JP2008276543A - Interactive processing apparatus, response sentence generation method, and response sentence generation processing program - Google Patents

Abstract

<P>PROBLEM TO BE SOLVED: To provide an interactive processing apparatus for enabling various interaction with a user, a response sentence generation method, and a response sentence generation processing program. <P>SOLUTION: Input sentence data indicating an input sentence to be input from a user in interaction with a user are acquired by an input device 14, and a positive response sentence is generated with respect to the input sentence indicated by the input sentence data acquired by the input device 14 by a computer main body 16, and a response sentence generated by the computer main body 16 is displayed by a display 12. <P>COPYRIGHT: (C)2009,JPO&INPIT

Description

  The present invention relates to a dialogue processing apparatus, a response sentence generation method, and a response sentence generation processing program, and more particularly to a dialogue processing apparatus and response sentence generation for generating a response sentence for an input sentence from a user and interacting with the user. The present invention relates to a method and a response sentence generation processing program.

  As a technique for performing various expressions for a user in a dialog processing apparatus that performs a dialog with the user, Patent Literature 1 searches related words for a main word included in an input sentence input from the user. A technique for expressing a set of main words and related words as an expanded expression of an input is disclosed. For example, paraphrasing “wisdom, intelligence, intelligence, knowledge” is performed for “knowledge”.

  Patent Document 2 discloses a technique for performing a more appropriate search by matching a paraphrase phrase pattern to an input sentence and paraphrasing the input sentence into various expressions. For “+ numeric string +“ year ””, 1988 is added to the value of the numeric string, and this is used as the year of the year. With this paraphrase rule, for example, “2000” is converted to “2000 AD”.

Further, Patent Document 3 generates a paraphrase expression that is a synonym for the input language expression, and stores the language expression and a higher-level language expression that is a higher-level language expression common to those language expressions in association with each other. Extracting at least a part of the upper language expression from the electronic dictionary, displaying the extracted upper language expression and at least a part of the stored language expression at the same time, and displaying one or more languages from the displayed language expression A technique for registering an appropriate paraphrase expression by accepting designation of an expression and designation of one upper language expression from the displayed upper language expression is disclosed.
JP 2004-240605 A JP 2002-278988 A JP 2003-173338 A

  However, Patent Documents 1 to 3 are techniques for rephrasing a part of an input sentence as a related word, a synonym, or an appropriate paraphrased expression. There was a problem that various dialogues with the user could not be performed.

  SUMMARY An advantage of some aspects of the invention is that it provides an interactive processing device, a response sentence generation method, and a response sentence generation processing program that enable various interactions with a user.

  In order to achieve the above object, the invention according to claim 1 is acquired by an acquisition unit that acquires input sentence data indicating an input sentence to be input from the user when interacting with the user, and acquired by the acquisition unit. Generation means for generating a positive response sentence or a negative response sentence for the input sentence indicated by the input sentence data, and output means for outputting the response sentence generated by the generation means Yes.

  According to the first aspect of the present invention, input sentence data indicating an input sentence to be input from the user when interacting with the user is acquired by the acquiring means, and input sentence data acquired by the acquiring means is acquired by the generating means. A positive response sentence or a negative response sentence is generated with respect to the input sentence indicated by, and the output means outputs the response sentence generated by the generation means. The acquisition unit may acquire input sentence data by inputting an input sentence from a user such as a keyboard, for example. Also, for example, the input data may be received by receiving from an external device via a network. Or the input data may be acquired by recognizing a voice. Further, the output means may output by displaying a response sentence, may output information indicating the response sentence to an external device, or may be output by printing on paper or the like. The response sentence may be output by voice.

  Thus, according to the first aspect of the present invention, the input sentence data indicating the input sentence to be input from the user when the user interacts with the user is acquired, and the input sentence indicated by the acquired input sentence data Since a positive response sentence or a negative response sentence is generated and the generated response sentence is output, various dialogs with the user are possible.

In particular, when a positive response sentence is generated for an input sentence, the user thinks that the device is a friend of the user and a supporter, so it is easy to clarify troubles and to be positive with the user. Because it is possible to proceed with the talk, it is effective for listening etc. In addition, when a negative response sentence is generated for the input sentence, the user thinks that the device strictly trains the user, Users can be encouraged.

  The generating means according to claim 1 includes the analyzing means for analyzing the syntactic structure of the input sentence indicated by the input sentence data acquired by the acquiring means, and the syntactic structure as in the invention according to claim 2. According to the storage means storing the conversion rule information indicating the conversion rule for converting the input sentence into a positive response sentence or a negative response sentence according to the structure, and the conversion rule information stored in the storage means Among the conversion rules shown, the input sentence is converted into a positive response sentence or a negative response sentence using a conversion rule corresponding to the syntactic structure of the input sentence specified by the analysis by the analysis determining means. Conversion means.

  Further, in the invention according to claim 2, as in the invention according to claim 3, the analysis means performs a morphological analysis on the input sentence, and the verb included in the input sentence is a non-independent verb or a state verb. , And a non-state verb, the storage means stores a non-independent verb, a state verb, and a relaxation phrase corresponding to the non-state verb, respectively, and the conversion means includes the storage means. Non-independent verbs, state verbs, and relaxation expression words corresponding to non-state verbs stored in the table, the input sentence may be converted using a word corresponding to the analysis result of the verb by the analysis determining means. Good.

  In addition, when the syntactic structure of the input sentence corresponds to a plurality of conversion rules, the conversion means according to claim 2 or claim 3 uses the respective conversion rules. The input sentence is converted into the response sentence, and when each converted response sentence has a number of characters within a predetermined number of characters, the response sentence having the largest number of characters among the response sentences with the number of characters within the predetermined number of characters is output. If the number of characters in the converted response sentences is not within the predetermined number of characters, the response sentence with the smallest number of characters is selected as the output target by the output means. Also good.

  Further, the generation means according to any one of claims 2 to 4 further includes a derivation means for deriving a closeness with a user as in the invention according to claim 5, wherein the storage means A plurality of words, at least one of conjunctive particles, adverbs, and relaxation expressions are further stored according to the degree of intimacy, and the conversion means stores at least one of connection particles, adverbs, and relaxation expressions stored in the storage means. You may convert the said input sentence using the phrase according to the familiarity derived | led-out by the said deriving means among several types of words / phrases.

  According to a fifth aspect of the invention, as in the sixth aspect of the invention, the storage means further stores the number of times of interaction with the user for each user, and the derivation means is stored in the storage means. The degree of closeness may be derived as the number of conversations with the user performing the conversation increases.

  Further, in the invention according to any one of claims 2 to 6, as in the invention according to claim 7, the storage means affirms a predetermined negative word and the negative word. Corresponding phrase data associated with the expressed positive phrase is further stored, and the conversion means converts the input sentence into a positive response sentence, and the input sentence includes the negative phrase Or when the input sentence is converted into a negative response sentence and the positive sentence is included in the input sentence, it is associated with the corresponding phrase data stored in the storage means You may convert using words.

  On the other hand, in order to achieve the above object, the response sentence generation method of the invention according to claim 8 is obtained by acquiring input sentence data indicating an input sentence to be input from the user when interacting with the user. Analyzes the syntactic structure of the input sentence indicated by the input sentence data, and stores the input sentence as a positive response sentence or a negative response sentence according to the syntactic structure of the input sentence, which is stored in the storage unit in advance. Among the conversion rules indicated by the conversion rule information indicating the conversion rule for converting to the input rule, the input sentence is a positive response sentence using the conversion rule corresponding to the syntactic structure of the input sentence specified by the analysis, Or convert it to a negative response sentence.

  Therefore, according to the invention described in claim 8, when a dialogue with the user is performed, the syntactic structure of the input sentence to be input from the user is analyzed, and the conversion rule corresponding to the syntactic structure of the input sentence is used. Thus, since the input sentence can be converted into a positive response sentence or a negative response sentence, various dialogs with the user are possible.

  On the other hand, in order to achieve the above object, a response sentence generation processing program according to claim 9 includes an analysis step for analyzing a syntactic structure of an input sentence from a user indicated by input sentence data, and storage means Of the conversion rule indicated by the conversion rule information indicating the conversion rule for converting the input sentence into a positive response sentence or a negative response sentence in accordance with the syntactic structure of the input sentence. A conversion step of converting the input sentence into a positive response sentence or a negative response sentence using a conversion rule corresponding to the syntactic structure of the input sentence specified by the analysis in the analysis step. It is out.

  Therefore, according to the ninth aspect of the invention, when a dialogue with the user is performed, the syntactic structure of the input sentence that is input from the user is analyzed, and the conversion rule corresponding to the syntactic structure of the input sentence is used. Thus, since the input sentence can be converted into a positive response sentence or a negative response sentence, various dialogs with the user are possible.

  As described above, according to the present invention, when the user interacts with the user, the input sentence data indicating the input sentence to be input from the user is acquired, and the input sentence indicated by the acquired input sentence data is obtained. Thus, a positive response sentence or a negative response sentence is generated, and the generated response sentence is output, so that it is possible to have various effects with the user.

  Hereinafter, embodiments of the present invention will be described in detail with reference to the drawings. In the following, the present invention is realized by using a personal computer as an interactive processing device that listens by generating and outputting a positive response sentence when a negative input sentence is input from the user. This will be described as an example.

  FIG. 1 is a block diagram showing the main configuration of the electrical system of the dialogue processing apparatus 10 according to the present embodiment.

  As shown in the figure, the dialogue processing apparatus 10 includes a display 12 for displaying various types of information, an input device 14 such as a mouse or a keyboard for inputting various operation instructions from a user, a computer main body 16, and the like. It is equipped with.

  The computer main body 16 includes a CPU (Central Processing Unit) 20 that controls the operation of the entire apparatus, a RAM (Random Access Memory) 22 that is used as a work area when the CPU 20 executes various processing programs, and a response sentence generation process that will be described later. Various control programs including programs, ROM (Read Only Memory) 24 in which various parameters are stored in advance, a hard disk 26 used for storing various information, and various information such as operation screens and messages on the display 12 A display control unit 28 that controls display and an operation input detection unit 29 that detects an operation instruction input to the input device 14 are provided, and these units are electrically connected to each other via a system bus BUS. Has been.

  Therefore, the CPU 20 can access the RAM 22, ROM 24, and hard disk 26, obtain various input information via the input device 14, and display various information on the display 12.

  The input device 14 receives an input sentence that is input from the user when the user interacts with the user.

  The computer main body 16 generates a positive response sentence with respect to the input sentence input from the input device 14 and performs control to display the generated response sentence on the display 12.

  FIG. 2 is a functional block diagram showing a functional configuration of the dialogue processing apparatus 10 according to the present embodiment.

  As shown in the figure, the dialogue processing apparatus 10 includes a closeness deriving unit 30, an input sentence analyzing unit 32, an input sentence converting unit 34, an acceptance response generating unit 36, and a response selecting unit 38. ing.

  The dialogue processing apparatus 10 according to the present embodiment counts the number of times an input sentence is input from the user for each user as the number of dialogues. The hard disk 26 stores the number of dialogs for each user.

  The familiarity deriving unit 30 is connected to the hard disk 26, the input sentence converting unit 34, and the acceptance response generating unit 36.

  The closeness deriving unit 30 derives closeness with the user based on the number of conversations stored in the hard disk 26 when the user interacts with the user, and the derived closeness information is received by the input sentence conversion unit 34 and the acceptance unit. The response is output to the response generation unit 36. In this embodiment, the familiarity is derived at three levels of “high”, “medium”, and “low”. The familiarity deriving unit 30 according to the present embodiment derives the familiarity as “high” when the number of conversations is 50 times or more, and determines the familiarity as “medium” when the number of conversations is 20 times or more and less than 50 times. ”And when the number of dialogues is less than 20, the intimacy is derived as“ low ”. Note that the method of dividing the familiarity level and the range of the number of conversations at each level can be changed as appropriate.

  The input sentence analysis unit 32 is connected to the input device 14, the hard disk 26, and the input sentence conversion unit 34, and input sentence data indicating an input sentence is input from the input device 14.

  The input sentence analysis unit 32 analyzes the syntactic structure by performing morphological analysis on the input sentence indicated by the input sentence data input from the input device 14 to analyze the original form of the word and the part of speech (noun, Syntactic elements such as case particles, verbs, connective particles, adverbs, adjectives, auxiliary particles, auxiliary verbs, etc.) are specified, and syntax structure information indicating the syntactic structure of the specified input sentence is output to the input sentence conversion unit 34 Has been.

  In general morphological analysis, verbs are recognized as “independent verbs” and “non-independent verbs”, but in the morphological analysis according to the present embodiment, independent verbs are classified into state verbs and non-state verbs. . The “state verb” is a verb representing a certain state, and generally refers to a verb that does not take a progressive form (...). “Non-state verb” refers to a verb other than a state verb.

  For this reason, in the present embodiment, dictionary data of state verbs as shown in FIG. The input sentence analysis unit 32 classifies the independent verb as either “state verb” or “non-state verb” using the dictionary data of the state verb stored in the hard disk 26.

  Further, in the present embodiment, it is specified as a result of the morphological analysis whether or not the words of the identified connected particles and the associated particles have a negative meaning.

  For this reason, in the present embodiment, as shown in FIGS. 4A to 4C, the dictionary data of negative connective particles having a negative meaning, the dictionary data of negative particle having a negative meaning, and Dictionary data of negative responsive adverbs having negative meanings are stored in advance in the hard disk 26, respectively. In FIGS. 4A to 4C, the words of the negative connective particle, the negative counselor, and the negative responsive adverb are classified according to the intimacy, but in the present embodiment, the level is not necessarily classified. do not have to.

  The input sentence analyzing unit 32 stores each of the connected particle or the word of the particle associated with the morphological analysis, the negative connected particle dictionary data stored in the hard disk 26, the negative particle particle dictionary data, and the negative responsive adverb dictionary data. By comparing with a phrase, it is specified whether or not the phrase of the identified connected particle, coordinating particle, and responsive adverb is a negative connected particle, a negative associated particle, or a negative responsive adverb.

  Furthermore, the hard disk 26 stores in advance conversion rule information indicating a conversion rule for converting the input sentence into a positive response sentence in accordance with the syntactic structure of the input sentence.

  FIG. 5 shows an example of the data structure of the conversion rule information according to the present embodiment.

  As shown in the figure, the conversion rule information defines a syntactic conversion rule for converting a negative input sentence into a positive response sentence in accordance with the syntactic structure of the input sentence.

  In the conversion rule shown in the figure, items surrounded by [] indicate that they are essential items, and items surrounded by () indicate that they are selection items. That is, negative connective particles, negative auxiliary verbs, positive connective particles, and reverse connective particles enclosed in [] are required items, and negative responsive adverbs, relaxed expressions, and sentence ending relaxed expressions enclosed in () are optional items. It is.

  In the present embodiment, as shown in FIGS. 6A to 6E, the positive meaning used when converting the input sentence according to the degree of closeness (high, medium, low), respectively. Is stored in the hard disk 26. The dictionary data of the positive conjunctive particle having the above, the dictionary data of the paradoxical connecting particle having the paradoxical meaning, the dictionary data of the relaxation expression having the positive meaning, and the dictionary data of the sentence end relaxation expression having the positive meaning are stored in the hard disk 26. It is memorized beforehand. In the dialogue processing apparatus 10 according to the present embodiment, as shown in FIGS. 6C and 6D, relaxation expressions are stored separately for state verbs and verbs other than state verbs (non-state verbs). Yes.

  The hard disk 26 stores corresponding phrase data that associates a predetermined negative phrase with a positive phrase that positively expresses the negative phrase in advance.

  FIG. 7 shows an example of the data structure of the corresponding phrase data according to the present embodiment.

  As shown in the figure, the corresponding phrase data is associated with a predetermined negative phrase and a positive phrase that positively expresses the negative phrase.

  Further, the hard disk 26 stores in advance an acceptance response sentence database that defines acceptance response sentences according to the degree of familiarity.

  FIG. 8 shows an example of the data structure of the acceptance response sentence database according to the present embodiment.

  As shown in the figure, in the acceptance response sentence database, acceptance response sentences according to the degree of familiarity (high, medium, low) are defined.

  The input sentence conversion unit 34 (see FIG. 2) is connected to the input device 14, the hard disk 26, and the response selection unit 38, input sentence data is input from the input device 14, and familiarity information from the familiarity deriving unit 30. Is input, and syntax structure information is input from the input sentence analysis unit 32.

  The input sentence conversion unit 34 determines the affirmative stored in the hard disk 26 according to the conversion rule corresponding to the syntactic structure of the input sentence indicated in the syntax structure information among the conversion rules indicated by the conversion rule information stored in the hard disk 26. The input sentence is converted into an affirmative response sentence by using a phrase according to the familiarity indicated by the familiarity information among the respective words of the dictionary data of the connected particle, the paradoxical conjunction particle, the relaxation expression, and the sentence end relaxation expression. It is supposed to be. Note that the relaxation expression is converted using a morphological analysis performed by the input sentence analysis unit 32 and corresponding to the result of classifying the verb into a state verb or a non-state verb. In addition, when there are a plurality of words / phrases according to the familiarity, any word / phrase is arbitrarily selected.

  Further, when the input sentence indicated by the input sentence data includes a negative word / phrase, the input sentence conversion unit 34 performs conversion using the word / phrase associated with the corresponding word / phrase data stored in the hard disk 26. In addition, when there are a plurality of phrases associated with the corresponding phrase data, any phrase is arbitrarily selected.

  Then, the input sentence conversion unit 34 outputs response sentence data indicating the converted response sentence to the response selection unit 38.

  The acceptance response generation unit 36 is connected to the hard disk 26 and the response selection unit 38, and the familiarity information is input from the familiarity deriving unit 30.

  The acceptance response generation unit 36 selects an acceptance response sentence corresponding to the familiarity indicated by the familiarity information from the acceptance response sentence database stored in the hard disk 26, and returns an acceptance response sentence data indicating the selected acceptance response sentence. The data is output to the selection unit 38. When there are a plurality of acceptance response sentences corresponding to the familiarity, any of the acceptance response sentences is arbitrarily selected.

  The response selection unit 38 is connected to the display 12, and response text data is input from the input text conversion unit 34, and acceptance response text data is input from the acceptance response generation unit 36.

  When the response text data is input from the input text conversion unit 34, the response selection unit 38 displays the response text indicated by the response text data on the display 12. Further, when the input sentence conversion unit 34 cannot convert the input sentence and the response sentence data is not input from the input sentence conversion unit 34, the response selection unit 38 indicates by the acceptance response sentence data input from the acceptance response generation unit 36. Is displayed on the display 12.

  By the way, the processing by each component (intimacy deriving unit 30, input sentence analyzing unit 32, input sentence converting unit 34, acceptance response generating unit 36, and response selecting unit 38) of the dialogue processing apparatus 10 configured as described above. Can be realized by a software configuration using a computer by executing a program. However, the present invention is not limited to realization by a software configuration, and needless to say, it can also be realized by a hardware configuration or a combination of a hardware configuration and a software configuration.

  Hereinafter, a case will be described in which the interactive processing device 10 according to the present embodiment realizes processing by each of the above-described components by executing a response sentence generation processing program. In this case, the response sentence generation processing program is installed in the ROM 24 or the hard disk 26 in advance, provided in a state stored in a computer-readable recording medium, or distributed via wired or wireless communication means. The form etc. which are made can be applied.

  Next, with reference to FIG. 9, the operation of the dialogue processing apparatus 10 according to the present embodiment will be described. FIG. 9 is a flowchart showing the flow of processing of the response sentence generation processing program executed by the CPU 20. The program is executed by the CPU 20 when an input sentence is input to the input device 14 by the user. Further, in the following, a case will be described as an example in which “input milk” is input but “not milky at all” is input to the input device 14.

  In step 100 in the figure, the familiarity with the user is derived based on the number of dialogues stored in the hard disk 26 with the user who has input the input sentence.

  In the next step 102, the syntactic structure is analyzed by performing morphological analysis on the input sentence indicated by the input sentence data input from the input device 14, and the syntax element is specified.

  FIG. 10A schematically shows an example of the result of performing morphological analysis on the input sentence “I'm giving milk but it doesn't grow at all”.

  In the next step 104, it is determined whether or not the syntactic structure of the input sentence analyzed in step 102 corresponds to any of the syntactic structures on the input side of the conversion rule indicated by the conversion rule information. If the determination is affirmative, the process proceeds to step 104. If the determination is negative, the process proceeds to step 106.

  For example, the syntactic structure of the input sentence shown in FIG. 10A corresponds to the first and third lines on the input side of the conversion rule shown in FIG.

  In step 106, the input sentence is converted based on the corresponding conversion rule. When conversion is performed, the positive connection particles, paradoxical connection particles, relaxation expressions, and sentence ending relaxation expressions shown in FIGS. 6A to 6E are words corresponding to the familiarity derived in step 100 above. The conversion is performed using. In addition, the relaxation expression is converted by using morphological analysis in step 102 and corresponding to the result of classifying the verb into a state verb or a non-state verb.

  For example, when the familiarity is derived as “medium” and the input sentence shown in FIG. 10A is converted using the conversion rule in the first line as shown in FIG. 10B, FIG. Converted as shown in C).

  Also, for example, when the familiarity is derived as “high” and the input sentence shown in FIG. 10A is converted using the conversion rule in the third line as shown in FIG. 11 (B).

  In step 108, if the input sentence is converted in step 106, the input sentence is converted. If the input sentence is not converted, the input sentence is negative. If a positive determination is made, the process proceeds to step 110. If a negative determination is made, the process proceeds to step 112.

  In step 110, a negative word / phrase included in the input sentence is converted into a word / phrase associated with the corresponding word / phrase data stored in the hard disk 26.

  In step 112, it is determined whether or not the input sentence has been converted in at least one of step 106 and step 110. If the determination is affirmative, the process proceeds to step 114 while the determination is negative. Goes to step 116.

  In step 114, a positive response sentence generated by converting the input sentence is displayed on the display 12, and the response sentence generation processing program is terminated. Although there are a plurality of selection methods when the syntactic structure of the input sentence corresponds to a plurality of conversion rules, for example, the number of characters of the generated positive response sentence is counted as one of them. Then, when the number of characters of any one of the generated response sentences is, for example, 20 characters or less, a response sentence having 20 characters or less is displayed on the display 12, and when both characters are 20 characters or less, The response sentence with the larger number of characters is displayed on the display 12, and when the number of characters is both greater than 20, the response sentence with the smaller number of characters is displayed on the display 12, and when the number of characters is the same, either one is arbitrarily selected. The response sentence is selected and displayed on the display 12.

  That is, when the response sentence is short, the user may feel that the dialogue processing apparatus 10 does not sympathize with the content of the input sentence. When the response sentence is long, the user overlaps with the content of the input sentence. Because of this, the contents of the response sentence may be complicated. For this reason, in the dialogue processing apparatus 10 according to the present embodiment, the number of characters in the response sentence is set to about 20.

  On the other hand, in step 116, an acceptance response sentence corresponding to the familiarity indicated by the familiarity information is selected from the acceptance response sentence database stored in the hard disk 26.

  In the next step 118, the acceptance response sentence selected in the above step 116 is displayed on the display 12, and this response sentence generation processing program is terminated.

  That is, when the syntactic structure of the input sentence does not correspond to any conversion rule and the input sentence does not include negative words, the acceptance response sentence selected according to the familiarity is displayed on the display 12. Is done.

  Here, FIGS. 12A to 12C show an example of a result of a dialogue between the user and the dialogue processing apparatus 10. In the examples of FIGS. 12A to 12C, the familiarity is assumed to be “medium”.

  In the input sentence 1 from the user shown in FIG. 12A, the syntactic structure does not correspond to any conversion rule, and the input sentence does not include negative words. For this reason, an acceptance response sentence (response sentence 1) is displayed on the display 12.

  The input sentence 2 from the user has a syntactic structure corresponding to the conversion rule. For this reason, a response sentence (response sentence 2) generated according to the conversion rule is displayed on the display 12.

  Furthermore, the input sentence 3 from the user has a syntactic structure that does not correspond to any conversion rule, and the input sentence does not include negative words. For this reason, an acceptance response sentence (response sentence 3) is displayed on the display 12.

  On the other hand, the input sentence 4 from the user shown in FIG. 12 (B) does not have a syntactic structure corresponding to any conversion rule, but includes a negative phrase such as “It is a crybaby”. Therefore, the display 12 displays a response sentence (response sentence 4) obtained by converting a negative phrase with a phrase associated with the corresponding phrase data stored in the hard disk 26.

  On the other hand, the input sentence 5 from the user shown in FIG. 12C has a syntactic structure corresponding to the fifth line on the input side of the conversion rule. For this reason, a response sentence (response sentence 5) generated according to the conversion rule is displayed on the display 12. 13A to 13C show, as a result of the morphological analysis on the input sentence 5, the conversion rule on the fifth line corresponding to the syntactic structure of the input sentence 5 and the conversion rule. An example of the result of using and converting is shown.

  As described above, according to the present embodiment, the input device 14 acquires input sentence data indicating an input sentence that is input from the user when interacting with the user, and the computer main body 16 acquires the input device. 14 generates a positive response sentence with respect to the input sentence indicated by the input sentence data acquired by 14, and the response sentence generated by the computer main body 16 is displayed on the display 12, so that various interactions with the user can be performed. It becomes possible.

  In addition, according to the present embodiment, the expression of the response sentence to the user is changed according to the familiarity with the user, and when the familiarity increases, the expression that can be understood by the response sentence is used. Is easier to do.

  In the present embodiment, the case where the input sentence is converted into a positive response sentence has been described. However, the present invention is not limited to this. For example, the input sentence is converted into a negative response sentence. Also good. In this case, if the input side and the output side shown in FIG. 5 are reversed, a conversion rule for generating a negative response sentence can be obtained. Also, with respect to the corresponding phrase data shown in FIG. 7, if the input side and the output side are reversed, a positive phrase can be converted into a negative phrase.

  Further, in the present embodiment, the case where an input sentence is input by the keyboard as the input device 14 has been described. However, the present invention is not limited to this, and for example, the input device 14 is a voice that recognizes a voice. An input sentence may be input by applying a recognition apparatus and recognizing the spoken voice by the voice recognition apparatus.

  In the present embodiment, the case where the response sentence is displayed on the display 12 has been described. However, the present invention is not limited to this, and for example, the response sentence may be output by voice.

  Furthermore, in the present embodiment, the case where the familiarity is derived based on the number of conversations has been described. However, the present invention is not limited to this. For example, familiarity information indicating the familiarity is stored in advance for each user. The familiarity may be derived by reading the familiarity information. In addition, the intimacy can be determined by comparing the phrases shown in FIGS. 4A to 4C with the phrases shown in FIGS. It is good also as what calculates | requires and converts into the response sentence according to the said familiarity.

  In addition, the main configuration (see FIG. 1) of the electrical system of the dialog processing device 10 described in the present embodiment and the functional configuration (see FIG. 2) of the dialog processing device 10 are examples, and the present invention. Needless to say, changes can be made as appropriate without departing from the spirit of the present invention.

  Also, each dictionary data described in the present embodiment (see FIGS. 3, 4A to 4C, 6A to 6E), conversion rule information (see FIG. 5), and correspondence. The data structure of the phrase data (see FIG. 7) and the acceptance response sentence database (see FIG. 8) is also an example, and it goes without saying that it can be changed as appropriate without departing from the gist of the present invention.

  Furthermore, the processing flow of the response sentence generation processing program (see FIG. 9) described in the present embodiment is also an example, and it goes without saying that it can be changed as appropriate without departing from the gist of the present invention.

It is a block diagram which shows the principal part structure of the electric system of the dialogue processing apparatus which concerns on embodiment. It is a block diagram which shows the functional structure of the dialogue processing apparatus which concerns on embodiment. It is a figure which shows an example of the data structure of the dictionary data of the state verb which concerns on embodiment (A) is a figure which shows an example of the data structure of the dictionary data of a negative connection particle, (B) is a figure which shows an example of the data structure of the dictionary data of a negative charge particle, (C) is a figure of a negative responsive adverb. It is a figure which shows an example of the data structure of dictionary data. It is a figure which shows an example of the data structure of the conversion rule information which concerns on embodiment. (A) is a figure which shows an example of the data structure of the dictionary data of an affirmative connection particle, (B) is a figure which shows an example of the data structure of the dictionary data of a paradoxical connection particle, (C) is a figure other than a state verb It is a figure which shows an example of the data structure of the dictionary data of the relaxation expression with respect to a verb, (D) is a figure which shows an example of the data structure of the dictionary data of the relaxation expression with respect to a state verb, (E) is a dictionary of sentence ending relaxation expression It is a figure which shows an example of the data structure of data. It is a figure which shows an example of the data structure of the corresponding phrase data based on Embodiment. It is a figure which shows an example of the data structure of the acceptance response sentence database which concerns on embodiment. It is a flowchart which shows the flow of a process of the response sentence production | generation process program which concerns on embodiment. (A)-(C) are the figures which showed typically an example of the process which converts an input sentence into a response sentence. (A) (B) is the figure which showed typically the other example of the process which converts an input sentence into a response sentence. (A)-(C) are the figures which showed the input sentence from a user, and the response sentence with respect to the said input sentence. (A)-(C) are the figures which showed typically an example of the process which converts an input sentence into a response sentence.

Explanation of symbols

10 Dialogue processing device 12 Display (output means)
14 Input device (acquisition means)
16 Computer body (generation means)
26 Hard disk (storage means)
30 Intimacy deriving section (derivation means)
32 Input sentence analysis part (analysis means)
34 Input sentence conversion part (conversion means)

Claims (9)

An acquisition means for acquiring input sentence data indicating an input sentence to be input from the user when interacting with the user;
Generating means for generating a positive response sentence or a negative response sentence for the input sentence indicated by the input sentence data acquired by the acquisition means;
Output means for outputting a response sentence generated by the generating means;
A dialogue processing apparatus comprising: The generating means includes
Analyzing means for analyzing the syntactic structure of the input sentence indicated by the input sentence data acquired by the acquiring means;
Storage means for storing in advance conversion rule information indicating a conversion rule for converting the input sentence into a positive response sentence or a negative response sentence according to the syntactic structure;
Among the conversion rules indicated by the conversion rule information stored in the storage means, the input sentence is positively answered using a conversion rule corresponding to the syntactic structure of the input sentence specified by the analysis by the analysis determination means. A conversion means for converting into a sentence or a negative response sentence;
The dialogue processing apparatus according to claim 1, further comprising: The analysis means performs morphological analysis on the input sentence to analyze whether the verb included in the input sentence is a non-independent verb, a state verb, or a non-state verb,
The storage means stores non-independent verbs, state verbs, and relaxation expressions corresponding to non-state verbs, respectively.
The converting means uses a phrase corresponding to the analysis result of the verb by the analysis determining means, out of words of relaxation expressions corresponding to non-independent verbs, state verbs, and non-state verbs stored in the storage means. The dialogue processing apparatus according to claim 2, wherein the input sentence is converted. When the syntactic structure of the input sentence corresponds to a plurality of conversion rules, the conversion means converts the input sentence to the response sentence using each conversion rule, and the number of characters in each converted response sentence If there are those within the predetermined number of characters, the response sentence having the largest number of characters is selected as an output target by the output means among the response sentences within the predetermined number of characters, and the number of characters in the converted response sentence is the predetermined number of characters. 4. The dialogue processing apparatus according to claim 2, wherein, when there is no response, the response sentence having the smallest number of characters among the response sentences is selected as an output target by the output unit. 5. The generating means includes
A derivation unit for deriving the intimacy with the user;
The storage means further stores a plurality of connective particles, adverbs, and relaxation expressions in accordance with the degree of intimacy.
The conversion means uses the input sentence by using a phrase according to the familiarity derived by the derivation means from among a plurality of words of at least one of connected particles, adverbs, and relaxation expressions stored in the storage means. The dialogue processing apparatus according to any one of claims 2 to 4, wherein: The storage means further stores the number of interactions with the user for each user,
The dialogue processing apparatus according to claim 5, wherein the derivation unit derives the intimacy as the number of dialogues with the user who performs dialogue stored in the storage unit increases. The storage means further stores corresponding phrase data in which a predetermined negative phrase is associated with a positive phrase that positively expresses the negative phrase,
The conversion means is a case where the input sentence is converted into a positive response sentence and the negative sentence is included in the input sentence, or a case where the input sentence is converted into a negative response sentence. 7. When the input sentence includes the positive word / phrase, the input sentence is converted using the word / phrase associated with the corresponding word / phrase data stored in the storage unit. 7. Interactive processing device. Get input sentence data indicating the input sentence that will be input from the user when interacting with the user,
Analyzing the syntactic structure of the input sentence indicated by the acquired input sentence data,
Conversion indicated by conversion rule information indicating a conversion rule for converting the input sentence into a positive response sentence or a negative response sentence in accordance with the syntactic structure of the input sentence stored in advance in a storage unit A response sentence generation method for converting an input sentence into a positive response sentence or a negative response sentence using a conversion rule corresponding to the syntactic structure of the input sentence specified by the analysis. An analysis step for analyzing the syntactic structure of the input sentence from the user indicated by the input sentence data;
Conversion indicated by conversion rule information indicating a conversion rule for converting the input sentence into a positive response sentence or a negative response sentence in accordance with the syntactic structure of the input sentence stored in advance in a storage unit A conversion step of converting the input sentence into a positive response sentence or a negative response sentence using a conversion rule corresponding to the syntactic structure of the input sentence specified by the analysis in the analysis step among the rules;
Is a response sentence generation processing program for causing a computer to execute.

Images