JP2020060846A5 - - Google Patents
Download PDFInfo
- Publication number
- JP2020060846A5 JP2020060846A5 JP2018189979A JP2018189979A JP2020060846A5 JP 2020060846 A5 JP2020060846 A5 JP 2020060846A5 JP 2018189979 A JP2018189979 A JP 2018189979A JP 2018189979 A JP2018189979 A JP 2018189979A JP 2020060846 A5 JP2020060846 A5 JP 2020060846A5
- Authority
- JP
- Japan
- Prior art keywords
- question
- candidate
- tree
- frequent
- explanations
- Prior art date
- Legal status (The legal status 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 status listed.)
- Granted
Links
- 239000011159 matrix material Substances 0.000 claims description 24
- 238000003066 decision tree Methods 0.000 claims description 15
- 230000000877 morphologic Effects 0.000 claims description 8
- 238000004364 calculation method Methods 0.000 claims description 6
- 238000000034 method Methods 0.000 claims 11
Description
続いて、質問生成装置１００は、ユーザ端末２００からの問い合わせを受信し、当該問い合わせが含むキーワードに基づき、ｓ４０で特定している頻出回答のうち当該キーワードを含むものを優先候補説明文として特定する（ｓ４２）。 Subsequently, the question generator 100 receives the inquiry from the user terminal 200, and based on the keyword included in the inquiry, specifies the frequently-used answer specified in s40 that includes the keyword as the priority candidate explanatory text. (S42).
また、質問生成装置１００は、ｓ４１で生成している頻出回答向けマトリクスにおける、上述の優先候補説明文（ｓ４２で特定したもの）に関するものを当該優先候補説明文のＩＤに基づき特定し、ここで特定した要素に関して決定木分析を実行する（ｓ４３）。 Also, the question generator 100, the frequent answer for matrix are generated in s41, identified on the basis of the related priority candidate description above (those identified in s42) the ID of the priority candidate explanation, here A decision tree analysis is performed on the identified elements (s43).
また、本実施形態の質問生成装置において、前記演算部は、前記各説明文のうち、所定期間において回答として出力された実績が所定基準以上の頻出回答であるものに関する形態素解析結果に基づき、前記頻出回答における各単語の有無を記述した頻出回答向けマトリクスの生成を行い、所定端末から問い合わせを受信し、当該問い合わせが含むキーワードに基づき、前記頻出回答のうち前記キーワードを含むものを優先候補説明文として特定し、前記頻出回答向けマトリクスにおける前記優先候補説明文に関する記述に基づき決定木分析を実行して、分岐判定項目を質問文とした頻出回答向け質問木を生成するものである、としてもよい。 Further, in the question generation device of the present embodiment, the calculation unit describes the above-mentioned explanations based on the morphological analysis result of each of the above-mentioned explanatory texts in which the actual result output as an answer in a predetermined period is a frequent answer of a predetermined standard or more. A matrix for frequent answers that describes the presence or absence of each word in the frequent answers is generated, an inquiry is received from a predetermined terminal, and based on the keywords included in the inquiry, those containing the keywords are given priority candidate explanations. identified as, by executing a decision tree analysis on the basis of description of the priority candidate legends in the frequent answer for matrix, and generates frequent answers for questions trees where the branch judgment items and the query sentence may be ..
また、本実施形態の質問生成方法において、前記情報処理装置が、前記各説明文のうち、所定期間において回答として出力された実績が所定基準以上の頻出回答であるものに関する形態素解析結果に基づき、前記頻出回答における各単語の有無を記述した頻出回答向けマトリクスの生成を行い、所定端末から問い合わせを受信し、当該問い合わせが含むキーワードに基づき、前記頻出回答のうち前記キーワードを含むものを優先候補説明文として特定し、前記頻出回答向けマトリクスにおける前記優先候補説明文に関する記述に基づき決定木分析を実行して、分岐判定項目を質問文とした頻出回答向け質問木を生成する、としてもよい。 Further, in the question generation method of the present embodiment, based on the morphological analysis result regarding each of the above-mentioned explanatory texts in which the actual result output as an answer in a predetermined period is a frequent answer equal to or higher than a predetermined standard. A matrix for frequent answers that describes the presence or absence of each word in the frequent answers is generated, an inquiry is received from a predetermined terminal, and based on the keywords included in the inquiry, those containing the keywords are given priority candidate explanations. identified as a statement, the frequent answer for running based on said description of the priority candidate legend decision tree analysis in the matrix, it generates frequent answers for questions trees where the branch judgment items and the query sentence may be.
Claims (12)
前記各説明文の形態素解析結果に基づき、前記各説明文における各単語の有無を記述したマトリクスの生成を行う処理と、所定端末から問い合わせを受信し、当該問い合わせが含むキーワードに基づき、前記各説明文のうち前記キーワードを含むものを候補説明文として特定する処理と、前記マトリクスにおける前記候補説明文に関する記述に基づき決定木分析を実行し、分岐判定項目を質問文とした質問木を生成する処理と、を実行する演算部と、
を備えることを特徴とする質問生成装置。 A storage unit that holds each explanation for each of the specified items,
Based on the morphological analysis result of each of the explanations, a process of generating a matrix describing the presence or absence of each word in each of the explanations, and receiving an inquiry from a predetermined terminal, and based on the keywords included in the inquiry, each of the explanations A process of specifying a sentence including the keyword as a candidate explanation sentence, and a process of executing a decision tree analysis based on the description of the candidate explanation sentence in the matrix and generating a question tree using a branch judgment item as a question sentence. And the arithmetic unit that executes
A question generator, characterized in that it comprises.
前記生成した質問木の段数が所定基準を越える場合、
前記端末に追加キーワードの要求メッセージを通知し、当該端末から追加キーワードを取得する処理を更に実行し、
前記候補説明文の特定に際し、前記キーワードおよび前記追加キーワードに基づき、前記各説明文のうち前記キーワードおよび前記追加キーワードを含むものを新たな候補説明文として特定し、
前記質問木の生成に際し、前記マトリクスにおける前記新たな候補説明文に関する記述に基づき決定木分析を実行し、分岐判定項目を質問文とした新たな質問木を生成し、
前記新たな質問木の段数が前記所定基準を超えなくなるまで、前記追加キーワードの取得、前記新たな候補説明文の特定、および、前記新たな質問木の生成、を繰り返し行うものである、
ことを特徴とする請求項１に記載の質問生成装置。 The calculation unit
When the number of steps of the generated question tree exceeds the predetermined standard
Notify the terminal of the request message of the additional keyword, further execute the process of acquiring the additional keyword from the terminal, and then execute the process.
In identifying the candidate explanatory text, based on the keyword and the additional keyword, each of the explanatory texts including the keyword and the additional keyword is specified as a new candidate explanatory text.
When generating the question tree, a decision tree analysis is executed based on the description regarding the new candidate explanatory text in the matrix, and a new question tree with the branch judgment item as the question text is generated.
The acquisition of the additional keyword, the identification of the new candidate explanation, and the generation of the new question tree are repeated until the number of stages of the new question tree does not exceed the predetermined criterion.
The question generator according to claim 1.
前記候補説明文のうち、各候補説明文を横断して頻出する単語を含んでいるものを優先候補説明文として特定し、
前記質問木の生成に際し、前記マトリクスにおける前記優先候補説明文に関する記述に基づき決定木分析を実行して、分岐判定項目を質問文とした頻出単語向け質問木を生成するものである、
ことを特徴とする請求項１に記載の質問生成装置。 The calculation unit
Among the candidate explanations, those containing words that frequently appear across each candidate explanation are specified as priority candidate explanations.
When generating the question tree, a decision tree analysis is executed based on the description of the priority candidate explanation sentence in the matrix to generate a question tree for frequently-used words with the branch determination item as the question sentence.
The question generator according to claim 1.
前記頻出単語向け質問木における所定段の分岐判定項目で前記候補説明文を区分しきれていない場合、前記候補説明文のうち、前記優先候補説明文以外の非優先候補説明文を特定し、
前記マトリクスにおける前記非優先候補説明文に関する記述に基づき、前記所定段の分岐判定項目を起点にした決定木分析を実行して、非頻出単語向け質問木を生成するものである、
ことを特徴とする請求項３に記載の質問生成装置。 The calculation unit
When the candidate explanations are not completely classified by the branch determination items in the predetermined stage in the question tree for frequently-used words, the non-priority candidate explanations other than the priority candidate explanations are specified among the candidate explanations.
Based on the description of the non-priority candidate explanatory text in the matrix, a decision tree analysis starting from the branch determination item of the predetermined stage is executed to generate a question tree for non-frequent words.
The question generator according to claim 3.
前記各説明文のうち、所定期間において回答として出力された実績が所定基準以上の頻出回答であるものに関する形態素解析結果に基づき、前記頻出回答における各単語の有無を記述した頻出回答向けマトリクスの生成を行い、
所定端末から問い合わせを受信し、当該問い合わせが含むキーワードに基づき、前記頻出回答のうち前記キーワードを含むものを優先候補説明文として特定し、
前記頻出回答向けマトリクスにおける前記優先候補説明文に関する記述に基づき決定木分析を実行して、分岐判定項目を質問文とした頻出回答向け質問木を生成するものである、
ことを特徴とする請求項１に記載の質問生成装置。 The calculation unit
Generation of a matrix for frequent answers that describes the presence or absence of each word in the frequent answers based on the morphological analysis result of each of the above explanations for which the actual result output as an answer in a predetermined period is a frequent answer exceeding a predetermined standard. And
Upon receiving an inquiry from a predetermined terminal, based on the keyword included in the inquiry, the frequently-used answers including the keyword are specified as priority candidate explanations.
A decision tree analysis is executed based on the description of the priority candidate explanatory text in the matrix for frequent answers to generate a question tree for frequent answers with the branch determination item as the question text.
The question generator according to claim 1.
前記頻出回答向け質問木における所定段の分岐判定項目で前記優先候補説明文を区分しきれていない場合、
前記各説明文のうち、前記頻出回答以外の非頻出回答であるものに関する形態素解析結果に基づき、前記非頻出回答における各単語の有無を記述した非頻出回答向けマトリクスの生成を行い、
前記非頻出回答向けマトリクスにおける前記非頻出回答に関する記述に基づき、前記所定段の分岐判定項目を起点にした決定木分析を実行して、非頻出回答向け質問木を生成するものである、
ことを特徴とする請求項５に記載の質問生成装置。 The calculation unit
When the priority candidate explanation is not completely classified by the branch judgment item of the predetermined stage in the question tree for frequent answers.
Based on the morphological analysis results for the non-frequent answers other than the frequent answers, a matrix for the non-frequent answers describing the presence or absence of each word in the non-frequent answers was generated.
Based on the description of the non-frequent answer in the matrix for non-frequent answers, a decision tree analysis starting from the branch determination item of the predetermined stage is executed to generate a question tree for non-frequent answers.
The question generator according to claim 5.
前記各説明文の形態素解析結果に基づき、前記各説明文における各単語の有無を記述したマトリクスの生成を行う処理と、
所定端末から問い合わせを受信し、当該問い合わせが含むキーワードに基づき、前記各説明文のうち前記キーワードを含むものを候補説明文として特定する処理と、
前記マトリクスにおける前記候補説明文に関する記述に基づき決定木分析を実行し、分岐判定項目を質問文とした質問木を生成する処理と、
を実行することを特徴とする質問生成方法。 An information processing device equipped with a storage unit that holds explanations for each of the predetermined items
Based on the morphological analysis result of each explanatory sentence, a process of generating a matrix describing the presence or absence of each word in each explanatory sentence, and a process of generating a matrix.
A process of receiving an inquiry from a predetermined terminal and, based on the keyword included in the inquiry, specifying one of the above explanations including the keyword as a candidate explanation.
A process of executing a decision tree analysis based on the description of the candidate explanatory text in the matrix and generating a question tree using the branch judgment item as the question text.
A question generation method characterized by executing.
前記生成した質問木の段数が所定基準を越える場合、
前記端末に追加キーワードの要求メッセージを通知し、当該端末から追加キーワードを取得する処理を更に実行し、
前記候補説明文の特定に際し、前記キーワードおよび前記追加キーワードに基づき、前記各説明文のうち前記キーワードおよび前記追加キーワードを含むものを新たな候補説明文として特定し、
前記質問木の生成に際し、前記マトリクスにおける前記新たな候補説明文に関する記述に基づき決定木分析を実行し、分岐判定項目を質問文とした新たな質問木を生成し、
前記新たな質問木の段数が前記所定基準を超えなくなるまで、前記追加キーワードの取得、前記新たな候補説明文の特定、および、前記新たな質問木の生成、を繰り返し行う、
ことを特徴とする請求項７に記載の質問生成方法。 The information processing device
When the number of steps of the generated question tree exceeds the predetermined standard
Notify the terminal of the request message of the additional keyword, further execute the process of acquiring the additional keyword from the terminal, and then execute the process.
In identifying the candidate explanatory text, based on the keyword and the additional keyword, each of the explanatory texts including the keyword and the additional keyword is specified as a new candidate explanatory text.
When generating the question tree, a decision tree analysis is executed based on the description regarding the new candidate explanatory text in the matrix, and a new question tree with the branch judgment item as the question text is generated.
The acquisition of the additional keyword, the identification of the new candidate explanation, and the generation of the new question tree are repeated until the number of stages of the new question tree does not exceed the predetermined criterion.
The question generation method according to claim 7, wherein the question is generated.
前記候補説明文のうち、各候補説明文を横断して頻出する単語を含んでいるものを優先候補説明文として特定し、
前記質問木の生成に際し、前記マトリクスにおける前記優先候補説明文に関する記述に基づき決定木分析を実行して、分岐判定項目を質問文とした頻出単語向け質問木を生成する、
ことを特徴とする請求項７に記載の質問生成方法。 The information processing device
Among the candidate explanations, those containing words that frequently appear across each candidate explanation are specified as priority candidate explanations.
When generating the question tree, a decision tree analysis is executed based on the description of the priority candidate explanation sentence in the matrix to generate a question tree for frequently-used words with the branch determination item as the question sentence.
The question generation method according to claim 7, wherein the question is generated.
前記頻出単語向け質問木における所定段の分岐判定項目で前記候補説明文を区分しきれていない場合、前記候補説明文のうち、前記優先候補説明文以外の非優先候補説明文を特定し、
前記マトリクスにおける前記非優先候補説明文に関する記述に基づき、前記所定段の分岐判定項目を起点にした決定木分析を実行して、非頻出単語向け質問木を生成する、
ことを特徴とする請求項９に記載の質問生成方法。 The information processing device
When the candidate explanations are not completely classified by the branch determination items in the predetermined stage in the question tree for frequently-used words, the non-priority candidate explanations other than the priority candidate explanations are specified among the candidate explanations.
Based on the description of the non-priority candidate explanatory text in the matrix, a decision tree analysis starting from the branch determination item of the predetermined stage is executed to generate a question tree for non-frequent words.
The question generation method according to claim 9, wherein the question is generated.
前記各説明文のうち、所定期間において回答として出力された実績が所定基準以上の頻出回答であるものに関する形態素解析結果に基づき、前記頻出回答における各単語の有無を記述した頻出回答向けマトリクスの生成を行い、
所定端末から問い合わせを受信し、当該問い合わせが含むキーワードに基づき、前記頻出回答のうち前記キーワードを含むものを優先候補説明文として特定し、
前記頻出回答向けマトリクスにおける前記優先候補説明文に関する記述に基づき決定木分析を実行して、分岐判定項目を質問文とした頻出回答向け質問木を生成する、
ことを特徴とする請求項７に記載の質問生成方法。 The information processing device
Generation of a matrix for frequent answers that describes the presence or absence of each word in the frequent answers based on the morphological analysis result of each of the above explanations for which the actual result output as an answer in a predetermined period is a frequent answer exceeding a predetermined standard. And
Upon receiving an inquiry from a predetermined terminal, based on the keyword included in the inquiry, the frequently-used answers including the keyword are specified as priority candidate explanations.
A decision tree analysis is executed based on the description of the priority candidate explanatory text in the matrix for frequent answers to generate a question tree for frequent answers with the branch judgment item as the question text.
The question generation method according to claim 7, wherein the question is generated.
前記頻出回答向け質問木における所定段の分岐判定項目で前記優先候補説明文を区分しきれていない場合、
前記各説明文のうち、前記頻出回答以外の非頻出回答であるものに関する形態素解析結果に基づき、前記非頻出回答における各単語の有無を記述した非頻出回答向けマトリクスの生成を行い、
前記非頻出回答向けマトリクスにおける前記非頻出回答に関する記述に基づき、前記所定段の分岐判定項目を起点にした決定木分析を実行して、非頻出回答向け質問木を生成する、
ことを特徴とする請求項１１に記載の質問生成方法。 The information processing device
When the priority candidate explanation is not completely classified by the branch judgment item of the predetermined stage in the question tree for frequent answers.
Based on the morphological analysis results for the non-frequent answers other than the frequent answers, a matrix for the non-frequent answers describing the presence or absence of each word in the non-frequent answers was generated.
Based on the description of the non-frequent answer in the matrix for non-frequent answers, a decision tree analysis starting from the branch determination item of the predetermined stage is executed to generate a question tree for non-frequent answers.
The question generation method according to claim 11, wherein the question is generated.
Priority Applications (2)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| JP2018189979A JP7028746B2 (en) | 2018-10-05 | 2018-10-05 | Question generator and question generation method |
| PCT/JP2019/038845 WO2020071398A1 (en) | 2018-10-05 | 2019-10-02 | Question generation device and question generation method |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| JP2018189979A JP7028746B2 (en) | 2018-10-05 | 2018-10-05 | Question generator and question generation method |
Publications (3)
| Publication Number | Publication Date |
|---|---|
| JP2020060846A JP2020060846A (en) | 2020-04-16 |
| JP2020060846A5 true JP2020060846A5 (en) | 2021-03-04 |
| JP7028746B2 JP7028746B2 (en) | 2022-03-02 |
Family
ID=70055554
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| JP2018189979A Active JP7028746B2 (en) | 2018-10-05 | 2018-10-05 | Question generator and question generation method |
Country Status (2)
| Country | Link |
|---|---|
| JP (1) | JP7028746B2 (en) |
| WO (1) | WO2020071398A1 (en) |
Families Citing this family (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| UA126005C2 (en) | 2017-12-06 | 2022-07-27 | Цзянсу Хенжуй Медісін Ко., Лтд. | Salt of phenylpropionamide derivative and preparation method therefor |
| CN112948608B (en) * | 2021-02-01 | 2023-08-22 | 北京百度网讯科技有限公司 | Picture searching method and device, electronic equipment and computer readable storage medium |
Family Cites Families (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| JP2011108085A (en) | 2009-11-19 | 2011-06-02 | Nippon Hoso Kyokai <Nhk> | Knowledge construction device and program |
| JP6419859B2 (en) | 2014-06-30 | 2018-11-07 | アマゾン・テクノロジーズ・インコーポレーテッド | Interactive interface for machine learning model evaluation |
| JP6562355B2 (en) | 2015-12-02 | 2019-08-21 | パナソニックＩｐマネジメント株式会社 | Search support method, search support device, and program |
| JP6719082B2 (en) | 2016-07-04 | 2020-07-08 | パナソニックＩｐマネジメント株式会社 | Decision tree generation device, decision tree generation method, decision tree generation program, and question system |
-
2018
- 2018-10-05 JP JP2018189979A patent/JP7028746B2/en active Active
-
2019
- 2019-10-02 WO PCT/JP2019/038845 patent/WO2020071398A1/en active Application Filing
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| US9582757B1 (en) | Scalable curation system | |
| RU2019116123A (en) | REAL-TIME ANSWER SYSTEM FOR QUESTIONS FROM DIFFERENT AREAS OF KNOWLEDGE | |
| WO2019084810A1 (en) | Information processing method and terminal, and computer storage medium | |
| CN108573702A (en) | System with the enabling phonetic function that domain ambiguity is eliminated | |
| CN110781276A (en) | Text extraction method, device, equipment and storage medium | |
| CN110162780B (en) | User intention recognition method and device | |
| RU2015138140A (en) | SYSTEMS AND METHODS OF MULTI-USER MULTI-LANGUAGE INFORMATION EXCHANGE | |
| CN111325037A (en) | Text intention recognition method and device, computer equipment and storage medium | |
| JP2003141075A5 (en) | ||
| US20120221656A1 (en) | Tracking message topics in an interactive messaging environment | |
| CN109408821B (en) | Corpus generation method and device, computing equipment and storage medium | |
| JP2014222402A (en) | Utterance candidate generation device, utterance candidate generation method, and utterance candidate generation program | |
| JP2020060846A5 (en) | ||
| CN111178081B (en) | Semantic recognition method, server, electronic device and computer storage medium | |
| CN110765776A (en) | Method and device for generating return visit labeling sample data | |
| CN108710653B (en) | On-demand method, device and system for reading book | |
| JP2018206263A (en) | Predicate term structure model generation apparatus, predicate term structure analysis apparatus, method and program | |
| US20180329983A1 (en) | Search apparatus and search method | |
| JP2013003663A (en) | Information processing apparatus, information processing method, and program | |
| CN113220854B (en) | Intelligent dialogue method and device for machine reading and understanding | |
| CN115345177A (en) | Intention recognition model training method and dialogue method and device | |
| Grant et al. | Vector space analysis of software clones | |
| JP2019046412A (en) | Information processing device, information processing method, and program | |
| WO2023144413A1 (en) | Augmenting machine learning language models using search engine results | |
| CN108959327B (en) | Service processing method, device and computer readable storage medium |