JP2018181259A - Dialogue rule collating apparatus, dialogue apparatus, dialogue rule collating method, dialogue method, dialogue rule collating program, and dialogue program - Google Patents

Abstract

When a response utterance to a user utterance is determined using a dialogue rule, the user utterance and the dialogue rule are appropriately matched. A user utterance is input, and a response utterance for a user utterance is determined by collating the user utterance and the utterance pattern based on a plurality of dialogue rules including a utterance pattern and a response utterance pair for the utterance of the utterance pattern. To do. At this time, in the dialogue, a case element that is omitted from the user utterance is supplemented or a focal word is added to the user utterance, a noun phrase in the user utterance is replaced with a synonym, and the user utterance is converted to the user utterance. The user utterance is rewritten by at least one process of replacing with the topic, and the user utterance before rewriting and the user utterance after rewriting are collated with the utterance pattern. [Selection] Figure 1

Description

  The present invention relates to an interaction rule matching device, an interaction device, an interaction rule matching method, an interaction method, an interaction rule matching program, and an interaction program that use rewriting of user utterances.

  In the dialogue system, a human being interacts with a computer to obtain various information and fulfill a request. In addition to achieving a predetermined task, there is also a dialogue system for daily conversation, which allows a person to gain mental stability, satisfy a desire for approval, and build a relationship of trust.

  The type of dialogue system is described in detail in Non-Patent Document 1.

  The most commonly used implementation of dialog systems in commercial systems is the use of dialog rules. The dialogue rule consists of an utterance pattern (pattern) and a response utterance (template). In the dialogue system using this dialogue rule, when the user's utterance matches the utterance pattern, for example, by matching a word string, the system responds using the response utterance corresponding to the utterance pattern. For example, an example of an interaction rule described in a language that describes interaction rules, which is called AIML (Artificial Intelligence Markup Language), is shown below. The asterisk "*" included in the utterance pattern is a symbol that matches any word string.

<category>
<pattern> Sake * Drink * Or </ pattern>
<template> I like alcohol </ template>
</ category>

  For example, if the user's utterance is "Can you drink alcohol?", The user's utterance matches the description content of <pattern> indicating the utterance pattern. Therefore, the system utters "I like alcohol" which is the descriptive content of <template> that indicates the response utterance.

  Non-Patent Document 2 describes AIML in detail.

Mikio Nakano, Kazunori Komaya, Kotaro Funakoshi, Yukiko Nakano, Manabu Okumura (Supervised) Dialogue System, Corona, 2016. WALLACE, Richard S. The anatomy of ALICE. In: Parsing the Turing Test. Springer Netherlands, 2009. pp. 181-210.

  When performing response processing using interaction rules, problems may occur because context is not considered. For example, suppose that there is the following rule.

<category>
<pattern>It's pretty </ pattern>
<template> Thank you </ template>
</ category>

  According to this dialogue rule, the system answers "Thank you" when the user says "You are pretty". However, it is assumed that the user and the system talk on the topic of "cat" before the utterance "It is cute". In that case, the utterance "You are pretty" probably refers to a cat, and the response "Thank you" will be inappropriate.

  In addition, in the language, it is possible to utter the same content in various ways. For example, when confirming whether the other party can drink alcohol, not only the expression "Can you drink alcohol" but also the expression "Can you drink alcohol?" As described above, it is difficult to create in advance a dialogue rule that can respond to all possible expressions of user utterances.

  Furthermore, when the user's speech and the speech pattern of the dialogue rule are collated, it may be desired to match if the contents roughly match even if the words do not match. Even in this case, in the method using the current dialogue rule, the user's utterance is compared with the utterance pattern of the dialogue rule based on whether or not the word matches, so if the words differ, the contents are roughly Even if it matches, it can not be matched.

  The present invention relates to a dialogue rule collating apparatus, a dialogue apparatus, a dialogue rule collating method, and a dialogue method, which can appropriately collate a user speech and a dialogue rule when determining a response speech to a user speech using the dialogue rule. It is an object of the present invention to provide a dialogue rule matching program and a dialogue program.

  In order to achieve the above object, the dialogue rule collating apparatus according to the present invention receives the user's utterance, and based on a plurality of dialogue rules including a combination of an utterance pattern and a response utterance to the utterance of the utterance pattern, An utterance rule matching device that determines a response utterance to the user's utterance by collating the utterance pattern, and complements a case element omitted in the user's utterance in a dialogue, or the user speaks a focused word The user's utterance before rewriting and rewriting the user's utterance by at least one process of adding to the user's utterance, replacing the noun phrase in the user's utterance with a synonym, and replacing the user's utterance with the topic of the user's utterance And a speech rule for matching each of the user speech after rewriting with the speech pattern Tsu, including the switch unit.

  The utterance rule matching unit obtains similarity with each of the utterance pattern before the rewriting and the utterance pattern compared with each of the user utterance after the rewriting and determines the similarity to each of the at least one processing. The response utterance to the user utterance may be determined based on a predetermined priority and the similarity.

  In order to achieve the above object, the dialogue apparatus according to the present invention collates a user utterance with the utterance pattern based on a plurality of dialogue rules consisting of a combination of an utterance pattern and a response utterance to the utterance of the utterance pattern. A speech apparatus for determining a response speech to the user speech, the input unit for inputting a user speech, extracting a noun phrase and a focused word from the user speech, and storing the extracted noun phrase and the focused word as dialogue history Supplementing a case element omitted in the user's speech based on at least one of a focal point extraction unit stored in a unit and the noun phrase and the focal word stored in the dialogue history storage unit, or the focal word A zero pronoun complementing unit that adds the user utterance to the user utterance; a synonym substitution unit that replaces a noun phrase in the user utterance with a synonym; The user's utterance after being rewritten by the topic extraction unit for replacing the topic in the utterance, the process by the zero pronoun complementation unit, the process by the synonym substitution unit, and the process by the topic extraction unit, and before being rewritten An utterance rule matching unit that determines the response utterance to the user utterance by matching each of the user utterances with the utterance pattern; an output unit that outputs the response utterance determined by the utterance rule matching unit; ,including.

  The utterance rule matching unit obtains the degree of similarity with each of the utterance pattern before the rewriting and the utterance pattern compared with each of the user utterance after the rewriting, and the processing by the zero pronoun complementing unit, The response utterance to the user utterance may be determined based on a predetermined priority for each of the processing by the synonym substitution unit and the processing by the topic extraction unit and the similarity.

  In order to achieve the above object, the dialogue rule matching method of the present invention includes a speech rule matching unit, a plurality of dialogue rules each having a user speech as an input, and a combination of a speech pattern and a response speech to the speech of the speech pattern. A dialogue rule matching method in a speech rule collating apparatus according to claim 4, wherein said speech rule matching unit determines the response speech to said user speech by collating said user speech and said speech pattern, wherein said speech rule matching unit Processing for complementing case elements omitted in the utterance or adding a focus word to the user utterance, processing for replacing a noun phrase in the user utterance with a synonym, and replacing the user utterance with the topic of the user utterance Rewriting the user utterance by at least one of the processing; and the utterance rule Pitch portion includes the steps of collating said speech pattern of each of said user's utterance after said user's utterance and rewriting before rewriting, the.

  In the dialogue rule collating method according to the present invention, the speech rule matching unit determines the similarity to each of the speech pattern collated with each of the user speech before rewriting and the user speech after rewriting. The method may further include determining the response utterance to the user utterance based on a predetermined priority for each of the at least one process and the similarity.

  In order to achieve the above-mentioned object, a dialogue rule collation program of the present invention is a program for making a computer function as each part of the dialogue rule collation device.

  In order to achieve the above object, the dialogue program of the present invention is a program for causing a computer to function as each part of the dialogue device.

  ADVANTAGE OF THE INVENTION According to this invention, when determining the response speech with respect to a user speech using a dialogue rule, it becomes possible to match a user speech and a dialogue rule appropriately.

It is a block diagram showing functional composition of a dialog device concerning an embodiment. It is a flow chart which shows a flow of dialogue processing concerning an embodiment. It is a flowchart which shows the flow of the subroutine of the zero pronouns complementation process which concerns on embodiment.

  Hereinafter, the present embodiment will be described using the drawings.

  FIG. 1 is a block diagram showing a functional configuration of the dialog device 10 according to the present embodiment. As shown in FIG. 1, the dialogue apparatus 10 according to the present embodiment includes an input unit 12, a focus extraction unit 14, a dialogue history storage unit 16, a zero pronoun complementation unit 18, a synonym substitution unit 20, a topic extraction unit 22, dialogue rules A storage unit 24, a speech rule matching unit 26, and an output unit 28 are provided.

  The input unit 12 receives a user utterance. The user utterance may be an utterance inputted as a text by the user, or may be an utterance in which the voice of the user utterance is converted into text using speech recognition. The input unit 12 causes the zero pronoun complementation unit 18, the synonym substitution unit 20, and the topic extraction unit 22 to output the input user's utterance, and causes the dialogue history storage unit 16 to store the same.

  The focus extraction unit 14 extracts a focus word included in the user's utterance. The focus word is a word string (mainly a noun phrase) representing a topic included in the user's utterance. For example, if the user's utterance is "I ate ramen yesterday", then "ramen" is the focus word. In this embodiment, a focus extractor is learned using a machine learning method called Conditional Random Fields (CRF), and a focus word is extracted using the learned focus extractor. The CRF is a general learning method of obtaining an estimator capable of labeling an unknown sequence from learning data when there is learning data in which a label is added to a predetermined sequence. The CRF is described in Non-Patent Document 3 below.

[Non-patent literature 3] Introduction to machine learning for language processing (Natural language processing series), Takaya Takaya (Author), Manabu Okumura (Supervised), Corona, 2010.

  In this embodiment, utterances such as "Yesterday / <center> Ramen </ center> // Eat / Tan / I / E" and "<center> England </ center> / I like /" Prepare a large amount of data tagged with <center> </ center>, which means that it is a focus word, at the location where it is a focus word, and learn the focus word extractor from these data using CRF Do.

  In the above data, slashes indicate morpheme delimiters, but these morpheme delimiters can be obtained using a morpheme analyzer. In this embodiment, a morphological analyzer called JTAG, which is disclosed in Non-Patent Document 4 below, is used. In addition, the feature template used for CRF learning includes surface forms, parts of speech, and end forms of words, but semantic information (for example, general noun attributes obtained from Japanese lexical system) and the like may be used.

[Non-patent document 4] T. Fuchi and S. Takagi, "Japanese Morphological Analyzer using Word Co-currence-JTAG", Proc. Of COLING-ACL, pp. 409-413, 1998.

  Thus, according to the learned focus extractor, "coffee" is extracted as a focus word from the user's utterance "I like coffee", for example.

  The focus extraction unit 14 also extracts noun phrases from the user's utterance using the result of morphological analysis. In the present embodiment, noun succession (a part of speech is a concatenation of noun words) is extracted as a noun phrase.

  The focus extraction unit 14 stores the extracted noun phrases and the focus words in the dialogue history storage unit 16. Multiple noun phrases and focused words may be extracted from one user utterance. Also, the noun phrase and the focused word may be the same.

  The dialogue history storage unit 16 stores, as the dialogue history, past user utterances and noun phrases and focus words extracted from past user utterances, together with the utterance numbers. The utterance number is a number for identifying each user utterance, and a continuous utterance number which is incremented each time a user utterance is input is given to each user utterance.

  The zero pronoun complementation unit 18 performs anaphoric analysis on the input user utterance, and eliminates the zero pronoun by complementing the case element omitted in the user utterance in the dialogue. Zero pronouns are abbreviated case elements. For example, when the user speaks "cute" when talking about "cat", the case element (and case particle (or cousin)) "cat is" or "cat is" is omitted. ing. In the present embodiment, the technology disclosed in Non-Patent Document 5 below is used as a Japanese technique for eliminating zero pronouns.

[Non-patent document 5] Kenji Imamura, Ryuichiro Higashinaka and Reiko Izumi, "Predential term analysis with zero pronouns for dialogue analysis," Natural language processing, Vol. 22, No. 1, pp. 3-26 , 2015.

  In this technique, for a predicate (verb, adjective, da sentence, etc.) in the user's speech, the required case dictionary is referenced, and if it is recognized that the required case is missing in the predicate, the case element is a noun phrase in the dialogue history And search from among the focus words. The way of search is determined statistically based on learning data.

  In this embodiment, for each of the predicates included in the input user's utterance, omission is permitted based on the required case dictionary for the case of ga case, case, ni case, de case, to case, kera case, and medeon case. If it is done, the case element is extracted. Although the above Non-Patent Document 5 deals with Ga-grade, Qualification, and Ni-grade, the case element is extracted using a case frame dictionary separately prepared for other case elements in addition to this technique. Specifically, noun phrases statistically significantly co-occurring in a predicate and a case are listed in advance using a text corpus, and the noun phrases listed in the dialogue history are matched with the enumerated noun phrases. If there is something to do, it extracts matching noun phrases as abbreviated case elements.

  For example, it is assumed that there are "house", "convenience store", "restaurant" etc. as noun phrases that co-occur with "de-rating" significantly with respect to the predicate "eating". If you want to extract the case element of "de-rating" omitted for the "eat" predicate, if there is "convenience store" in the noun phrase or the focus word in the context, extract "convenience store" as an omitted element .

  Further, the zero pronoun complementing unit 18 adds a focused word to the user's speech when the case element omitted in the user's speech is not complemented and the focused word is not included in the user's speech.

  The zero pronoun complementation unit 18 generates a new user utterance in which the zero pronoun is complemented or a focused word is added according to the following procedure for user utterance.

There is no focal word in the IF user's speech
Zero pronouns (abbreviated case elements) exist in the IF user's speech
For each of the omitted case elements
IF Case element omitted is not "I", "YOU", or "exog" (Rewriting 1) Add a string obtained by connecting the case element and case particle to the beginning of the utterance
There is a focused word obtained from IF dialogue history
IF Rewriting 1 has not been performed, and the focal word is not included in the user's utterance (Rewriting 2) Add a character string connecting the focal word and "ha" at the beginning of the user's utterance

  However, “I”, “YOU”, and “exog” are special contacts, respectively, in first person (when “I” is omitted) and second person (when “you” is omitted), Refers to external context (if an object not present in the context is omitted). The focus word obtained from the dialog history is the focus word added last to the dialog history. That is, it represents the latest topic in the previous dialogues.

  According to the above-described procedure, when the user utterance “I went to Italy” is stored as the dialogue history, the user utterance “good” is rewritten to “Italy good”. In addition, when the user's utterance "I ate ramen" is stored as the dialogue history, the user's utterance "I'm delicious" is rewritten as "I'm delicious ramen".

  In addition, based on at least one of the noun phrase and the focus word stored in the dialogue history storage unit 16 performed by the zero pronoun complementing unit 18 described above, the zero pronoun is complemented to the user's speech or the focus word is The zero pronoun complementation process for generating the added new user speech will be described in more detail later using the flowchart shown in FIG.

  The zero pronoun complementation unit 18 outputs the user utterance thus rewritten together with the user utterance before the rewriting to the utterance rule matching unit 26.

  The synonym substitution unit 20 refers to the synonym dictionary for the noun phrase in the input user utterance, and if there is a synonym, creates an utterance sentence in which the noun phrase is replaced with the synonym. For example, it is assumed that the synonym dictionary has "alcohol" as a synonym of "sake". In that case, when the user utterance "I like alcohol" is input, the user utterance "I like alcohol" is generated. If there are synonyms to be rewrite candidates in multiple noun phrases, and if there are multiple synonyms to be rewrite candidates in any noun phrase, etc., list all utterance sentences in which each noun phrase has been rewritten with each synonym. Alternatively, an utterance sentence in which only some noun phrases are rewritten with synonyms may be generated, or an utterance sentence in which all noun phrases are rewritten with only some synonyms may be generated.

  If many speech sentences are generated by replacing noun phrases with synonyms, it takes time to match the user speech and the speech pattern of the dialogue rule in the speech rule matching unit 26 in the latter stage. Therefore, in the present embodiment, among a plurality of noun phrases in the user's utterance, one of noun phrases that can be rewritten as synonyms (for example, the first noun phrase) is rewritten with one synonym.

  The synonym substitution unit 20 outputs the user utterance thus rewritten together with the user utterance before the rewriting to the utterance rule matching unit 26.

  The topic extraction unit 22 acquires, for the input user utterance, a character string representing a topic (topic) of the user utterance. Specifically, first, user utterances are classified into a plurality of classes using a machine learning method. The class is, for example, "sports", "cooking", "beauty" and the like. In the present embodiment, about 40 classes in total are used, and the classifier is learned in advance. When learning a classifier, prepare a large number of user utterances, assign a correct answer class as a label to each user utterance, and enter the user utterances into any of these classes when the user utterances are input. Learn the classifier to classify.

  A general text classification method can be used as a classification method for classifying user speech into any class. For example, using a word n-gram obtained from natural language utterances through morphological analysis as a feature amount, based on learning data (that is, data labeled with a class for each of a plurality of utterances), a method of machine learning You can use it to learn a classifier. Examples of learning algorithms include logistic regression, support vector machines, and deep learning methods. The general algorithm of the text classification method is disclosed in the above non-patent document 3. The deep learning method is disclosed in Non-Patent Document 6 below.

[Non-patent document 6] Yoon Kim, "Convolutional Neural Networks for Sentence Classification", Proceedings of the 2014 Conference on Empirical Methods in Natural Language Processing (EMNLP), pp. 1746-1751, 2014.

  The topic extraction unit 22 outputs the class name (for example, "sports", "cooking", etc.) of the class into which the user's speech has been classified as the user's speech after rewriting, together with the user's speech before the rewriting, to the speech rule matching unit 26. Do.

  The dialogue rule storage unit 24 stores data indicating a plurality of dialogue rules each including a combination of an utterance pattern and a response utterance to the utterance of the utterance pattern. The dialogue rule is a set of dialogue rules described in the above-described AIML format. In the present embodiment, tens of thousands of dialogue rules are created and stored in the dialogue rule storage unit 24.

  The utterance rule matching unit 26 utters each of the user utterances after being rewritten by the processing by the zero pronoun complementation unit 18, the processing by the synonym substitution unit 20, and the processing by the topic extraction unit 22, and the user utterances before being rewritten Match the pattern. As the collation method, for example, one based on forward coincidence as shown in the above-mentioned Non-Patent Document 2 may be used.

  In this embodiment, for each of the utterance patterns of the plurality of dialogue rules stored in the dialogue rule storage unit 24, a word set consisting of other than the asterisk "*" described in the utterance pattern and the user utterance are decomposed into morphemes The word set obtained by the calculation is expressed as a word vector, weighted by TF-IDF (Term Frequency-Inverse Document Frequency), and then calculated the cosine similarity between the word vectors, and the cosine similarity is The response utterance corresponding to the highest utterance pattern is output. The weight of TF-IDF is calculated by using a large amount of dialogue data, newspaper articles and the like.

  Here, let U1 be the user utterance before rewriting, let U2 be the user utterance rewritten by the processing by the zero pronoun complementing unit 18, let U3 be the user utterance rewritten by the processing by the synonym substitution unit 20, and let the topic extracting unit 22 do so. Let U4 be the user utterance rewritten by the process. In this case, each of U1, U2, U3, and U4 is compared with the utterance patterns of the plurality of dialogue rules stored in the dialogue rule storage unit 24, and the response utterance of the system is determined based on the collation result.

  If the cosine similarity obtained for each of U1, U2, U3 and U4 is cos (U1), cos (U2), cos (U3) and cos (U4), the system is the speech pattern with the highest cosine similarity. The response utterance is determined as the response utterance to the user utterance, and the determined response utterance is output to the output unit 28.

  Alternatively, the response utterance to the user's utterance based on the priority determined in advance for each of the processing by the zero pronoun complementation unit 18, the processing by the synonym substitution unit 20, and the processing by the topic extraction unit 22 and the degree of cosine similarity. You may decide For example, when the priority order is determined in the order of the processing by the zero pronoun complementing unit 18, the processing by the synonym replacing unit 20, and the processing by the topic extracting unit 22 before rewriting, cos (U1) is a predetermined threshold (for example, If U exceeds 0.8), output U1; otherwise, check whether cos (U2) exceeds a predetermined threshold. Also, if cos (U2) exceeds the predetermined threshold, U2 is output, and if not, it is checked whether cos (U3) exceeds the predetermined threshold. If cos (U3) exceeds a predetermined threshold, U3 is output, otherwise U4 is output.

  The method of calculating the similarity is not limited to the method using cosine similarity, but a method using general Jaccard coefficients as the similarity between sets, such as a word set included in the utterance pattern and a word included in the user's utterance Any method may be used as long as it can calculate the degree of similarity with the set.

  The output unit 28 outputs the response utterance determined by the utterance rule matching unit 26 as a system utterance. As an output method, there is a method of displaying data indicating a response utterance on a display means such as a display, transmitting data indicating a response utterance to an external device, or outputting a voice indicating a response utterance by a voice output unit. It can be mentioned.

  The dialog device 10 according to the present embodiment is configured by, for example, a computer device provided with a central processing unit (CPU), a random access memory (RAM), and a read only memory (ROM) for storing various programs. Moreover, the computer which comprises the interactive apparatus 10 may be equipped with memory | storage parts, such as a hard-disk drive and non-volatile memory. In the present embodiment, when the CPU reads and executes the program stored in the storage unit such as the ROM and the hard disk, the above-described hardware resource and the program cooperate with each other to realize the above-described function.

  Next, the flow of dialogue processing by the dialogue device 10 according to the present embodiment will be described using the flowchart shown in FIG. In the present embodiment, the dialogue processing is started at the timing when predetermined data for starting the execution of the dialogue processing is input to the dialogue device 10, but the timing at which the dialogue processing is started is not limited to this, For example, the dialogue processing may be started at the timing when the user's speech is input.

  In step S101, the input unit 12 inputs a user utterance.

  In step S103, the focus extraction unit 14 extracts noun phrases and focus words from the input user speech.

  In step S105, the focus extraction unit 14 causes the dialogue history storage unit 16 to store the input user utterance and the noun phrase and the focused word extracted from the user utterance.

  In step S107, the zero pronoun complementation unit 18 performs the above-described zero pronoun complementation processing.

  Here, the flow of the zero pronoun complementation process by the zero pronoun complementation unit 18 will be described using the flowchart shown in FIG.

  In step S201, the zero pronoun complementation unit 18 acquires the input user utterance.

  In step S203, the zero pronoun complementation unit 18 determines whether or not a focused word is present in the user's speech. If it is determined in step S203 that a focused word is present in the user's utterance (S203, Y), the execution of the program of this zero pronoun complementation processing is ended, and the process proceeds to step S109 in FIG. If it is determined in step S203 that no focused word is present in the user's utterance (S203, N), the process proceeds to step S205.

  In step S205, the zero pronoun complementation unit 18 determines whether a zero pronoun exists in the user's speech. If it is determined in step S205 that a zero pronoun exists in the user utterance (S205, Y), the process proceeds to step S207. If it is determined in step S205 that no zero pronoun exists in the user's speech (S205, N), the process proceeds to step S215.

  In step S207, the zero pronoun complementation unit 18 selects one omitted case element from the user utterance.

  In step S209, the zero pronoun complementing unit 18 determines whether the case element selected in step S207 is any of "I", "YOU", and "exog". If it is determined in step S209 that the case element selected in step S207 is any of "I", "YOU", and "exog" (S209, Y), the process proceeds to step S213. If it is determined in step S209 that the case element selected in step S207 is not one of “I”, “YOU”, and “exog” (S209, N), the process proceeds to step S211.

  In step S211, the zero pronoun complementing unit 18 adds a character string in which the case element and the case particle are connected at the beginning of the user's speech.

  In step S213, the zero pronoun complementing unit 18 determines whether there are unprocessed case elements, that is, case elements for which the processes in steps S207 to S211 have not been performed. If it is determined in step S213 that there is an unprocessed case element (S213, Y), the process returns to step S207 to select an unprocessed case element. If it is determined in step S213 that there is no unprocessed case element (S213, N), the process proceeds to step S215.

  In step S215, the zero pronoun complementation unit 18 determines whether the above-described focused word obtained from the dialogue history exists. If it is determined in step S215 that the in-focus word obtained from the dialogue history is present (S215, Y), the process proceeds to step S217. If it is determined in step S215 that the in-focus word obtained from the dialogue history does not exist (S215, N), the execution of the program of the zero pronoun complementing process is ended, and the process proceeds to step S109 in FIG.

  In step S217, the zero pronoun complementation unit 18 determines whether or not the completion of the zero pronoun (the process of step S211) is not performed, and the focused word is not included in the user utterance. When it is determined in step S217 that the zero pronoun complementation has not been performed and the in-focus word is not included in the user's utterance (S217, Y), the process proceeds to step S219. If it is determined in step S217 that the zero pronoun has been complemented or if the in-focus word is included in the user's utterance (S217, N), the execution of the program of this zero pronoun complement processing is terminated, and the figure is It transfers to step S109 of 2.

  In step S219, the zero pronoun complementing unit 18 adds a character string obtained by concatenating the focused word and "ha" at the beginning of the user speech, and the process proceeds to step S109 in FIG.

  In step S109 of FIG. 2, the synonym substitution unit 20 replaces the input noun phrase in the user's utterance with the synonym.

  In step S111, the topic extraction unit 22 replaces the input user utterance with the topic of the user utterance.

  In step S113, the speech rule matching unit 26 rewrites the user speech after the user speech is rewritten by the processing by the zero pronoun complementing unit 18, the processing by the synonym substitution unit 20, and the processing by the topic extracting unit 22, and rewriting The response utterance to the user utterance is determined by matching each of the user's previous utterances with the utterance pattern of the dialogue rule.

  In step S115, the output unit 28 outputs the determined response utterance, and ends the execution of the program of the dialogue processing.

  As described above, in the present embodiment, processing for complementing the case element omitted in the user's utterance or adding a focused word to the user's utterance in the dialogue, processing for replacing the noun phrase in the user's utterance with the synonym, and A response to the user's utterance by rewriting the user's utterance by at least one process of replacing the user's utterance with the topic of the user's utterance, and collating the user's utterance before the rewriting and the user's utterance after the rewriting with the utterance pattern Determine your speech.

  In this way, it is possible to realize an utterance response using only interaction rules that are useful to the business operator. Then, the possibility of business using the dialogue system is enhanced, and human beings and computers can cooperate efficiently.

  In the present embodiment, the operation of the component of the function shown in FIG. 1 is constructed as a program, installed in a computer used as the interactive apparatus 10 and executed, but the present invention is not limited to this. It is good.

  Further, the constructed program may be stored in a portable storage medium such as a hard disk or a CD-ROM, and may be installed in a computer or distributed.

DESCRIPTION OF SYMBOLS 10 Dialogue device 12 Input part 14 Focus extraction part 16 Dialogue history storage part 18 Zero pronoun complement part 20 Synonym substitution part 22 Topic extraction part 24 Dialogue rule storage part 26 Speech rule matching part 28 Output part

Claims (8)

A response utterance to the user utterance is determined by collating the user utterance with the utterance pattern based on a plurality of dialog rules including a user utterance as an input and a combination of an utterance pattern and a response utterance to the utterance of the utterance pattern. A speech rule matching device that
Processing for complementing the case element omitted in the user's speech or adding a focus word to the user's speech in dialogue, processing for replacing a noun phrase in the user's speech with a synonym, and the user's speech Rewrite the user utterance by at least one process of processing to replace the topic of the utterance;
A dialogue rule matching device comprising: a speech rule matching unit which matches each of the user speech before rewriting and the user speech after rewriting with the speech pattern. The speech rule matching unit obtains similarity between each of the user speech before rewriting and each of the speech patterns collated with each of the user speech after rewriting, and for each of the at least one process in advance The dialogue rule collating apparatus according to claim 1, wherein the response utterance to the user utterance is determined based on a predetermined priority and the similarity. An utterance apparatus that determines a response utterance to a user's utterance by collating the user's utterance with the utterance pattern based on a plurality of dialogue rules including a combination of an utterance pattern and an utterance response to the utterance of the utterance pattern.
An input unit for inputting user utterances;
A focus extraction unit which extracts a noun phrase and a focused word from the user's speech and stores the extracted noun phrase and the focused word in a dialogue history storage unit;
A zero pronoun that complements the case element omitted in the user's utterance or adds the focal word to the user's utterance based on at least one of the noun phrase and the focal word stored in the dialogue history storage unit The complement section,
A synonym substitution unit that replaces noun phrases in the user utterance with synonyms;
A topic extraction unit that replaces the user utterance with a topic of the user utterance;
The user utterance after being rewritten by the processing by the zero pronoun complementation unit, the processing by the synonym substitution unit, and the processing by the topic extraction unit, and the user utterance before being rewritten are the utterance patterns An utterance rule matching unit that determines the response utterance to the user utterance by matching;
An output unit that outputs the response utterance determined by the utterance rule matching unit;
Interactive device including: The utterance rule matching unit obtains the degree of similarity with each of the utterance pattern before rewriting and the utterance pattern matched with each of the user utterance after rewriting, and the processing by the zero pronoun complementation unit, the synonym substitution The interactive apparatus according to claim 3, wherein the response utterance to the user's utterance is determined based on a predetermined priority for each of processing by a part and processing by the topic extraction unit, and the degree of similarity. The user's utterance is input by including an utterance rule matching unit, and the user's utterance is compared with the utterance pattern based on a plurality of dialogue rules including an utterance pattern and a response utterance to the utterance of the utterance pattern. A dialogue rule matching method in a speech rule collating device for determining a response speech to a user speech, comprising:
A process of the speech rule matching unit complementing a case element omitted from the user's speech or adding a focused word to the user's speech in a dialogue, a process of replacing a noun phrase in the user's speech with a synonym, Rewriting the user utterance by at least one process of replacing the user utterance with a topic of the user utterance.
The speech rule matching unit matching the user speech before rewriting and the user speech after rewriting with the speech pattern;
Interaction rule matching method including. The speech rule matching unit determines similarity with each of the user speech before rewriting and each of the speech patterns collated with each of the user speech after rewriting, and for each of the at least one process in advance The dialogue method according to claim 5, further comprising the step of determining the response utterance to the user utterance based on a predetermined priority and the similarity.   An interactive rule matching program for causing a computer to function as each unit of the interactive rule matching device according to claim 1 or 2.   An interactive program for causing a computer to function as each unit of the interactive device according to claim 3 or 4.