含まれない推薦単語を抽出する推薦単語抽出部６０とを備える。 As described above, the problem solving support device 100 according to the present embodiment is a problem solving support device that supports creation of a document representing a problem to be solved, and is a recommended word in which a recommended word to be included in the document is displayed. A document input unit 10 for inputting a document created by a user while referring to the display unit 70, a morphological analysis unit 20 for morphologically analyzing the above-mentioned document and outputting a morpheme of the document, and a problem solved by AI. A word co-occurrence, which is a combination of two different words to which an AI score and a system score are added, which represents the degree of whether the task is a task to be solved or a task to be solved in system development, is a constant. A word co-occurrence that extracts a noun and a verb from each dictionary obtained by morphologically analyzing a document and a word co-occurrence including the extracted word if the dictionary 30 stores the word co-occurrence. The extraction unit 40, the AI score and the system score of the extracted word co-occurrence are added together, and the score determination unit 50 that determines which of the combined AI score and the system score is larger, and the dictionary 30 are referred to. Word extraction unit that extracts a word that has the highest score of the type determined to be large by the score determination unit 50 and that is not included in a phrase among the words associated with each word of the co-occurred words. 60.
　これにより、利用者が作成する解決したい課題を表す文書の作成を支援することができる。つまり本実施形態に係る課題解決支援装置１００は、利用者が解決したい課題を表す文書を作成する際に、入力途中の文章から利用者が解決したい課題の種別を判別し、判別した種別の文書を作成するのに好適な推薦単語を表示する。したがって、利用者が推薦単語を参照しながら作成する文書は、解決したい課題の種別が明確な文書になる。よって、課題解決を任された担当者間で認識のギャップが生じ難い文書の作成を支援することができる。 This can support creation of a document that represents a problem that the user wants to solve. That is, the problem solving support device 100 according to the present embodiment, when creating a document representing a problem that the user wants to solve, determines the type of the problem that the user wants to solve from the text being input, and Display recommended words suitable for creating. Therefore, the document created by the user while referring to the recommended words is a document in which the type of the problem to be solved is clear. Therefore, it is possible to support creation of a document in which a recognition gap is unlikely to occur between persons in charge of task resolution.
 〔第２実施形態〕
　図８は、本発明の第２実施形態に係る課題解決支援装置の機能構成例を示すブロック図である。図８に示す課題解決支援装置２００は、課題解決支援装置１００（図１）に対して、単語共起抽出部２４０と未登録単語共起記録部２８０を備える点で異なる。 [Second embodiment]
FIG. 8 is a block diagram illustrating a functional configuration example of the problem solving support device according to the second embodiment of the present invention. The problem solving support device 200 shown in FIG. 8 differs from the problem solving supporting device 100 (FIG. 1) in that a word co-occurrence extraction unit 240 and an unregistered word co-occurrence recording unit 280 are provided.
　図９は、課題解決支援装置２００の処理手順を示すフローチャートである。図９に示すフローチャートは、図２に示したフローチャートのステップＳ４以降の処理手順を示す。 FIG. 9 is a flowchart showing a processing procedure of the problem solving support device 200. The flowchart shown in FIG. 9 shows the processing procedure after step S4 of the flowchart shown in FIG.
　単語共起抽出部２４０は、利用者が入力する文書を形態素解析した文節から辞書３０に記憶されていない名詞と動詞も抽出する（ステップＳ４０のＹＥＳ）。本実施形態では、ＡＩスコアとシステムスコアの和である定数は1.0と定める。 The word co-occurrence extraction unit 240 also extracts nouns and verbs that are not stored in the dictionary 30 from phrases obtained by morphological analysis of a document input by the user (YES in step S40). In the present embodiment, the constant that is the sum of the AI score and the system score is set to 1.0.
　未登録単語共起記録部２８０は、辞書３０に記憶されていない単語と、文書を形態素解析した文節から取り出した他の単語の組みである未登録単語共起を生成し、ＡＩスコア＝0.5、システムスコア＝0.5として生成した未登録単語共起を辞書３０に記録する（ステップＳ４１）。 The unregistered word co-occurrence recording unit 280 generates an unregistered word co-occurrence which is a combination of a word not stored in the dictionary 30 and another word extracted from a phrase obtained by morphologically analyzing the document, and has an AI score = 0.5, The unregistered word co-occurrence generated with the system score = 0.5 is recorded in the dictionary 30 (step S41).
　スコア判定部５０は、新たに記録された単語共起（未登録単語共起）を含めて、抽出された単語共起のＡＩスコアとシステムスコアをそれぞれ合算する（ステップＳ５）。ステップＳ５～Ｓ９は、図２で説明済みの処理ステップである。 The score determination unit 50 adds together the AI score and the system score of the extracted word co-occurrence, including the newly recorded word co-occurrence (unregistered word co-occurrence) (step S5). Steps S5 to S9 are the processing steps described in FIG.
　未登録単語共起記録部２８０は、推薦単語を抽出した後、ＡＩスコアが高いと判定された場合、新たに記録した単語共起のＡＩスコアを+0.1とし、システムスコアを-0.1として再記録する（ステップＳ４３）。また、システムスコアが高いと判定された場合は、新たに記録した単語共起のＡＩスコアを-0.1とし、システムスコアを+0.1として再記録する（ステップＳ４５）。 The unregistered word co-occurrence recording unit 280 re-records the newly recorded word co-occurrence with the AI score of +0.1 and the system score of -0.1 if the AI score is determined to be high after extracting the recommended word. (Step S43). If it is determined that the system score is high, the AI score of the newly recorded word co-occurrence is set to -0.1, and the system score is set to +0.1 and re-recorded (step S45).
　図１０は、新たに記録された単語共起（未登録単語共起）の例を示す図であり、図１０（ａ）はステップＳ４１で記録された単語共起の例、図１０（ｂ）はステップＳ４３で再記録された単語共起の例を示す。 FIG. 10 is a diagram showing an example of a newly recorded word co-occurrence (unregistered word co-occurrence). FIG. 10A shows an example of the word co-occurrence recorded in step S41, and FIG. Shows an example of the word co-occurrence re-recorded in step S43.
　このように動作することで、新たに記録された単語共起（未登録単語共起）のＡＩスコアとシステムスコアは逐次更新され、適切なスコア値に自動的に調整される。なお、予め辞書３０に記憶されている単語共起のスコアは不変である。 動作 By operating in this manner, the AI score and system score of the newly recorded word co-occurrence (unregistered word co-occurrence) are sequentially updated and automatically adjusted to an appropriate score value. The score of the word co-occurrence stored in the dictionary 30 in advance is unchanged.
　以上述べたように本実施形態に係る課題解決支援装置２００は、単語共起抽出部２４０と未登録単語共起記録部２８０を備え、単語共起抽出部２４０は、文書を形態素解析した文節に含まれ且つ辞書３０に記憶されていない名詞と動詞を抽出し、未登録単語共起記録部２８０は、単語共起抽出部２４０で抽出された辞書に記憶されていない単語と、上記の文節から取り出した他の単語の組みである未登録単語共起を生成し、スコア判定部５０で大きいと判定された一方の種別のスコアを大きく、また他方の種別のスコアを小さくして辞書３０に記録する。 As described above, the problem solving support device 200 according to the present embodiment includes the word co-occurrence extraction unit 240 and the unregistered word co-occurrence recording unit 280, and the word co-occurrence extraction unit 240 Nouns and verbs that are included and not stored in the dictionary 30 are extracted, and the unregistered word co-occurrence recording unit 280 extracts the words that are not stored in the dictionary extracted by the word co-occurrence extraction unit 240 from the above phrases. An unregistered word co-occurrence, which is a set of other words taken out, is generated, and the score of one type determined to be large by the score determination unit 50 is increased, and the score of the other type is reduced and recorded in the dictionary 30. I do.
　これにより、利用者が辞書３０に記憶されていない単語を入力した際に、新たな単語共起が自動的に増加する。したがって、解決したい課題の種別がより明確な文書の作成を支援することができる。 Thereby, when the user inputs a word that is not stored in the dictionary 30, a new word co-occurrence is automatically increased. Therefore, it is possible to support creation of a document in which the type of the problem to be solved is clearer.
　以上述べたように、本実施形態に係る課題解決支援装置１００，２００によれば、例えば業務上の課題を解決する目的で、その課題を文書で表す場合に、その課題がシステム開発的な課題であるのか又はＡＩで解決すべき課題であるのかを分別して記述することの支援ができる。その結果、課題解決を任された担当者間で認識のギャップを生じさせない。 As described above, according to the problem solving support devices 100 and 200 according to the present embodiment, when the problem is represented by a document for the purpose of solving a business problem, for example, the problem is a system development problem. Or the problem to be solved by AI can be described separately. As a result, there is no perception gap between persons in charge of task resolution.
　なお、上記の実施形態は、本装置を一つのコンピュータで構成する例、または、クライアントサーバシステムで構成する例で説明を行ったが、本発明はこの例に限定されない。文書入力部１０と推薦単語表示部７０を利用者が同時に参照できるとすれば、他はどのように構成しても良い。また、辞書３０には、機械学習、回帰。予測、チャットポット、最適割合、テキストマイニング、データマイニング、識別・分類、シミュレーション、推論、強化学習、クラスタリング、推薦、近傍探索、統計的因果推論、知識表現・探索・推論、確率推論、状態推定、系列ラベリング、及び特徴抽出等に関連する単語共起（単語）を記憶させておいても良い。 Although the above embodiment has been described with respect to an example in which the present apparatus is configured by one computer or an example in which the apparatus is configured by a client server system, the present invention is not limited to this example. If the user can simultaneously refer to the document input unit 10 and the recommended word display unit 70, any other configuration may be used. The dictionary 30 includes machine learning and regression. Prediction, chat pot, optimal ratio, text mining, data mining, identification / classification, simulation, inference, reinforcement learning, clustering, recommendation, neighborhood search, statistical causal inference, knowledge expression / search / inference, probability inference, state estimation, Word co-occurrence (words) related to sequence labeling, feature extraction, etc. may be stored.
　このように本発明は、上記の実施形態に限定されるものではなく、その要旨の範囲内で変形が可能である。 As described above, the present invention is not limited to the above embodiment, and can be modified within the scope of the gist.
１０：文書入力部
１１：推薦実行ボタン
２０：形態素解析部
３０：辞書
４０、２４０：単語共起抽出部
５０：スコア判定部
６０：推薦単語抽出部
７０：推薦単語表示部
２８０：未登録単語共起記録部
１００、２００：課題解決支援装置 10: document input unit 11: recommendation execution button 20: morphological analysis unit 30: dictionaries 40 and 240: word co-occurrence extraction unit 50: score determination unit 60: recommended word extraction unit 70: recommended word display unit 280: unregistered word Recorder 100, 200: Problem solving support device

Claims (3)

  1. 　解決したい課題を表す文書の作成を支援する課題解決支援装置であって、
    　前記文書に含めたい推薦単語が表示される推薦単語表示部を参照しながら利用者が作成する前記文書が入力される文書入力部と、
    　前記文書を形態素解析し、該文書の形態素を出力する形態素解析部と、
    　前記課題がＡＩで解決すべき課題であるか又はシステム開発で解決すべき課題であるかの程度を表す両者の和が定数となるＡＩスコアとシステムスコアが付与された2つの異なる単語の組みである単語共起が記憶された辞書と、
    　前記文書を形態素解析したそれぞれの文節から名詞と動詞を取り出し、該取り出した単語を含む前記単語共起が前記辞書に記憶されていれば該単語共起を抽出する単語共起抽出部と、
    　抽出された前記単語共起のＡＩスコアとシステムスコアをそれぞれ合算し、合算したＡＩスコアとシステムスコアのどちらが大きいかを判定するスコア判定部と、
    　前記辞書を参照し、抽出された前記単語共起の各単語に紐付けられた単語の中で、前記スコア判定部で大きいと判定された種別のスコアが最も高い単語で且つ前記文節に含まれない前記推薦単語を抽出する推薦単語抽出部と
    　を備えることを特徴とする課題解決支援装置。     A problem solving support device for supporting creation of a document representing a problem to be solved,
    A document input unit in which the document created by the user is input with reference to a recommended word display unit in which a recommended word to be included in the document is displayed;
    A morphological analyzer for morphologically analyzing the document and outputting a morpheme of the document;
    A combination of two different words with an AI score and a system score where the sum of the two is a constant indicating the degree to which the problem is a problem to be solved by AI or a problem to be solved in system development. A dictionary in which a certain word co-occurrence is stored,
    A word co-occurrence extraction unit that extracts a noun and a verb from each of the phrases obtained by morphologically analyzing the document, and extracts the word co-occurrence if the word co-occurrence including the extracted word is stored in the dictionary;
    A score determining unit that adds together the extracted AI score and the system score of the word co-occurrence, and determines which of the combined AI score and the system score is larger,
    Referring to the dictionary, among the words linked to each of the extracted words of the word co-occurrence, the score of the type determined to be large by the score determination unit is the highest word and is included in the phrase. And a recommended word extracting unit that extracts the no recommended word.
  2. 　前記単語共起抽出部は、前記文節に含まれ且つ前記辞書に記憶されていない名詞と動詞を抽出し,
    　前記単語共起抽出部で抽出された前記辞書に記憶されていない単語と、前記文節から取り出した他の単語の組みである未登録単語共起を生成し、前記スコア判定部で大きいと判定された一方の種別のスコアを大きく、また他方の種別のスコアを小さくして前記辞書に記録する未登録単語共起記録部を
    　備えることを特徴とする請求項１に記載の課題解決支援装置。     The word co-occurrence extraction unit extracts nouns and verbs included in the phrase and not stored in the dictionary,
    An unregistered word co-occurrence, which is a combination of a word not stored in the dictionary extracted by the word co-occurrence extraction unit and another word extracted from the phrase, is determined by the score determination unit to be large. 2. The problem solving support device according to claim 1, further comprising: an unregistered word co-occurrence recording unit that records in the dictionary with a score of one type being large and a score of the other type being small.
  3. 　解決したい課題を表す文書の作成を支援する課題解決支援装置が実行する課題解決支援方法であって、
    　前記課題がＡＩで解決すべき課題であるか又はシステム開発で解決すべき課題であるかの程度を表す両者の和が定数となるＡＩスコアとシステムスコアが付与された2つの異なる単語の組みである単語共起が記憶された辞書を備え、
    　前記文書に含めたい推薦単語が表示される推薦単語表示部を参照しながら利用者が作成する前記文書が入力される文書入力ステップと、
    　前記文書を形態素解析し、該文書の形態素を出力する形態素解析ステップと、
    　前記文書を形態素解析したそれぞれの文節から名詞と動詞を取り出し、該取り出した単語を含む前記単語共起が前記辞書に記憶されていれば該単語共起を抽出する単語共起抽出ステップと、
    　抽出された前記単語共起のＡＩスコアとシステムスコアをそれぞれ合算し、合算したＡＩスコアとシステムスコアのどちらが大きいかを判定するスコア判定ステップと、
    　前記辞書を参照し、抽出された前記単語共起の各単語に紐付けられた単語の中で、前記スコア判定ステップで大きいと判定された種別のスコアが最も高い単語で且つ前記文節に含まれない前記推薦単語を抽出する推薦単語抽出ステップと
    　を行うことを特徴とする課題解決支援方法。     A problem solving support method executed by a problem solving support device that supports creation of a document representing a problem to be solved,
    A combination of two different words with an AI score and a system score where the sum of the two is a constant indicating the degree to which the problem is a problem to be solved by AI or a problem to be solved in system development. A dictionary in which certain word co-occurrences are stored,
    A document input step in which the document created by the user is input while referring to a recommended word display unit in which a recommended word to be included in the document is displayed;
    A morphological analysis step of morphologically analyzing the document and outputting a morpheme of the document;
    A word co-occurrence extraction step of extracting a noun and a verb from each phrase obtained by morphologically analyzing the document, and extracting the word co-occurrence if the word co-occurrence including the extracted word is stored in the dictionary;
    A score determination step of summing the extracted AI co-occurrence AI score and the system score, and determining which of the summed AI score and the system score is greater,
    With reference to the dictionary, among the words linked to each of the extracted words of the word co-occurrence, the score of the type determined to be large in the score determination step is the highest word and is included in the phrase. Performing a recommended word extracting step of extracting the recommended word that does not exist.
PCT/JP2019/034221 2018-09-13 2019-08-30 Problem solution assistance device and method therefor WO2020054465A1 (en)

Applications Claiming Priority (2)

Application Number Priority Date Filing Date Title
JP2018-171256 2018-09-13
JP2018171256A JP2020042695A (en) 2018-09-13 2018-09-13 Problem solution support device and method thereof

Publications (1)

Publication Number Publication Date
WO2020054465A1 true WO2020054465A1 (en) 2020-03-19

Family

ID=69777583

Family Applications (1)

Application Number Title Priority Date Filing Date
PCT/JP2019/034221 WO2020054465A1 (en) 2018-09-13 2019-08-30 Problem solution assistance device and method therefor

Country Status (2)

Country Link
JP (1) JP2020042695A (en)
WO (1) WO2020054465A1 (en)

Citations (1)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
JP2006072436A (en) * 2004-08-31 2006-03-16 Ricoh Co Ltd Document generation support system, its method, program, and recording medium

Patent Citations (1)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
JP2006072436A (en) * 2004-08-31 2006-03-16 Ricoh Co Ltd Document generation support system, its method, program, and recording medium

Non-Patent Citations (1)

* Cited by examiner, † Cited by third party
Title
SANOMACHI, YUKI ET AL.: "A Proposal and Evaluation of the Writing Rule Dataset for Requirements Specification Written in Japanese", IPSJ SIG TECHNICAL REPORT: SOFTWARE ENGINEERING (SE) 2018- SE -19 8, 2 March 2018 (2018-03-02), pages 1 - 8 *

Also Published As

Publication number Publication date
JP2020042695A (en) 2020-03-19

Similar Documents

Publication Publication Date Title
Kumar et al. Text mining: concepts, process and applications
US20040073874A1 (en) Device for retrieving data from a knowledge-based text
KR102032091B1 (en) Method And System of Comment Emotion Analysis based on Artificial Intelligence
JP2005165958A (en) Information retrieval system, information retrieval support system and method therefor, and program
JP7297458B2 (en) Interactive content creation support method
WO2020080155A1 (en) Technology name/service name generation device and method therefor
JP2021022211A (en) Inquiry response support device, inquiry response support method, program and recording medium
JPH06215049A (en) Document summarizing device
US7383171B2 (en) Semantic stenography using short note input data
JP2020166735A (en) Generation method, learning method, generation program, and generation device
Begum et al. Analysis of legal case document automated summarizer
US11599580B2 (en) Method and system to extract domain concepts to create domain dictionaries and ontologies
WO2020054465A1 (en) Problem solution assistance device and method therefor
JP2007011774A (en) Sentence analysis device, sentence analysis method, program, and storage medium
JP4169618B2 (en) Text information management device
KR20210012606A (en) Method of bio information analysis and storage medium storing a program for performing the same
JP7234010B2 (en) Information processing device and information processing method
Wolff et al. " I’ve Got a Feeling": Performing Sentiment Analysis on Critical Moments in Beatles History
US11971915B2 (en) Language processor, language processing method and language processing program
JP2003141132A (en) Information processor and method
JPH1145249A (en) Information retrieval device and computer-readable recording medium where program for making computer function as same device is recorded
JP2002366556A (en) Information retrieving method
JP4088176B2 (en) Question answering apparatus and question answering program
Yulianton et al. Sentiment Analysis Of Internet Service Providers
JP2016218853A (en) Information processor, information processing method, and program

Legal Events

Date Code Title Description
NENP Non-entry into the national phase

Ref country code: DE

122 Ep: pct application non-entry in european phase

Ref document number: 19858828

Country of ref document: EP

Kind code of ref document: A1