JPH0431146B2 - - Google Patents
Info
- Publication number
- JPH0431146B2 JPH0431146B2 JP60225188A JP22518885A JPH0431146B2 JP H0431146 B2 JPH0431146 B2 JP H0431146B2 JP 60225188 A JP60225188 A JP 60225188A JP 22518885 A JP22518885 A JP 22518885A JP H0431146 B2 JPH0431146 B2 JP H0431146B2
- Authority
- JP
- Japan
- Prior art keywords
- question
- input
- response
- answering system
- question answering
- 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.)
- Expired - Lifetime
Links
- 230000004044 response Effects 0.000 claims description 23
- 238000000034 method Methods 0.000 claims description 16
- 239000002245 particle Substances 0.000 claims description 6
- 230000006403 short-term memory Effects 0.000 claims description 4
- 239000000284 extract Substances 0.000 claims description 3
- 238000010586 diagram Methods 0.000 description 5
- 230000008569 process Effects 0.000 description 4
- 230000014509 gene expression Effects 0.000 description 2
- 230000002452 interceptive effect Effects 0.000 description 2
- 230000007787 long-term memory Effects 0.000 description 2
- 230000007774 longterm Effects 0.000 description 2
- 230000000877 morphologic effect Effects 0.000 description 2
- 230000000694 effects Effects 0.000 description 1
- 230000006872 improvement Effects 0.000 description 1
- 230000003993 interaction Effects 0.000 description 1
- 239000003550 marker Substances 0.000 description 1
- 230000007246 mechanism Effects 0.000 description 1
- 230000015654 memory Effects 0.000 description 1
Landscapes
- Machine Translation (AREA)
- Information Retrieval, Db Structures And Fs Structures Therefor (AREA)
- Devices For Executing Special Programs (AREA)
Description
【発明の詳細な説明】
〈産業上の利用分野〉
本発明は診断装置等の対話形式の質問応答シス
テムに関し、特に質問応答システムの入力方式の
改良に関するものである。DETAILED DESCRIPTION OF THE INVENTION <Industrial Application Field> The present invention relates to an interactive question answering system for a diagnostic device or the like, and particularly relates to an improvement in the input method of the question answering system.
〈従来の技術〉
最近、エキスパートシステムと称し、コンピユ
ータシステムを基本とし、専問的知識をデータベ
ースとして利用した推論機構を有する種々の対話
形式の質問応答システムが開発されている。<Prior Art> Recently, various interactive question answering systems called expert systems have been developed, which are based on computer systems and have reasoning mechanisms that utilize specialized knowledge as a database.
それらの中には、第１図に示す如く内部に予め
ルール、事実を書込んだ長期記憶領域１２と、シ
ステムが推論中にユーザーの応答に応じて事実を
書込む短期記憶領域１３の２つの記憶領域を保有
し、上記両記憶領域のルール・事実に基づいてユ
ーザーの質問の内部形式を決めるようにしたもの
がある。こゝで内部形式とは第２図にその一例を
示す如くシステム内部における事実の記述形式を
いう。第４図はそのような従来の質問応答システ
ムの入力文の処理フローである。ユーザーからの
入力文（「もりが９じにしゆつちようする。」）は
先ず形態素解析S32（森／が／９／時／に／出張／
する）で文単語単位に分離され、次いで構文解析
S33（森が／９時に／出張する／）により入力文の
構文及び係り受けが判断される。その後意味解析
S34（森⇒ag，９⇒time，時⇒接続詞，等）によ
り各部の意味が解釈され入力文に対する内部形式
が決められる（S35）。 As shown in Fig. 1, there are two long-term storage areas 12 in which rules and facts are written in advance, and a short-term storage area 13 in which facts are written in response to the user's responses during the system's inference. There is a system that has a storage area and determines the internal format of the user's question based on the rules and facts of both storage areas. The internal format here refers to the description format of facts within the system, as shown in FIG. FIG. 4 shows a processing flow of input sentences in such a conventional question answering system. The input sentence from the user (``Mori is going to do it at 9/day.'') is first subjected to morphological analysis S 32 (Mori/ga/9/hour/ni/business trip/
), the sentence is separated into word units, and then parsed.
The syntax and dependencies of the input sentence are determined by S 33 (Mori ga /9 o'clock/Going on a business trip/). Then semantic analysis
The meaning of each part is interpreted by S 34 (forest⇒ag, 9⇒time, time⇒conjunction, etc.) and the internal format for the input sentence is determined ( S35 ).
この様な質問応答システムではユーザーの応答
から内部形式を抽出する際の意味解析において、
ユーザーからの応答が意味解析に用いる諸条件の
一つでも満足しない場合、再度同一問合せを繰返
しシステムからの質問に正確に対応する応答を求
めていた。 In this type of question answering system, semantic analysis is performed when extracting the internal form from the user's response.
If the response from the user does not satisfy even one of the conditions used for semantic analysis, the same query is repeated again in order to obtain a response that accurately corresponds to the question from the system.
例えば、出張申請用質問応答システムにおい
て、システムからの「だれが出張しますか？」と
の質問に対し、「森です」とか「森」とかの省略、
入力をユーザーが行つた場合、意味解析が出来な
いことからシステムはそれらの応答から必要とす
る事実を抽出できず、「森が出張します。」との応
答が得られるまで同じ質問を繰返すとかより正確
な応答を求めるコメントを出力するとかの処理を
行つていた。 For example, in a question-and-answer system for business trip applications, when the system asks, "Who is going on a business trip?"
If the user enters input, the system cannot extract the necessary facts from the responses because it cannot perform semantic analysis, and may repeat the same question until it gets the response "Mori is going on a business trip." It was performing processing such as outputting comments to request more accurate responses.
〈発明が解決しようとする問題点〉
このように従来の質問応答システムに於ては、
ユーザーはシステムからの質問に対して正確に応
答する必要があり、システムの知りたい事実以外
の個所でも正確に質問文に対応した応答、即ち省
略のない応答を行わなければならず、システムか
らの質問への応答が煩雑であるという問題点があ
つた。<Problems to be solved by the invention> As described above, in the conventional question answering system,
The user must respond accurately to the questions asked by the system, and even in areas other than the facts that the system wants to know, the user must respond accurately to the question, that is, respond without omissions, and the user must respond accurately to the questions asked by the system. There was a problem that responding to questions was complicated.
本発明はこのような点に鑑みてなされたもの
で、システムからの質問に対して省略された応答
でも入力できる質問応答システムの入力方式の提
供を目的としている。 The present invention has been made in view of the above points, and an object of the present invention is to provide an input method for a question answering system in which even omitted responses to questions from the system can be input.
〈問題点を解決するための手段〉
上記目的を達成するため、本発明の質問応答シ
ステムの入力方式はシステムから発せられる質問
文の疑問代名詞に、質問者から前記質問文に対応
して入力される応答文を意味解析して得られた格
体言を対応せしめることにより推論のための事実
を抽出するように為し、且つ前記意味解析を格助
詞、接尾語及び意味マーカーのいずれか一つ又は
一つ以上の組合せで行なうようにしたことを特徴
としている。<Means for Solving the Problems> In order to achieve the above object, the input method of the question answering system of the present invention is such that an interrogative pronoun in a question sentence issued from the system is input by the questioner in correspondence with the question sentence. The system extracts facts for inference by semantically analyzing a response sentence and associating the obtained case phrases, and the semantic analysis is performed using any one of case particles, suffixes, and semantic markers It is characterized in that it is performed in combination of one or more.
〈作用〉
本発明に係る質問応答システムの入力方式は上
述の通り質問者からの応答文の他に、システムか
ら質問者へ発せられる質問文も考慮しているか
ら、前記応答文の意味解析を厳重にしなくても推
論に必要な事実を抽出することができる。<Operation> The input method of the question answering system according to the present invention takes into consideration not only the response sentences from the questioner as described above but also the question sentences sent from the system to the questioner. It is possible to extract the facts necessary for inference without being too strict.
〈実施例〉
第１図は本発明を実施する場合に用いる機能ブ
ロツク図である。図中、一点鎖線の左側がユーザ
（質問者）側で、右側が質問応答システム側１０
である。質問応答システム１０は大きく分けて推
論部１６、長期記憶部１２及び短期記憶部１３か
ら構成され、それにユーザ側から長期記憶部１２
への入力時に使用するエデイタ１１とユーザと推
論部１６間の入出力時に使用する入力インターフ
エース（Ｉ／Ｆ）１４、及び出力Ｉ／Ｆ１５が付
加されている。長期記憶部１２には質問応答シス
テムで使用する各種規則（ルール）と固定的なデ
ータベースが予め外部（ユーザ側）から書込まれ
ており、推論部１６からその内容を書込むことは
できない。一方短期記憶部１３は、システムとユ
ーザとのやりとりの過程で得られた事実等を書込
んだり読み出したりするメモリ領域であり、推論
部１６からの読出し・書込みが可能な構成になつ
ている。入力Ｉ／Ｆ１４はユーザーが音声若しく
はKB等で入力した入力文（応答文）を推論部１
６で利用できる内部形式に変換するところであ
り、具体的には形態素、構文及び意味の各解析を
行つている。出力Ｉ／Ｆ１５は推論部１６からの
出力をユーザーが理解できる文章若しくは音声に
変換する部分である。<Embodiment> FIG. 1 is a functional block diagram used when implementing the present invention. In the figure, the left side of the dashed line is the user (questioner) side, and the right side is the question answering system side 10.
It is. The question answering system 10 is broadly divided into an inference section 16, a long-term memory section 12, and a short-term memory section 13.
An input interface (I/F) 14 and an output I/F 15 are added, which are used for input/output between an editor 11, a user, and an inference section 16. Various rules and a fixed database used in the question answering system are written in advance from the outside (user side) into the long-term storage unit 12, and the contents thereof cannot be written from the inference unit 16. On the other hand, the short-term memory section 13 is a memory area in which facts etc. obtained in the process of interaction between the system and the user are written and read, and is configured to be readable and writable by the inference section 16. The input I/F 14 inputs the input sentence (response sentence) inputted by the user by voice or KB etc. into the inference unit 1.
6. Specifically, morphological, syntactic, and semantic analysis is performed. The output I/F 15 is a part that converts the output from the inference section 16 into sentences or sounds that the user can understand.
第２図は質問応答システム１０で用いる内部形
式の一例であり、用言単位２１に格体言２２ｂと
その深層格２２ａ、推尾語２２ｃ、文番号２２
ｅ、肯定／否定の別２２ｄ等の情報２２を対応付
ける構成になつている。 FIG. 2 is an example of an internal format used in the question answering system 10, in which the predicate unit 21 includes a case 22b, its deep case 22a, a tail 22c, and a sentence number 22.
The configuration is such that information 22 such as e, affirmation/negation distinction 22d, etc. are associated with each other.
次に本発明の動作を第３図の動作フローに従つ
て詳細に説明する。ユーザーからの入力（応答）
文に対し、それがシステムからの質問に対してな
されたのか否かのチエツクS42が行われ、質問に
対応していない場合には従来の意味解析S48を経
た後内部形式が決められる（S46）が、システム
からの質問に対応した入力文の場合には省略入力
か否かのチエツクS43が行われる。省略入力の場
合には接尾語、格助詞及び意味マーカーチエツク
による本発明に係る意味解析が行われ、その結果
得られた格体言をシステムからユーザーに発した
質問文の疑問代名詞の個所に挿入する（S45）こ
とにより内部形式が決められる（S46）。又、省略
入力でない場合には質問と入力の用言の異同チエ
ツクを経て同じ場合には本発明に係る意味解析
S45を、異る場合には従来方式の意味解析S48を経
て内部形式S46が決められる。これを「何時に出
発するか？」とのシステムからの質問に「９時」
とユーザーが省略入力（応答）した場合について
説明する。この場合、システムは「出発する」と
いう用言に対して時間の格体言が不明であること
から上記質問を発しており、これに対する応答
「９時」には「時」という接尾語がある。従つて
「時」の前にある「９」を質問文で不確定であつ
た格体言と判断し、格体言に「９」を挿入して
「出発する」という用言の内部形式を確定する。 Next, the operation of the present invention will be explained in detail according to the operation flow shown in FIG. Input (response) from the user
A check S 42 is performed on the sentence to see if it corresponds to a question from the system, and if it does not correspond to a question, the internal format is determined after passing through conventional semantic analysis S 48 ( If S46 ) is an input sentence corresponding to a question from the system, a check S43 is performed to see if it is an omitted input. In the case of omitted input, semantic analysis according to the present invention is performed by checking suffixes, case particles, and semantic markers, and the resulting case pronoun is inserted into the interrogative pronoun in the question sent from the system to the user. (S 45 ) The internal format is determined (S 46 ). In addition, if the input is not an omitted input, the question and the input phrase are checked for the sameness, and if they are the same, the semantic analysis according to the present invention is performed.
S 45 and, if different, an internal form S 46 is determined through a conventional semantic analysis S 48 . When the system asked, "What time do you want to leave?", the answer was "9 o'clock."
The following describes the case where the user makes an abbreviated input (response). In this case, the system issues the above question because the time case is unknown for the phrase "leaving," and the response "9 o'clock" has the suffix "hour." Therefore, we judge the ``9'' before ``time'' to be a case phrase, which was uncertain in the question sentence, and insert ``9'' into the case phrase to determine the internal form of the predicate ``depart''. .
尚、意味解析の方法としては種々のものが考え
られるが、本発明においては、格助詞、接尾語及
び意味マーカーをチエツクすることで行つてい
る。格助詞及び接尾語は格体言の次に来るもので
あり格体言の抽出に有効なものである。意味マー
カーは表層格の種類を示す符号であり、構文解析
処理S33において使用する辞書の各表層格に付さ
れている。従つて、構文解析処理S33で付された
入力文の各表層格の意味マーカー、接尾語、格助
詞の各情報を元にして入力文に対する深層格を判
男している。 Various methods can be considered for semantic analysis, but in the present invention, it is performed by checking case particles, suffixes, and semantic markers. Case particles and suffixes come after case expressions and are effective in extracting case expressions. The meaning marker is a code indicating the type of surface case, and is attached to each surface case in the dictionary used in the parsing process S33 . Therefore, the deep case of the input sentence is determined based on the information on the semantic markers, suffixes, and case particles of each surface case of the input sentence added in the syntactic analysis process S33 .
〈発明の効果〉
以上の説明で明らかなように、本発明は質問応
答システムからユーザーへ発せられる質問文を考
慮してユーザーの入力（応答）を判断するように
したから、ユーザーからの入力が省略されたもの
であつても容易に内部形式を決定できる。従つ
て、ユーザーは質問応答システムからの質問に簡
潔に応答できると共にシステムからの同一質問の
繰返しに対する応答を除去でき、入力の煩雑さを
除去できる点で実用上極めて有用なものである。<Effects of the Invention> As is clear from the above explanation, the present invention determines the user's input (response) by considering the question text issued to the user from the question answering system. Even if it is omitted, the internal format can be easily determined. Therefore, the user can easily respond to questions from the question answering system, and can eliminate responses to repeated questions from the system, which is extremely useful in practice in that it can eliminate the complexity of input.
第１図は本発明方式に係る質問応答システムの
機能ブロツク図、第２図は内部形式の一例を示す
図、第３図は本発明方式に係る入力処理フロー
図、第４図は従来の入力処理フロー図である。
１１はエデイタ、１２は長期記憶部、１３は短
期記憶部、１６は推論部。
Fig. 1 is a functional block diagram of the question answering system according to the method of the present invention, Fig. 2 is a diagram showing an example of the internal format, Fig. 3 is an input processing flow diagram according to the method of the present invention, and Fig. 4 is a conventional input method. It is a process flow diagram. 11 is an editor, 12 is a long-term memory section, 13 is a short-term memory section, and 16 is an inference section.
Claims (1)
知識が与えられた装置に対して質問、検索、命令
が行えると共に、装置がその質問、検索、命令に
対する応答を行うようにした質問応答システムで
あつて、該質問応答システムからの質問文に対応
して入力された使用者からの応答文を意味解析
し、推論のための事実を抽出するようにした質問
応答システムの入力方式において、 該質問応答システムの内部知識における変数を
具体化する事実が存在しない場合にその変数と変
数を含む具体化しようとする事実に対応する内部
形式を基に質問文を短期記憶領域に生成し、使用
者に対して質問文を問い掛け、該質問文に対する
応答をユーザーが入力手段より自然言語で入力し
た際に、入力された応答文に対して意味解析が行
なわれ、解析の結果得られた用言と格体言のペア
とそれに付随する意味的情報の集合の中から必要
とする応答文に固有とする格体言と質問文とを対
応させることにより、推論のための事実を抽出す
るようにしたことを特徴とする質問応答システム
の入力方式。 ２ 応答文の意味解析を格助詞、接続語及び意味
マーカーのいずれか一つまたは一つ以上の組み合
わせで行うようにしたことを特徴とする特許請求
の範囲第１項記載の質問応答システムの入力方
式。[Claims] 1. A question answering system in which questions, searches, and commands can be made to a device to which knowledge is given in advance through natural language or programming, and the device responds to the questions, searches, and commands. In an input method of a question answering system, the input method of a question answering system is configured to semantically analyze a response text input from a user corresponding to a question text from the question answering system and extract facts for inference. If there is no fact that embodies a variable in the internal knowledge of the response system, a question sentence is generated in the short-term memory area based on the variable and the internal form that corresponds to the fact that is to be instantiated, and the question is sent to the user. When a user asks a question and inputs a response to the question in natural language using an input means, semantic analysis is performed on the input response, and the predicates and cases obtained as a result of the analysis are analyzed. The feature is that the facts for inference are extracted by matching the case phrases specific to the required response sentences and the question sentences from a set of pairs of nominal phrases and the accompanying semantic information. An input method for a question answering system. 2. Input to the question answering system according to claim 1, characterized in that the semantic analysis of the response sentence is performed using any one or a combination of case particles, conjunctions, and meaning markers. method.
Priority Applications (4)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| JP60225188A JPS6282429A (en) | 1985-10-07 | 1985-10-07 | Input system of interrogation and response system |
| US06/915,021 US4837689A (en) | 1985-10-07 | 1986-10-03 | Inputting and editing system in a knowledge based inquiry and answer system |
| EP86113730A EP0218218A3 (en) | 1985-10-07 | 1986-10-03 | An inputting system and an editing system in an inquiry-and-answer system |
| US07/333,565 US4931926A (en) | 1985-10-07 | 1989-04-05 | Inputting system and an editing system in an inquiry-and-answer system |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| JP60225188A JPS6282429A (en) | 1985-10-07 | 1985-10-07 | Input system of interrogation and response system |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| JPS6282429A JPS6282429A (en) | 1987-04-15 |
| JPH0431146B2 true JPH0431146B2 (en) | 1992-05-25 |
Family
ID=16825340
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| JP60225188A Granted JPS6282429A (en) | 1985-10-07 | 1985-10-07 | Input system of interrogation and response system |
Country Status (1)
| Country | Link |
|---|---|
| JP (1) | JPS6282429A (en) |
Families Citing this family (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| JP3575242B2 (en) * | 1997-09-10 | 2004-10-13 | 日本電信電話株式会社 | Keyword extraction device |
-
1985
- 1985-10-07 JP JP60225188A patent/JPS6282429A/en active Granted
Also Published As
| Publication number | Publication date |
|---|---|
| JPS6282429A (en) | 1987-04-15 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| Gleich et al. | Ambiguity detection: Towards a tool explaining ambiguity sources | |
| US10460028B1 (en) | Syntactic graph traversal for recognition of inferred clauses within natural language inputs | |
| EP1217533A2 (en) | Method and computer system for part-of-speech tagging of incomplete sentences | |
| Callaway et al. | Pronominalization in generated discourse and dialogue | |
| Loritz | GPARS: A suite of grammar assessment systems | |
| JPH0431146B2 (en) | ||
| Core | Analyzing and predicting patterns of DAMSL utterance tags | |
| Vandeventer Faltin | Syntactic error diagnosis in the context of computer assisted language learning | |
| Horacek et al. | From Meaning to Meaning: A Walk through WISBER’s Semantic-Pragmatic Processing | |
| Fong et al. | Treebank parsing and knowledge of language | |
| Godea et al. | Automatic generation and classification of minimal meaningful propositions in educational systems | |
| Rush | A Computer Assisted Language Analysis System. | |
| Prakash et al. | Domain Specific Text Preprocessing for Open Information Extraction | |
| CN119741914B (en) | Robot training method, device, equipment, storage medium and program product | |
| Limber | Syntax and sentence interpretation | |
| Deveci | Transformer Models for Translating Natural Language Sentences into Formal Logical Expressions | |
| US11106717B2 (en) | Automatic identification and clustering of patterns | |
| Biébow et al. | Acquisition and validation: from text to semantic network | |
| Vadali et al. | Artificial Intelligence-Based Automatic Evaluation Engine | |
| Wildenburg | Investigations into Semantic Underspecification in Language Models | |
| Shardlow | Lexical simplification: optimising the pipeline | |
| Saffran | Non-linguistic constraints on the acquisition of phrase structure | |
| HK40088467A (en) | Data structuring method, device, equipment, storage medium and program product | |
| HK40088467B (en) | Data structuring method, device, equipment, storage medium and program product | |
| Dorotich | A prolog implementation of a subset of Marcus’ parser and its relation to the handling of extragrammatical input |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| LAPS | Cancellation because of no payment of annual fees |