JP5740272B2 - Dialog system, dialog method and program - Google Patents

Description

  The present invention relates to a dialog system, a dialog method, and a program in a social network, and more particularly, to a dialog system, a dialog method, and a program that perform communication suitable for a social network by performing a dialog action that does not have an explicit target.

  Currently, the social network community has been developed. In particular, Twitter (registered trademark) has grown rapidly with the number of users in Japan approaching 10 million (according to Nielsen Online survey in April 2010). For this reason, in a community on a social network such as Twitter, there is an increasing expectation for a dialogue system (bot) that performs a dialogue with a user using a natural language to perform advertising activities and user support.

  Therefore, recently, as shown in Non-Patent Document 1, an interactive system corresponding to Twitter has already been commercialized in combination with a product search system (for example, see Non-Patent Document 1). This system presents products by matching and searching for keywords in response to user comments.

  Further, in the system described in Non-Patent Document 1, since simple keyword matching is used, only one-to-one response can be performed. However, in the system described in Non-Patent Document 2, the knowledge of the system, By structuring user utterances, it is possible to interactively present information while the system listens to insufficient user input (see, for example, Non-Patent Document 2). Further, in the system described in Non-Patent Document 3, the system described in Non-Patent Document 2 is applied to a voice dialog, and a dialog system that enables interaction including a response to a voice recognition error is constructed (for example, Non-Patent Document 3).

  These systems of Non-Patent Documents 1 to 3 are dialogue control technologies for reproducing dialogues between humans in the real world, and are systems that always respond to inputs. However, humans do not always respond to all inputs on social networks, but have unique communications that do not exist in the real world.

  Here, the structure of communication between humans constructed on Twitter is roughly classified into the three shown in FIG. That is, (a) shows a dialogue between two parties. In Twitter, a third party can view the conversation between these two parties. Also, as shown in (b), Twitter has a function to perform a dialogue that calls to the entire audience like a meeting or a class because an unspecified number of users can participate in the conversation. On Twitter, communication that simulates human conversation in the real world is performed as in (a) and (b), but Twitter does not exist in the real world. Communication is carried out on social networks.

  Each user on Twitter is building a network by following other users he likes. FIG. 6 shows a user's follow list. Twitter presents the user's remarks included in the follow list in chronological order as shown in FIG. In Twitter, this screen is called a timeline (TL). Basically, on Twitter, it is not possible to control who follows him / her, so as shown in (c), a remark that is similar to a monolog that does not expect a response is often made without specifying a target. When the user finds a favorite remark on his / her TL, he / she speaks alone without specifying a target regarding the remark.

  As described above, the biggest feature of the social network built on Twitter is that it does not obligate other users to respond to their utterances, and its implicit understanding lowers the threshold of utterances. It is one of the factors to acquire active users.

Corecamo http://korekamo.net/ Yoji Kiyota, Ikuo Kurohashi, Toko Kido: Automatic question answering based on large-scale text knowledge base-Dialog navigation, Natural language processing, Vol. 10, no. 4, pp. 145-175, July, 2003. Teruhisa Tsuji, Tatsuya Kawahara, Tetsuro Shoji, Nobuhiko Mino: A voice information guidance system with question answering and information recommendation functions. Transactions of Information Processing Society of Japan, Vol. 48, no. 12, pp. 3602--3611, 2007.

  However, the systems described in Non-Patent Documents 1 to 3 described above are dialogue control technologies that enable dialogues belonging to (a) and (b) that are close to dialogues between humans in the real world. The system always responds. Therefore, while it is possible to take communication belonging to (a) and (b), in these systems, it is not possible to follow a user to be noticed on Twitter and extract a notice to be noticed from TL. Therefore, there was a problem that the communication shown in (c) could not be taken.

  Therefore, the present invention has been made in view of the above-described problems, and provides a dialog system, a dialog method, and a program for performing communication suitable for a social network by performing a dialog action without an explicit target. For the purpose.

  The present invention proposes the following items in order to solve the above-described problems. In addition, in order to make an understanding easy, although the code | symbol corresponding to embodiment of this invention is attached | subjected and demonstrated, it is not limited to this.

(1) The present invention relates to a dialogue system to interact with the social network, the interactive information collection device for collecting interactive information on the social network (e.g., corresponding to interactive information collection device 100 of FIG. 1) and A response determination device (for example, corresponding to the response determination device 200 in FIG. 1) that determines whether or not to respond to the collected dialogue information, and the social network from the dialogue information for which the response determination has been made A response filter (for example, corresponding to the response filter 300 in FIG. 1) that extracts only dialogue information having a strong influence on the content , understanding of the dialogue information, creation of an utterance sentence, and speech generation apparatus for controlling the response by utterance (e.g., corresponding to the speech generation apparatus 400 of FIG. 1) and a dialogue system, comprising the Hisage It is.

According to the present invention, the dialogue information collection device collects dialogue information on a social network. The response determination device determines whether to respond to the collected dialogue information. The response filter extracts only dialogue information having a strong influence on the social network from the dialogue information for which the response is determined. Based on the extracted dialogue information , the utterance sentence generation device understands the dialogue information, creates the utterance sentence, and controls a response by the utterance sentence . Therefore, the communication action suitable for the social network can be performed by performing the dialog act having no explicit target based on the dialog information on the social network, for example, the remarks to be noted from the timeline .

(2) According to the present invention, in the dialog system of (1), the dialog information collection device determines an extracted person determining means (for example, the extraction of FIG. and equivalent) to the person determining unit 110, the search keyword determining means for determining a search keyword in accordance with a determination result of the extraction person determination means (e.g., a corresponding) to the search keyword determining section 150 of FIG. 2, the search keyword the determined A dialogue system characterized by comprising dialogue information collection means (for example, equivalent to the dialogue information collection unit 160 in FIG. 2) for searching and collecting dialogue information on a social network by using it.

According to the present invention, the extracted person determining means determines which range of groups the person having an influence is to be extracted. The search keyword determination means determines the search keyword according to the determination result of the extracted person determination means. The dialogue information collecting means searches and collects dialogue information on the social network using the determined search keyword. Therefore, it is possible to efficiently collect the influence level of the dialogue information.

(3) According to the present invention, in the dialog system of (1), the dialog information collection device determines which range of groups the dialog information that influences the dialog information extraction means (for example, FIG. 2). and equivalent) to extract interactive information determination unit 120 of the search keyword determining means for determining a search keyword in accordance with a determination result of the extraction interactive information determining means (e.g., corresponding to the search keyword determining section 150 of FIG. 2), the determined Proposing a dialogue system characterized by comprising dialogue information collecting means (for example, equivalent to the dialogue information collecting unit 160 in FIG. 2) for searching and collecting dialogue information on a social network using the searched keyword Yes.

According to the present invention, the extracted dialogue information determining means determines which range of groups the dialogue information having an influence is to be extracted. The search keyword determining means determines the search keyword according to the determination result of the extracted dialogue information determining means. The dialogue information collecting means searches and collects dialogue information on the social network using the determined search keyword. Therefore, it is possible to efficiently collect the influence level of the dialogue information.

(4) The present invention is, for interactive system (1), wherein the dialog information collecting device, the reply destination information referring unit for referring to the obtained reply destination information in the social network (e.g., destination information see FIG. 2 A dialogue system characterized in that it is provided with a portion equivalent to the section 130).

  According to the present invention, the reply destination information reference means refers to the acquired reply destination information in the social network. Therefore, it is possible to estimate which dialogue information is the response, and to make the dialogue information having a high response number with a certain number or more of the responses.

  (5) The present invention relates to the user name reference means (for example, equivalent to the user name reference unit 140 in FIG. 2) in which the dialog information collection device refers to the user name included in the speech in the dialog system of (1). We have proposed a dialogue system characterized by having

  According to the present invention, the user name reference means refers to the user name included in the speech. Therefore, it is possible to estimate which dialogue information is the response, and to make the dialogue information having a high response number with a certain number or more of the responses.

(6) The present invention is, for interactive system (1), the response determination apparatus, aggregates the impact on the social network of each user, the user influence degree calculating means for calculating a user impact (e.g., FIG. 3 and a response determination unit (for example, corresponding to the response determination unit 230 in FIG. 3) that performs a response determination based on whether or not the calculated user influence level exceeds a predetermined threshold. ) Is proposed.

  According to the present invention, the user influence degree calculating means adds up the influence degree of each user on the social network and calculates the user influence degree. The response determination unit performs a response determination based on whether or not the calculated user influence level exceeds a predetermined threshold value. Therefore, it is possible to find and respond to a user who should pay attention on the social network.

  (7) According to the present invention, in the dialog system of (1), the response determination device predicts the degree of influence using a keyword that is periodically updated (for example, the degree of influence prediction in FIG. 3). And a response determination unit (for example, corresponding to the response determination unit 230 in FIG. 3) for performing a response determination based on whether or not the calculated user influence level exceeds a predetermined threshold value. We have proposed a dialogue system characterized by this.

  According to the present invention, the influence degree predicting means predicts the influence degree using important words that are periodically updated. The response determination unit performs a response determination based on whether or not the calculated user influence level exceeds a predetermined threshold value. Thereby, it is possible to respond to TL in real time. As a method for predicting the influence degree of important words, for example, the importance degree of a phrase is converted into a numerical value, and a value obtained by multiplying this by an arbitrary coefficient is calculated. The largest one is extracted, thereby predicting the influence of important words. In addition, since the combination of dialogue information for which the degree of influence of an important word is calculated and the important word that contains it is stored in the database, how much influence is given when there is a predetermined important word based on this information. Prediction may be performed by learning control whether there is a degree.

  (8) The present invention proposes an interactive system characterized in that the response determination device responds to a user who frequently makes a one-to-one response to the interactive system of (1). Yes.

  According to the present invention, a response is made to a user who has made a one-to-one response frequently. Thereby, the affinity with respect to a user can be improved.

(9) The present invention proposes an interactive system characterized in that the social network is a Twitter for the interactive system of (1) to (8).

According to the present invention, the social network is Twitter.

(10) The present invention proposes a dialogue system in which the dialogue information collection device collects dialogue information from a timeline for the dialogue system of (9).

According to the present invention, the dialogue information collection device collects dialogue information from the timeline. Thereby, it is possible to extract a remark that should be noted from the timeline.

(11) The present invention is a dialog method in a dialog system that includes a dialog information collection device, a response determination device, a response filter, and an utterance sentence generation device, and performs a dialog on a social network, and the dialog information collection device includes: A first step of collecting dialogue information on the social network (for example, corresponding to step S101 in FIG. 4), and whether or not the response determination device responds to the collected dialogue information A second step (e.g., corresponding to step S102 in FIG. 4), and a third filter in which the response filter extracts only dialog information having a strong influence on the social network from the dialog information for which the response determination has been made. Step (equivalent to step S103 in FIG. 4) and the utterance sentence generation device, based on the extracted dialogue information, Proposing a dialogue method characterized by comprising a fourth step (for example, corresponding to step S104 in FIG. 4) for understanding dialogue information, creating a spoken sentence, and controlling a response by the spoken sentence. ing.

According to the present invention, the dialogue information collection device collects the dialogue information on the social network, and determines whether or not to respond to the dialogue information collected by the response determination device. Then, only the dialogue information that has a strong influence on the social network is extracted from the dialogue information for which the response filter has made a response determination, and based on the dialogue information taken out by the utterance generation device, Creates sentences and controls responses by spoken sentences. Therefore, the communication action suitable for the social network can be performed by performing the dialog act having no explicit target based on the dialog information on the social network, for example, the remarks to be noted from the timeline.

(12) The present invention is interactive information collecting apparatus, response determination apparatus, response filters, and includes a speech generation apparatus, a program for executing an interactive methodologies to computers in dialogue system for performing a dialogue on social networks, the A dialogue information collection device collects dialogue information on the social network (for example, corresponding to step S101 in FIG. 4), and the response determination device responds to the collected dialogue information. Dialogue information that has a strong influence on the social network from the dialogue information for which the response filter has been judged and the second step for judging whether or not the response filter has been judged. the third step of taking out only (e.g., corresponding to step S103 in FIG. 4) and, said speech generation apparatus Based on the retrieved interactive information, understanding of the pair talk information, the creation of utterance, and a fourth step of controlling the response by emitting Hanashibun (e.g., corresponding to step S104 in FIG. 4), the A program to be executed by a computer is proposed.

According to the present invention, the dialogue information collection device collects the dialogue information on the social network, and determines whether or not to respond to the dialogue information collected by the response determination device . Then, only the dialogue information that has a strong influence on the social network is extracted from the dialogue information for which the response filter has made a response determination, and based on the dialogue information taken out by the utterance generation device, Creates sentences and controls responses by spoken sentences . Therefore, the communication action suitable for the social network can be performed by performing the dialog act having no explicit target based on the dialog information on the social network, for example, the remarks to be noted from the timeline .

  ADVANTAGE OF THE INVENTION According to this invention, there exists an effect which attracts the attention of another user because a dialog system always makes the speech regarding dialog information with a high influence degree. Therefore, this dialogue system is particularly effective for advertising purposes.

  In addition, by having the function to construct the same relationship between the user and the user built on the social network between the dialog system and the user, the dialog system can have a sociality close to humans in the social network. There is an effect that becomes possible. In the present invention, in order to build a relationship adapted to a social network, a social rule of “not obligating another user to respond to his / her speech”, which is an implicit understanding within the social network, interacts. It may also apply to the system. In other words, the existing dialogue system has modeled the relationship of dialogue in the real world, so it has become a system that allows users to expect a high level of responsiveness, such as “being able to respond to their comments”. However, when using natural language input, it is impossible to completely model various user inputs in advance. Therefore, in the conventional technique, the input pattern deviates from the input pattern previously described by the system as a rule. If the user performs, the system gives an answer that does not meet the user's expectations, resulting in a significant decrease in the user's evaluation. However, according to the present invention, the social rules described above are also applied to the dialogue system, so that it is possible to construct a system that suppresses user dissatisfaction in advance while attracting the user's attention.

It is a figure which shows the structure which concerns on the dialogue system of this invention. It is a figure which shows the structure of the dialogue information collection apparatus of this invention. It is a figure which shows the structure of the response determination apparatus of this invention. It is a figure which shows the process which concerns on the dialogue system of this invention. It is a figure which illustrates the interactive form in the social network which concerns on a prior art example. It is a figure which illustrates the user's follow list in the conventional Twitter. It is the figure which presented the user's utterance contained in the follow list in the conventional Twitter in time series order.

Hereinafter, embodiments of the present invention will be described in detail with reference to the drawings.
Note that the constituent elements in the present embodiment can be appropriately replaced with existing constituent elements and the like, and various variations including combinations with other existing constituent elements are possible. Therefore, the description of the present embodiment does not limit the contents of the invention described in the claims.

  Hereinafter, an embodiment according to the interactive system of the present invention will be described with reference to FIGS. 1 to 4. In the present embodiment, Twitter will be described as an example of a social network.

<Configuration of dialogue system>
As shown in FIG. 1, the dialogue system according to the present embodiment mainly includes a dialogue information collection device 100, a response determination device 200, a response filter 300, and an utterance sentence generation device 400.

  The dialogue information collecting apparatus 100 collects dialogue information (Twitter) on a social network (Twitter). The response determination device 200 determines whether to respond to the collected dialogue information (twits). The response filter 300 extracts only specific dialogue information from the dialogue information (twits) for which the response is determined. The utterance sentence generation device 400 performs language understanding, dialogue control, utterance sentence creation, and the like.

  That is, a set of dialogue information (twits) is collected from the social network (Twitter), and influential dialogue information (twits) is output to the response filter 300. On the other hand, the response filter 300 also inputs dialogue information (twits) from a client who has acquired the timeline (TL) of the social network (Twitter) from the social network (Twitter), and only influential dialogue information (twitter). Is output to the utterance sentence generator 400. Note that the client that has acquired the timeline (TL) acquires the timeline (TL) of the interactive system (BOT) in real time using the Twitter API.

  On the other hand, the dialogue information collection device 100 outputs influential user information to the response determination device 200, and the response determination device 200 outputs a responding user to the output client based on the input influential user information. To do. Furthermore, the output client inputs an utterance sentence from the utterance sentence generation device 400 using the Twitter API, transmits an utterance (twit) to the system, and outputs a follow request.

<Configuration of conversation information collection device>
As shown in FIG. 2, the dialogue information collection apparatus 100 according to the present embodiment includes an extracted person determination unit 110, an extracted dialogue information determination unit 120, a transmission destination information reference unit 130, a user name reference unit 140, and a search. The keyword determination unit 150 and the dialogue information collection unit 160 are configured.

  The extracted person determination unit 110 determines which range of groups the person having an influence is to be extracted. The extracted dialogue information determination unit 120 determines which range of groups the dialogue information (twits) having an influence is to be extracted.

  The transmission destination information reference unit 130 refers to the reply destination information in the acquired social network (Twitter). The user name reference unit 140 refers to the user name included in the speech.

  The search keyword determination unit 150 determines a search keyword according to the determination result of the extracted person determination unit 110 and / or the extracted dialogue information determination unit 120. The dialogue information collection unit 160 searches and collects dialogue information (twitter) on the social network (Twitter) using the keyword determined by the search keyword determination unit 150.

  That is, the dialogue information collecting apparatus 100 determines a search keyword according to which range of groups or persons and / or utterances that influence the group, and searches and collects utterances on the social network (Twitter) using the keywords. To do. Also, by referring to the reply destination information in_reply_to_status_id in the acquired dialogue information (Tweet) or referring to the user name included in the utterance, it is estimated which tweet is the response, and the number of responses above a certain number A large amount of dialogue information (twits) is output to the response filter 300 as dialogue information (twits) having a high degree of influence. Therefore, the extracted person determination unit 110, the transmission destination information reference unit 130, and the user name reference unit 140 may include all of them, or at least one of them.

<Configuration of response determination device>
As illustrated in FIG. 3, the response determination apparatus 200 according to the present embodiment includes a user influence level calculation unit 210, an influence level prediction unit 220, and a response determination unit 230.

  The user influence degree calculation unit 210 calculates the user influence degree by summing up the influence degree of each user on the social network (Twitter). The influence degree prediction unit 220 predicts the influence degree by using important words that are periodically updated. The response determination unit 230 performs a response determination based on whether or not the calculated user influence level exceeds a predetermined threshold. In addition, as the method of predicting the degree of influence of an important word in the influence degree prediction unit 220, for example, when the degree of importance of a phrase is digitized and a value obtained by multiplying this by an arbitrary coefficient is calculated, there are a plurality of phrases. Extracts the one having the maximum calculated value, and thereby predicts the influence degree of the important word. In addition, since the combination of dialogue information for which the degree of influence of an important word is calculated and the important word that contains it is stored in the database, how much influence is given when there is a predetermined important word based on this information. Prediction may be performed by learning control whether there is a degree.

  That is, the influence degree of each user's dialogue information (Tweet) is totaled to calculate the user influence degree. Alternatively, the degree of influence is predicted using important words that are regularly updated. In addition, by following a user whose user influence degree is equal to or greater than a threshold value, a user to be noticed on the social network (Twitter) is discovered and followed.

<Interaction system processing>
The process of the dialogue system according to this embodiment will be described with reference to FIG.

  First, the dialogue information collection device 100 collects dialogue information (twitter) on the social network (Twitter) (step S101), and whether or not the response determination device 200 responds to the collected dialogue information (twitter). Is determined (step S102).

  Then, the response filter 300 extracts only specific dialogue information from the dialogue information for which the response is determined (step S103), and the utterance sentence generation device 400 performs language understanding, dialogue control, utterance sentence creation, and the like. This is performed (step S104). In addition, in order to improve the affinity with respect to the user, the user who frequently made a one-to-one response to the dialogue system is followed.

  As described above, according to the present embodiment, communication suitable for a social network can be performed by performing an interactive action without an explicit target. In addition, it is determined which group or person and / or dialogue information (twitter) that has an influence on the group, and a search keyword is determined according to the determination result, and the keyword is used on the social network (Twitter). Since the dialogue information (twits) is retrieved and collected, the influence degree of the dialogue information (twits) can be efficiently collected. Further, the influence level of each user on the social network is aggregated, the user influence degree is calculated, and the response determination is performed based on whether or not the calculated user influence degree exceeds a predetermined threshold. ) It is possible to find and respond to a user to be noted above.

<Example>
As a specific example, a case will be described in which a Bot that promotes sales of a movie video distribution service is implemented using an API published by Twitter.

  In this case, first, a word related to a movie to be sold is extracted from the promotional text of the movie and used as a keyword. Use Twitter's Search API to collect tweets with these keywords. In the API published by Twitter, the amount of data that can be obtained differs depending on the form of the contract. Therefore, the search API is used periodically to collect data without omission as much as possible within the scope of the contract. Collect whispers including.

  The dialogue information collecting apparatus 100 calculates how many tweets are retweeted by the method described above. In this embodiment, if the retweet is performed a certain number of times or more, the degree of influence is high, and a value (STATUS_ID) that uniquely identifies the tweet is transmitted to the response filter 300. However, there are several methods for calculating the degree of influence in detail, and an appropriate selection is made according to the application, such as normalization based on the number of followers of the user who has spoken, or summing up the past degree of influence of the user who has spoken.

  The response filter 300 selects a tweet that matches the STATUS_ID output from the conversation information collection device 100 from the Bot TL obtained by using the Twitter API, transmits the tweet to the utterance sentence generation device 400, and the others on the TL Throw away your tweets. However, since Bot that does not respond to an explicit request from the user has a problem for advertising purposes, even if T_weet in_reply_to_user_id indicating Bot is detected from the TL, it is similarly sent to the dialog control unit. 5 (a) and (b) are also realized.

  Next, as an example of the utterance sentence generation device 400, the same technique as that described in Non-Patent Document 1 will be described. First, prepare a keyword that matches the movie you want to advertise in advance and a template that matches the keyword. By extracting a keyword having the longest match with a tweet input via a response filter and selecting a template corresponding to the keyword, it is possible to advertise advertising information to the user. At this time, if all utterances on the TL are responded, the user gets bored and the bot's utterance is ignored by the user. However, it is possible to effectively expose the promotional text by speaking only when a topic that many users pay attention to.

The following is an example of a speech template.
The performers are also appearing in movies like [Movie Title] [Promotional URL], the director name is [I also make [Movie Title] [Promotional URL]], and other keywords For example, “Some [movie titles are also there [advertisement URL]”.

  Note that the interactive system of the present invention can be realized by recording the processing of the interactive system on a computer-readable recording medium, reading the program recorded on the recording medium into the interactive system, and executing the program. The computer system here includes an OS and hardware such as peripheral devices.

  Further, the “computer system” includes a homepage providing environment (or display environment) if a WWW (World Wide Web) system is used. The program may be transmitted from a computer system storing the program in a storage device or the like to another computer system via a transmission medium or by a transmission wave in the transmission medium. Here, the “transmission medium” for transmitting the program refers to a medium having a function of transmitting information, such as a network (communication network) such as the Internet or a communication line (communication line) such as a telephone line.

  The program may be for realizing a part of the functions described above. Furthermore, what can implement | achieve the function mentioned above in combination with the program already recorded on the computer system, what is called a difference file (difference program) may be sufficient.

  The embodiments of the present invention have been described in detail with reference to the drawings. However, the specific configuration is not limited to the embodiments, and includes designs and the like that do not depart from the gist of the present invention.

DESCRIPTION OF SYMBOLS 100; Dialog information collection apparatus 110: Extraction person determination part 120; Extraction dialog information determination part 130; Transmission destination information reference part 140; User name reference part 150; Search keyword determination part 160; Dialog information collection part 200; User influence degree calculation unit 220; influence degree prediction unit 230; response determination unit 300; response filter 400;

Claims (12)

A dialogue system for conducting a dialogue on a social network,
And interactive information collecting apparatus for collecting interactive information on the social network,
A response determination device for determining whether to respond to the collected dialogue information;
A response filter that extracts only dialogue information having a strong influence on the social network from the dialogue information for which the response determination is made;
Based on the extracted dialogue information, an utterance sentence generation device for understanding the dialogue information, creating an utterance sentence, and controlling a response by the utterance sentence;
An interactive system characterized by comprising: The dialogue information collection device
Extraction person determination means for determining which range of groups the influencer is extracted from,
A search keyword determining means for determining a search keyword in accordance with a determination result of the extraction person determination means,
Dialog information collecting means for searching and collecting dialog information on the social network using the determined search keyword;
The interactive system according to claim 1, further comprising: The dialogue information collection device
Extraction dialog information decision means for deciding which range of group influence dialogue information is to be extracted;
A search keyword determining means for determining a search keyword in accordance with a determination result of the extraction interactive information determining means,
Dialog information collecting means for searching and collecting dialog information on the social network using the determined search keyword;
The interactive system according to claim 1, further comprising: The dialogue information collection device
Dialogue system according to claim 1, characterized in that with the reply destination information referring unit for referring to the obtained reply destination information in the social network. The dialogue information collection device
2. The dialogue system according to claim 1, further comprising user name reference means for referring to a user name included in the speech. The response determination device is
It aggregates impact on the social network of each user, and the user influence degree calculating means for calculating a user impact,
A response determination means for performing a response determination based on whether or not the calculated user influence level exceeds a predetermined threshold;
The interactive system according to claim 1, further comprising: The response determination device is
An impact prediction means for predicting the impact using important words that are regularly updated,
A response determination means for performing a response determination based on whether or not the calculated user influence level exceeds a predetermined threshold;
The interactive system according to claim 1, further comprising:   The interactive system according to claim 1, wherein the response determination device responds to a user who frequently makes a one-to-one response. 9. The dialogue system according to claim 1, wherein the social network is Twitter. The dialogue system according to claim 9, wherein the dialogue information collection device collects dialogue information from a timeline. A dialogue method in a dialogue system comprising a dialogue information collection device, a response determination device, a response filter, and an utterance sentence generation device, and performs dialogue on a social network,
A first step in which the dialogue information collecting device collects dialogue information on the social network;
A second step of determining whether or not the response determination device responds to the collected dialogue information;
A third step in which the response filter extracts only dialogue information having a strong influence on the social network from the dialogue information for which the response determination has been made;
When,
A fourth step in which the utterance sentence generation device performs understanding of the dialogue information, creation of the utterance sentence, and control of a response by the utterance sentence based on the extracted dialogue information;
A dialogue method characterized by comprising: A program comprising a dialog information collection device, a response determination device, a response filter, and an utterance sentence generation device, causing a computer to execute a dialog method in a dialog system for performing a dialog on a social network,
A first step in which the dialogue information collecting device collects dialogue information on the social network;
A second step of determining whether or not the response determination device responds to the collected dialogue information;
A third step in which the response filter extracts only dialogue information having a strong influence on the social network from the dialogue information for which the response determination has been made;
A fourth step in which the utterance sentence generation device performs understanding of the dialogue information, creation of the utterance sentence, and control of a response by the utterance sentence based on the extracted dialogue information;
A program that causes a computer to execute.