JP7080609B2 - Information processing equipment, information processing methods, and information processing programs - Google Patents

Description

The present invention relates to an information processing apparatus, an information processing method, and an information processing program.

A system for generating a response (hereinafter referred to as response information) corresponding to a user's utterance is known. For example, a dialogue agent system that receives a message from a user terminal and outputs a message corresponding to the received message is known.

Japanese Unexamined Patent Publication No. 2016-45584

However, with the conventional technique, it is not easy for the user to obtain response information, such as the need to purchase a dialogue-dedicated robot or the need to download a dialogue-dedicated application.

The present application has been made in view of the above, and an object of the present application is to enable a user to easily obtain response information.

The information processing apparatus according to the present application has an acquisition unit for acquiring a search query, a specific unit for specifying a conversation response corresponding to the content indicated by the search query, and an output control unit for outputting information based on the conversation response as a search result. And.

According to one aspect of the embodiment, the user can easily obtain the reply information.

FIG. 1 is a diagram showing the operation of the information processing system according to the embodiment. FIG. 2 is a diagram showing an example of a conversation information database. FIG. 3 is a diagram for explaining a model generated by the information processing apparatus. FIG. 4 is a diagram showing an example of a model generated by the information processing apparatus. FIG. 5 is a diagram showing an example of a response candidate information database. FIG. 6 is a diagram showing an example of a search page. FIG. 7 is a diagram showing how a vector, which is a distributed representation of a search query, is mapped to a vector space. FIG. 8 is a diagram showing a calculation example of the degree of similarity. FIG. 9 is a diagram showing an example of a page output to the terminal device. FIG. 10 is a diagram showing a configuration example of the terminal device according to the embodiment. FIG. 11 is a diagram showing a configuration example of the service providing server according to the embodiment. FIG. 12 is a diagram showing a configuration example of the information processing apparatus according to the embodiment. FIG. 13 is a diagram showing an example of information stored in the conversation information data. FIG. 14 is a diagram showing an example of information stored in the response candidate information database. FIG. 15 is a flowchart showing an example of the learning process. FIG. 16 is a flowchart showing an example of the distribution process. FIG. 17 is a hardware configuration diagram showing an example of a computer that realizes the functions of the information processing device.

Hereinafter, the information processing apparatus, the information processing method, and the embodiment for implementing the information processing program (hereinafter referred to as “the embodiment”) according to the present application will be described in detail with reference to the drawings. The following embodiments do not limit the information processing apparatus, information processing method, and information processing program according to the present application. Further, in each of the following embodiments, the same parts are designated by the same reference numerals, and duplicate explanations are omitted.

[1. Operation of information processing device]
First, the operation of the information processing apparatus will be described by taking as an example an information processing system 1 including an information processing apparatus 30 which is an example of the information processing apparatus of the present embodiment.

The information processing device of the present embodiment is a device that provides a search service to a user. As a search service, a service for searching posts such as Twitter (registered trademark), Facebook (registered trademark), Instagram (registered trademark), etc. using a query keyword (search query) as a clue can be considered. However, in such a search service, the information processing device can search for posts containing a given keyword, but cannot search for posts that can be answered conversationally to a query such as a question. For example, the information processing device posts answers such as "It was really amazing" and "It was a tremendous flight distance" in response to a query like the question "The home run of yesterday's game was amazing". It's difficult to find.

The information processing device of the present embodiment accepts query information in the form of conversational questions, and makes it possible to search for posts that can be regarded as responses to the post information from a database such as post information. This allows users to perform new searches that make them enjoy talking to their computers. Hereinafter, the information processing system 1 of the present embodiment will be described.

[1-1. Information processing system]
FIG. 1 is a diagram showing the operation of the information processing system 1 according to the embodiment. The information processing system 1 includes terminal devices 10 ₁ , 10 ₂ , 10 ₃ , 10 ₄ , a service providing server 20, and an information processing device 30. In the example of FIG. 1, four terminal devices are shown, but the number of terminal devices may be less than four or more than four. In the following description, the terminal device 10 ₁ , 10 ₂ , 10, ₃ , 10 ₄ and the like may be collectively referred to as the terminal device 10.

The terminal device 10 is, for example, a user terminal such as a smart device (smartphone or tablet), a mobile phone, or a personal computer. The terminal device 10 ₁ is a terminal used by the user A, the terminal device 10 ₂ is a terminal used by the user B, the terminal device 10 ₃ is a terminal used by the user C, and the terminal device 10 ₄ is a terminal used by the user D. The terminal to use. The terminal device 10 communicates with the service providing server 20 via the network. For example, the terminal device 10 (eg, terminal devices 10 ₁ , 10 ₂ , 10 ₃ ) acquires information on a posting page (eg, Twitter®) that allows a user to post. Further, the terminal device 10 transmits the information posted by the user to the posting page to the service providing server 20. Further, the terminal device 10 (for example, the terminal device 10 ₄ ) acquires the information of the search page from the information processing device 30.

The service providing server 20 is a server host computer that provides various services to client terminals. The service providing server 20 includes a portal site, an SNS (Social Networking Service) site, a search site, a bulletin board site, a game information distribution site, a news site, an auction site, a weather forecast site, a shopping site, a finance (stock price) site, and a route search site. , It functions as a map providing site, a travel site, a restaurant introduction site, etc.

In the case of the present embodiment, the service providing server 20 is a server that provides a service that enables a user to post information. For example, the service providing server 20 is a server that provides social networking services (SNS) such as Twitter (registered trademark), Facebook (registered trademark), Instagram (registered trademark), mixi (registered trademark), and LINE (registered trademark). be. The service provided by the service providing server 20 is not limited to the SNS. As long as the user can post information, the service providing server 20 may be a server that provides services other than the above-mentioned SNS. For example, the service providing server 20 may be a server that provides a bulletin board service. The service providing server 20 may be a server that provides a shopping service, as long as it is configured so that the user can post a review or the like of the purchased product. In the following explanation, the service that enables the user to post information is called a posting service, and the information posted by the user using the posting service is called posting information. The user communicates with another user through the service provided by the service providing server 20.

The information processing device 30 is a server host computer that provides various services to client terminals. The information processing device 30 functions as a search site. The information processing device 30 performs a process for allowing the user to easily obtain the response information. Specifically, the information processing device 30 regards the search query acquired from the terminal device 10 as the utterance of the user, and specifies the response corresponding to the utterance (search query). For example, the information processing apparatus 30 identifies response information suitable for a user's utterance from a plurality of response candidates prepared in advance. Then, the information processing apparatus 30 regards the response information as a search result and transmits it to the terminal apparatus 10. The user can easily obtain the response information via the search page.

The search page used by the user to send a search query to the information processing apparatus 30 is not limited to the web page. For example, the search page may be a page created for the purpose of being displayed on the user interface screen (for example, the screen of an application). The page may be generated based on the data (for example, the data of the search screen) pre-installed in the storage device included in the terminal device 10. The description of the "search page" shown in the following description can be replaced with a "search screen" or a "search interface". Further, the operation of the terminal device 10 described below includes not only the operation on the browser but also the operation on the pre-installed program (for example, on the application).

Hereinafter, the operation of the information processing system 1 including the information processing apparatus 30 will be described with reference to FIG. As described above, the information processing apparatus 30 identifies a response suitable for the user's utterance (search query). In order to specify the response, the information processing apparatus 30 generates in advance a learning model that converts the input sentence into a distributed representation, similar to the DSSM (Deep Semantic Similarity Model). The information processing device 30 generates a learning model based on information on conversations between users using a network (hereinafter referred to as conversation information). For example, the information processing apparatus 30 generates a learning model based on a user's post and another user's post for the post. Before explaining the specific operation of the response, the generation operation of the learning model will be described.

[1-2. Generation of learning model]
First, the service providing server 20 communicates with the terminal device 10 operated by the user regarding the posting service (steps S1a to S1c). When the service providing server 20 acquires the posted information from the terminal device 10, the service providing server 20 registers the acquired posted information in the database. Then, the service providing server 20 extracts the conversation pair from the database (step S2) and transmits it to the information processing apparatus 30 (step S3). When the information processing apparatus 30 acquires conversation pair information from the service providing server 20, the information processing device 30 registers the acquired information in the conversation information database. FIG. 2 is a diagram showing an example of a conversation information database. Each time the information processing apparatus 30 acquires information on a conversation pair from the service providing server 20, the information processing device 30 registers the information in the conversation information database. In the case of the example of FIG. 2, the information registered in the "message" item is the utterance on the user's network, and the information registered in the "reply" item is the response of another user to the utterance. be.

Subsequently, the information processing apparatus 30 generates a learning model (hereinafter, simply referred to as a model) based on the information of the conversation pair registered in the conversation information database (step S4). The model generated by the information processing apparatus 30 is, for example, a model that generates a distributed representation from a sentence. The generated distributed representation is, for example, a 1024-dimensional vector.

FIG. 3 is a diagram for explaining a model generated by the information processing apparatus 30. FIG. 3 shows how the message or response shown in FIG. 2 is mapped to the vector space via the model. The information processing apparatus 30 trains the model so that the distributed representation of the message and the distributed representation of the response paired with the message are mapped close to each other in the vector space. At the same time, the information processing apparatus 30 trains the model so that the distributed representation of the message and the distributed representation of the sentence unrelated to the message are mapped far away in the vector space. In the example of FIG. 3, in the information processing apparatus 30, the vector V22, which is a distributed expression of “let's eat rice”, and the vector V12, which is a distributed expression of the response “hungry”, are close to each other in the vector space. Train the model to be mapped. At the same time, in the information processing apparatus 30, the vector V22, which is a distributed expression of "let's eat rice", and the vector V16, which is a distributed expression of "Kyoto" unrelated to "let's eat rice", are mapped far away in the vector space. Train the model to be done. In addition, the information processing apparatus 30 trains the model using the vectors V24, V14, V25, V15, V26, V16, V28, V18 and the like.

FIG. 4 is an example of a model generated by the information processing apparatus 30. The model generated by the information processing apparatus 30 is, for example, a model using LSTM (Long Short-Term Memory) which is a kind of RNN (Recurrent Neural Network). In the example shown in FIG. 4, the model generated by the information processing apparatus 30 is composed of one embedding layer, three layers of LSTM RNN, and one layer of readout layer. The size of each layer is, for example, 256 dimensions for the embedding layer, 1024 dimensions for the LSTM, and 1024 dimensions for the readout layer. When a sentence (“Hello!” In the example of FIG. 4) is input to the embedding layer, a 1024-dimensional vector (distributed expression) is output from the readout layer.

In addition, in order to generate the response information, the information processing apparatus 30 generates the response candidate information in which the distributed representation of the response candidates is recorded in advance. At this time, the information processing apparatus 30 generates response candidate information using the model generated in step S4 (step S5). The response candidate information is registered in the response candidate database. FIG. 5 is a diagram showing an example of a response candidate information database. The information processing apparatus 30 acquires, for example, a sentence on the responding side as a response candidate among the conversation pairs acquired in step S3. Then, the information processing apparatus 30 generates a distributed representation of the response candidates using the model generated in step S4, and registers the response candidates in the response candidate information database. For example, the information processing apparatus 30 registers the vectors V11, V12, and V13, which are distributed expressions of "good morning", "hungry", and "Tokyo office", in the response candidate information database.

[1-3. Identification of reply]
Next, the specific operation of the reply will be described.

When the information processing device 30 receives the distribution request for the search page from the terminal device 10, the information processing device 30 transmits the search page to the terminal device 10. FIG. 6 is a diagram showing an example of a search page. The user inputs text (hereinafter, simply referred to as a search query) to be a search query in the text box T in the search page. In the example of FIG. 6, the sentence "Wait a minute?" Is input as a search query. When the user clicks the search button SB, the terminal device 10 sends a search query to the information processing device 30 (step S6).

When the search query is acquired from the terminal device 10, the information processing device 30 determines whether the search query satisfies a predetermined criterion. For example, the information processing apparatus 30 determines whether the search query is a natural sentence. It should be noted that various methods can be adopted for determining whether or not the sentence is a natural sentence. For example, the information processing apparatus 30 may perform morphological analysis of the search query and determine that the search query is a natural sentence when the search query contains a word as a predicate.

When the search query does not meet a predetermined criterion (for example, when the search query is not a natural sentence), the information processing apparatus 30 executes a normal search process. For example, the information processing apparatus 30 acquires a link to a page including a word designated as a search query as a search result, and transmits the acquired search result to the terminal apparatus 10.

When the search query meets a predetermined criterion (for example, when the search query is a natural sentence), the information processing apparatus 30 identifies the response of the conversation corresponding to the content indicated by the search query. At this time, the information processing apparatus 30 identifies the response of the conversation corresponding to the content indicated by the search query based on the model in which the conversation using the network between the users is learned. Specifically, the information processing apparatus 30 performs the following processing.

First, the information processing apparatus 30 generates a distributed representation of the search query (step S7). At this time, the information processing apparatus 30 generates a distributed representation of the search query using the model generated in step S4. For example, if the model generated in step S4 is configured to output a 1024-dimensional vector, the information processing apparatus 30 outputs a 1024-dimensional vector as a distributed representation of the search query. FIG. 7 is a diagram showing how the vector V35, which is a distributed representation of the search query, is mapped to the vector space. The vector V35 is a distributed representation (vector) of the search query "Wait a minute?". For ease of understanding, a part of the distributed representation of the response candidates (vectors V12, V14 to 17) is also mapped in the vector space shown in FIG. 7.

Then, the information processing apparatus 30 identifies the response of the conversation corresponding to the search query from the response candidates registered in the response candidate information database (step S8). For example, the information processing apparatus 30 specifies a response as follows.

First, the information processing apparatus 30 identifies the vector closest to the vector of the search query in the vector space (hereinafter, referred to as a similar vector) among the plurality of vectors of the response candidates. At this time, the information processing apparatus 30 specifies the similarity vector based on the similarity (for example, the cosine similarity). FIG. 8 is a diagram showing a calculation example of the degree of similarity. In the item of "distributed expression", the distributed expression (vector) of the sentence shown in "response candidate" is stored. In the item of "similarity", the cosine similarity between the vector shown in "distributed representation" and the vector V35 of the search query is stored. The larger the similarity value, the more similar the two vectors are. In the case of the example of FIG. 8, the vector V15 having a similarity of 0.506 is most similar to the vector V35 of the search query. In the case of the example of FIG. 8, the information processing apparatus 30 specifies the vector V15 as a vector similar to the vector V35 of the search query.

Then, the information processing apparatus 30 identifies the response candidate corresponding to the similar vector as the response of the conversation corresponding to the search query. In the example of FIG. 8, the information processing apparatus 30 specifies "I'm going first" corresponding to the vector V15 as a response to the search query "Wait a minute?".

Then, the information processing apparatus 30 generates the information of the page to be transmitted to the terminal apparatus 10 based on the response of the conversation specified in step S8. Then, the information processing apparatus 30 outputs the generated information as a search result (step S9). For example, the information processing device 30 outputs the generated information to the terminal device 10. FIG. 9 is a diagram showing an example of a page output to the terminal device 10. Specifically, FIG. 9 is a search result page corresponding to the search query "Wait a minute?". In the example of FIG. 9, the message M (“I'm going first”) is displayed as a response to the search query “Wait a minute?”. The page output by the information processing device 30 to the terminal device 10 may include information other than the message M. For example, the page may contain information about the page that contains the search query.

The information processing device 30 may output a link to a page including a conversation response as a search result. For example, the information processing apparatus 30 may specify a page including the response of the conversation specified in step S8, and output a link to the specified page as a search result. At this time, the information processing apparatus 30 may specify the page from the posting site of the acquisition source of the conversation pair registered in the conversation information database. In the example of FIG. 9, a link L is attached to the message M. When the user clicks the link L, the page displayed on the terminal device 10 transitions to the page including the message M.

According to the present embodiment, the information processing apparatus 30 is configured to output a response corresponding to the search query as a search result when the search query is a natural sentence. Therefore, the user can easily obtain the response information using the existing search interface. Since there is no need to prepare a new dedicated interface, the user can easily enjoy the conversation with the computer.

[2. Information processing system configuration example]
The operation of the information processing device has been described above by taking as an example the information processing system 1 including the information processing device 30 which is an example of the information processing device of the present embodiment. Hereinafter, the configuration of the information processing device of the present embodiment will be described. do. In the following description, the configuration of the information processing apparatus will be described using the information processing system 1 as an example. As described above, the information processing system 1 includes a terminal device 10, a service providing server 20, and an information processing device 30. The terminal device 10, the service providing server 20, and the information processing device 30 are connected via a network. The network is a communication network such as LAN (Local Area Network), WAN (Wide Area Network), telephone network (mobile telephone network, fixed telephone network, etc.), regional IP (Internet Protocol) network, and the Internet. The network may include a wired network or may include a wireless network.

[3. Configuration example of terminal device]
FIG. 10 is a diagram showing a configuration example of the terminal device 10 according to the embodiment. The terminal device 10 is an information display device (information processing device) used by a user to browse a page. The terminal device 10 is, for example, a communication terminal such as a smartphone, a tablet, or a mobile phone. As long as it has a communication function, the terminal device 10 may be an information processing terminal such as a personal computer (notebook PC, desktop PC) or a PDA (Personal Digital Assistant). In this case, the information processing terminal is also a kind of communication terminal. The terminal device 10 connects to the service providing server 20 via a network. As shown in FIG. 10, the terminal device 10 includes a communication unit 11, an input unit 12, a storage unit 13, an output unit 14, and a control unit 15. The configuration shown in FIG. 10 is a functional configuration, and the hardware configuration may be different from this.

The communication unit 11 is a communication interface that communicates with an external device. The communication unit 11 may be a network interface or a device connection interface. For example, the communication unit 11 may be a LAN interface such as a NIC (Network Interface Card), or may be a USB interface composed of a USB (Universal Serial Bus) host controller, a USB port, or the like. Further, the communication unit 11 may be a wired interface or a wireless interface. The communication unit 11 functions as a communication means of the terminal device 10. The communication unit 11 communicates with the service providing server 20 according to the control of the control unit 15.

The input unit 12 is an input device that receives various inputs from the outside. For example, the input unit 12 is an operation device for the user to perform various operations such as a keyboard, a mouse, and operation keys. The input unit 12 functions as an input means of the terminal device 10. When a touch panel is adopted for the terminal device 10, the touch panel is also included in the input unit 12. In this case, the user performs various operations by touching the screen with a finger or a stylus.

The storage unit 13 is a storage device capable of reading and writing data such as a DRAM (Dynamic Random Access Memory), an SRAM (Static Random Access Memory), a flash memory, and a hard disk. The storage unit 13 functions as a storage means for the terminal device 10. The storage unit 13 stores data of application software such as a web browser. The user communicates with the service providing server 20 via the application software.

The output unit 14 is a device that outputs various outputs such as sound, light, vibration, and an image to the outside. The output unit 14 functions as an output means of the terminal device 10. The output unit 14 includes a display device that displays various types of information. The display device is, for example, a liquid crystal display or an organic EL display. When a touch panel is adopted for the terminal device 10, the display device may be integrated with the input unit 12. The output unit 14 displays an image on the screen according to the control of the control unit 15.

The control unit 15 is a controller, and for example, various programs stored in a storage device inside the terminal device 10 by a processor such as a CPU (Central Processing Unit) or an MPU (Micro Processing Unit) store a RAM or the like. It is realized by being executed as a work area. Further, the control unit 15 is a controller, and may be realized by, for example, an integrated circuit such as an ASIC (Application Specific Integrated Circuit) or an FPGA (Field Programmable Gate Array).

[4. Service provision server configuration example]
Next, the configuration of the service providing server 20 will be described in detail. The service providing server 20 is a server host computer (hereinafter, simply referred to as “server”) that processes requests from client computers such as the terminal device 10 and the information processing device 30. The service providing server 20 may be a PC server, a midrange server, or a mainframe server. Further, the service providing server 20 may be configured by one server or may be configured by a plurality of servers that perform processing in cooperation with each other. When the service providing server 20 is composed of a plurality of servers, the installation locations of these servers may be separated. Even if the installation locations are far apart, these servers can be regarded as one service providing server if the processes are executed in cooperation with each other.

FIG. 11 is a diagram showing a configuration example of the service providing server 20 according to the embodiment. The service providing server 20 includes a communication unit 21, a storage unit 22, and a control unit 23. The configuration shown in FIG. 11 is a functional configuration, and the hardware configuration may be different from this.

The communication unit 21 is a communication interface that communicates with an external device. The communication unit 21 may be a network interface or a device connection interface. For example, the communication unit 21 may be a LAN interface such as a NIC, or may be a USB interface composed of a USB host controller, a USB port, or the like. Further, the communication unit 21 may be a wired interface or a wireless interface. The communication unit 21 functions as a communication means of the service providing server 20. The communication unit 21 communicates with the terminal device 10 according to the control of the control unit 23.

The storage unit 22 is a storage device that can read and write data such as a DRAM, SRAM, flash memory, and a hard disk. The storage unit 22 functions as a storage means for the service providing server 20. The storage unit 22 stores the user's posted information.

The control unit 23 is a controller, and is realized by, for example, using a processor such as a CPU or MPU to execute various programs stored in a storage device inside the service providing server 20 using RAM or the like as a work area. Further, the control unit 23 is a controller, and may be realized by an integrated circuit such as an ASIC or FPGA. The control unit 23 extracts conversation pair information (conversation information) from the posted information. Then, the control unit 23 transmits the conversation information extracted from the posted information to the information processing device 30 via the communication unit 21.

[5. Configuration example of information processing device]
Next, the configuration of the information processing apparatus 30 will be described in detail. The information processing device 30 is a computer that analyzes an auction. The information processing device 30 may be a terminal device or a server device. When the information processing device 30 is used as a terminal device, the information processing device 30 may be an information processing terminal such as a smartphone, a tablet, a mobile phone, a personal computer, or a PDA. When the information processing device 30 is used as a server device, the information processing device 30 may be a PC server, a midrange server, or a mainframe server. At this time, the information processing apparatus 30 may be configured by one server or may be configured by a plurality of servers that perform processing in cooperation with each other. When the information processing apparatus 30 is composed of a plurality of servers, the installation locations of these servers may be separated. Even if the installation locations are far apart, these servers can be regarded as one server device if the processes are executed in cooperation with each other.

FIG. 12 is a diagram showing a configuration example of the information processing apparatus 30 according to the embodiment. The information processing device 30 includes a communication unit 31, a storage unit 32, and a control unit 33. The configuration shown in FIG. 5 is a functional configuration, and the hardware configuration may be different from this.

The communication unit 31 is a communication interface that communicates with an external device. The communication unit 31 may be a network interface or a device connection interface. For example, the communication unit 31 may be a LAN interface such as a NIC, or may be a USB interface composed of a USB host controller, a USB port, or the like. Further, the communication unit 31 may be a wired interface or a wireless interface. The communication unit 31 functions as a communication means of the information processing device 30. The communication unit 31 communicates with the terminal device 10 and the service providing server 20 under the control of the control unit 33.

The storage unit 32 is a data-readable / writable storage device such as a DRAM, an SRAM, a flash memory, and a hard disk. The storage unit 32 functions as a storage means for the information processing device 30. The storage unit 32 stores the conversation information database 321, the response candidate information database 322, and the model 323.

The conversation information database 321 is a database that stores conversation information. FIG. 13 is a diagram showing an example of information stored in the conversation information database 321. In the example shown in FIG. 13, the conversation information database 321 has items such as "message" and "reply".

A "message" indicates a user's utterance using characters on the network. In the example of FIG. 13, information such as “good morning” and “let's eat rice” is stored in the “message”.

The "reply" indicates the response of another user to the user's utterance recorded in the "message" item. In the example of FIG. 13, information such as “good morning” and “hungry” is stored in the “reply”.

The response candidate information database 322 is a database that stores response candidate information. FIG. 14 is a diagram showing an example of information stored in the response candidate information database 322. In the example shown in FIG. 14, the response candidate information database 322 has items such as “response candidate” and “distributed expression”.

The "response candidate" indicates a candidate for a response to the user. In the example of FIG. 14, information such as "good morning" and "hungry" is stored in the "response candidate".

The "distributed expression" is a distributed expression of the sentence recorded in the item of "response candidate". The information stored in the "distributed representation" is, for example, a 1024-dimensional vector. In the example of FIG. 14, information such as vectors V11 and V12 is stored in the “response candidate”.

Data (learning model) for specifying a response to a message is stored in the model 323. For example, the model 323 stores data for converting a message into a distributed representation. Alternatively, the model 323 stores data for output in response to the input message. The model 323 includes an input layer into which a message is input, an output layer, a first element which is any layer from the input layer to the output layer and belongs to a layer other than the output layer, and the first element and the first element. A second element whose value is calculated based on the weight of is included, and a distributed representation (or response to the message) of the message input to the input layer is output according to the message input to the input layer. The information processing apparatus 30 may be made to function so as to output from.

Here, it is assumed that the model 323 is realized by the regression model shown by "y = a1 * x1 + a2 * x2 + ... + ai * xi". In this case, the first element included in the model 323 corresponds to input data (xi) such as x1 and x2. Further, the weight of the first element corresponds to the coefficient ai corresponding to xi. Here, the regression model can be regarded as a simple perceptron having an input layer and an output layer. When each model is regarded as a simple perceptron, the first element corresponds to any node of the input layer, and the second element can be regarded as the node of the output layer.

Further, it is assumed that the model 323 is realized by a neural network having one or a plurality of intermediate layers such as a DNN (Deep Neural Network). In this case, the first element included in the model 323 corresponds to either the node of the input layer or the intermediate layer. Further, the second element corresponds to a node in the next stage, which is a node to which a value is transmitted from a node corresponding to the first element. Further, the weight of the first element corresponds to a connection coefficient which is a weight considered for the value transmitted from the node corresponding to the first element to the node corresponding to the second element.

The information processing apparatus 30 calculates a distributed representation (or a response to a message) using a model having an arbitrary structure such as the regression model and the neural network described above. Specifically, the model 323 is set with a coefficient to output a distributed representation (or a response to the message) when a message is input. The information processing apparatus 30 uses such a model 323 to calculate a distributed representation (or a response to a message).

In the above example, the model 323 is a model (hereinafter referred to as model X) that outputs a distributed representation (or a response to the message) of the message when the message is input. However, the model 323 according to the embodiment may be a model generated based on the result obtained by repeating the input / output of data to the model X. For example, the model 323 may be a model (hereinafter referred to as model Y) trained to input a message and output a distributed expression (or a response to the message) output by the model X. Alternatively, the model 323 may be a model trained to input a message and output the output value of the model Y.

Further, when the information processing apparatus 30 performs estimation processing using GAN (Generative Adversarial Networks), the model 323 may be a model constituting a part of GAN.

Returning to FIG. 5, the control unit 33 is a controller, and for example, various programs stored in the storage device inside the information processing device 30 are executed by a processor such as a CPU or MPU using the RAM or the like as a work area. Is realized by. Further, the control unit 33 is a controller, and may be realized by an integrated circuit such as an ASIC or FPGA.

As shown in FIG. 5, the control unit 33 includes an acquisition unit 331, a generation unit 332, a discrimination unit 333, a specific unit 334, a search unit 335, and an output control unit 336. The blocks constituting the control unit 33 (acquisition unit 331 to output control unit 336) are functional blocks indicating the functions of the control unit 33, respectively. These functional blocks may be software blocks or hardware blocks. For example, each of the above-mentioned functional blocks may be one software module realized by software (including a microprogram), or may be one circuit block on a semiconductor chip (die). Of course, each functional block may be one processor or one integrated circuit. The method of configuring the functional block is arbitrary. The control unit 33 may be configured in a functional unit different from the above-mentioned functional block.

The acquisition unit 331 acquires a search query from the terminal device 10.

The generation unit 332 generates the model 323 based on the conversation information acquired from the service providing server 20.

The determination unit 333 determines whether the search query satisfies a predetermined criterion. For example, determine if the search query is a natural sentence.

The identification unit 334 identifies the response of the conversation corresponding to the content indicated by the search query. Further, the specific unit 334 specifies the response of the conversation corresponding to the content indicated by the search query when the determination unit 333 determines that the search query satisfies a predetermined criterion. For example, when the determination unit 333 determines that the search query is a natural sentence, the specific unit 334 specifies the response of the conversation corresponding to the natural sentence. In addition, the specific unit 334 specifies a plurality of conversation responses corresponding to the contents indicated by the search query. In addition, the specific unit 334 specifies a page containing a response to the conversation.

The identification unit 334 identifies the response of the conversation corresponding to the content indicated by the search query based on the model of learning the conversation using the network between the users. For example, the identification unit 334 identifies the response of the conversation corresponding to the content indicated by the search query, based on the model that learned the user's post and the posts of other users for the post. Alternatively, the identification unit 334 identifies the response of the conversation corresponding to the content indicated by the search query based on the model learned from the question in the user's predetermined support service and the answer of the operator to the question.

The search unit 335 executes a search using the response of the conversation as a search query.

The output control unit 336 outputs information based on the response of the conversation as a search result. For example, the output control unit 336 outputs information based on the response as a search result when the specific unit 334 specifies the response of the conversation. Further, the output control unit 336 outputs the responses of the specified plurality of conversations as the search result. Further, the output control unit 336 outputs the result searched by the search unit 335 as a search result. Further, the output control unit 336 outputs a link to a page including a conversation response as a search result.

[6. Information processing device processing flow]
Next, the procedure of the processing executed by the information processing apparatus 30 will be described. The process executed by the information processing apparatus 30 is divided into a learning process and a distribution process. The information processing apparatus 30 includes a multitasking OS, and can execute learning processing and distribution processing in parallel.

[6-1. Learning process]
First, the learning process will be described. FIG. 15 is a flowchart showing an example of the learning process. The information processing apparatus 30 starts the learning process when it receives a process start command from the user who operates the information processing apparatus 30.

First, the information processing apparatus 30 acquires conversation information from the service providing server 20 (step S11). Conversational information includes information about the message and the response to the message. Then, the information processing apparatus 30 generates a model 323 that generates a distributed representation from the message (step S12). The information processing apparatus 30 may generate a model 323 based on a conversation using a network between users.

Subsequently, the information processing apparatus 30 generates response candidate information (step S13). The response candidate information includes a plurality of information on the response candidates to the user. A response candidate is associated with a distributed representation of the response candidate. The information processing apparatus 30 inputs a response candidate into the model 323 to generate a distributed representation associated with the response candidate.

When the generation of the response candidate information is completed, the information processing apparatus 30 ends the process.

[6-2. Delivery process]
Next, the delivery process will be described. FIG. 16 is a flowchart showing an example of the distribution process. The information processing apparatus 30 starts the distribution process when the power is turned on.

First, the information processing apparatus 30 determines whether or not the search query from the terminal apparatus 10 has been received (step S21). If the search query has not been received (step S21: No), the information processing apparatus 30 executes step S21 again.

When the search query is received (step S21: Yes), the information processing apparatus 30 determines whether the search query is a natural sentence (step S22). If the search query is not a natural sentence (step S22: No), the information processing apparatus 30 proceeds to step S24.

When the search query is a natural sentence (step S22: Yes), the information processing apparatus 30 uses the model 323 to specify the response of the conversation to the search query (step S23).

Subsequently, the information processing apparatus 30 executes a search for a page including a search query (step S24). The information processing apparatus 30 may execute the search by regarding the response of the conversation specified in step S23 as a search query.

When the search is completed, the information processing apparatus 30 generates a page including the search result of step S24 (step S25). When the information processing apparatus 30 identifies the response of the conversation in step S23, the information processing apparatus 30 may generate a page in which the response is included as a search result.

When the page generation is completed, the information processing device 30 transmits the generated page to the terminal device 10 (step S26). When the transmission is completed, the information processing apparatus 30 returns to step S21 and waits for the reception of the search result.

[7. Modification example]
The above embodiment shows an example, and various modifications and applications are possible.

For example, in the above embodiment, the information processing apparatus 30 has specified one conversation response corresponding to the content indicated by the search query. However, the information processing apparatus 30 may specify a plurality of conversation responses corresponding to the contents indicated by the search query. Then, the information processing apparatus 30 may output a plurality of responses as search results to the terminal apparatus 10. As a result, the user can feel as if he / she receives a response from a plurality of persons to his / her utterance.

Further, the information processing apparatus 30 executes a search using the response of the conversation corresponding to the search query as a search query. Then, the information processing apparatus 30 may output a search result using the response of the conversation as a search query to the terminal apparatus 10. For example, the information processing device 30 may output page information (for example, a link to a page or a part of a page) including a conversation response to the terminal device 10. As a result, the user can obtain a response to the utterance and also obtain information for digging deep into the conversation. The user can use the obtained information to enhance the conversation with the person in the real world.

Further, in the above-described embodiment, the model 323 was generated based on the information of the conversation pair extracted from the posted information. However, the information processing apparatus 30 may generate the model 323 based on the question in the user's predetermined support service and the answer of the operator to the question. The predetermined support service may be a telephone support service relating to the sale and purchase of goods and the use of goods. Then, the information processing apparatus 30 may generate a conversation response corresponding to the content indicated by the search query based on the model 323.

Further, in the above-described embodiment, the model 323 outputs a distributed representation of the message, but the model 323 may be a learning model configured to output a response to the message.

The control device for controlling the information processing device 30 of the present embodiment may be realized by a dedicated computer system or a normal computer system. For example, a program for executing the above-mentioned operation is stored in a computer-readable recording medium such as an optical disk, a semiconductor memory, a magnetic tape, and a flexible disk and distributed, and the program is installed in the computer to perform the above-mentioned processing. The control device may be configured by executing. The control device may be an external device (for example, a personal computer) of the information processing device 30, or an internal device (for example, a control unit 33). Further, the above program may be stored in a disk device provided in a server device on a network such as the Internet so that it can be downloaded to a computer or the like. Further, the above-mentioned functions may be realized by the cooperation between the OS (Operating System) and the application software. In this case, the part other than the OS may be stored in a medium and distributed, or the part other than the OS may be stored in the server device so that it can be downloaded to a computer or the like.

Further, among the processes described in each of the above embodiments, all or part of the processes described as being automatically performed can be manually performed, or the processes described as being manually performed. It is also possible to automatically perform all or part of the above by a known method. In addition, information including processing procedures, specific names, various data and parameters shown in the above documents and drawings can be arbitrarily changed unless otherwise specified. For example, the various information shown in each figure is not limited to the information shown in the figure.

Further, each component of each of the illustrated devices is a functional concept, and does not necessarily have to be physically configured as shown in the figure. That is, the specific form of distribution / integration of each device is not limited to the one shown in the figure, and all or part of them may be functionally or physically distributed / physically in any unit according to various loads and usage conditions. Can be integrated and configured.

In addition, the above-described embodiments can be appropriately combined as long as the processing contents do not contradict each other.

[8. Hardware configuration]
The information processing apparatus 30 according to the embodiment and the modification can also be realized by, for example, a computer 1000 having a configuration as shown in FIG. FIG. 17 is a hardware configuration diagram showing an example of a computer that realizes the functions of the information processing apparatus 30. The computer 1000 includes a CPU (Central Processing Unit) 1100, a RAM 1200, a ROM 1300, an HDD (Hard Disk Drive) 1400, a communication interface (I / F) 1500, an input / output interface (I / F) 1600, and a media interface (I / F). ) Has 1700.

The CPU 1100 operates based on a program stored in the ROM 1300 or the HDD 1400, and controls each part. The ROM 1300 stores a boot program executed by the CPU 1100 when the computer 1000 is started, a program depending on the hardware of the computer 1000, and the like.

The HDD 1400 stores a program executed by the CPU 1100, data used by such a program, and the like. The communication interface 1500 receives data from another device via the network N and sends it to the CPU 1100, and transmits the data generated by the CPU 1100 to the other device via the network N.

The CPU 1100 controls an output device such as a display or a printer, and an input device such as a keyboard or a mouse via the input / output interface 1600. The CPU 1100 acquires data from the input device via the input / output interface 1600. Further, the CPU 1100 outputs the generated data to the output device via the input / output interface 1600.

The media interface 1700 reads a program or data stored in the recording medium 1800 and provides the program or data to the CPU 1100 via the RAM 1200. The CPU 1100 loads the program from the recording medium 1800 onto the RAM 1200 via the media interface 1700, and executes the loaded program. The recording medium 1800 is, for example, an optical recording medium such as a DVD (Digital Versatile Disc) or PD (Phase change rewritable Disk), a magneto-optical recording medium such as MO (Magneto-Optical disk), a tape medium, a magnetic recording medium, or a semiconductor memory. And so on.

For example, when the computer 1000 functions as the information processing apparatus 30 according to the embodiment, the CPU 1100 of the computer 1000 functions the control unit 33 by executing a program or data (for example, model 323) loaded on the RAM 1200. To realize. The CPU 1100 of the computer 1000 reads these programs or data (eg, model 323) from the recording medium 1800 and executes them, but as another example, even if these programs are acquired from another device via the network N. good.

Although some of the embodiments of the present application have been described in detail with reference to the drawings, these are examples, and various modifications are made based on the knowledge of those skilled in the art, including the embodiments described in the disclosure line of the invention. It is possible to carry out the present invention in other modified forms.

[9. effect〕
According to the present embodiment, the information processing apparatus 30 acquires a search query and identifies a conversation response corresponding to the content indicated by the search query. Then, the information processing apparatus 30 outputs information based on the response of the conversation as a search result. This allows the user to easily obtain response information using an existing search interface.

The information processing apparatus 30 determines whether the search query satisfies a predetermined criterion. When the information processing apparatus 30 determines that the search query satisfies a predetermined criterion, the information processing apparatus 30 identifies the response of the conversation corresponding to the content indicated by the search query. When the information processing apparatus 30 identifies a response to a conversation, the information processing apparatus 30 outputs information based on the response as a search result. This also allows the user to obtain response information while obtaining normal search results using an existing search interface.

The information processing device 30 determines whether the search query is a natural sentence. When the information processing apparatus 30 determines that the search query is a natural sentence, the information processing apparatus 30 identifies the response of the conversation corresponding to the natural sentence. This also allows the user to obtain response information while obtaining normal search results using an existing search interface.

The information processing apparatus 30 specifies a plurality of conversation responses corresponding to the contents indicated by the search query. The information processing device 30 outputs the responses of the specified plurality of conversations as a search result. As a result, the user can feel as if he / she receives a response from a plurality of persons to his / her utterance.

The information processing apparatus 30 executes a search using the response of the conversation as a search query. The information processing apparatus 30 outputs a search result using the response of the conversation as a search query. As a result, the user can obtain a response to the utterance and also obtain information for digging deep into the conversation. The user can use the obtained information to enhance the conversation with the person in the real world.

The information processing apparatus 30 identifies a page containing a conversation response. The information processing device 30 outputs a link to a page including a conversation response as a search result. This allows the user to get a response to the utterance and sell information to dig deeper into the conversation. The user can use the obtained information to enhance the conversation with the person in the real world.

The information processing apparatus 30 identifies the response of the conversation corresponding to the content indicated by the search query based on the model in which the conversation using the network between the users is learned. As a result, the information processing apparatus 30 can specify a response that matches the human senses.

The information processing apparatus 30 identifies the response of the conversation corresponding to the content indicated by the search query based on the model in which the user's post and the posts of other users in response to the post are learned. As a result, the information processing apparatus 30 can specify a response that matches the human senses.

The information processing apparatus 30 identifies the answer of the conversation corresponding to the content indicated by the search query based on the model in which the question in the predetermined support service of the user and the answer of the operator to the question are learned. As a result, the information processing apparatus 30 can specify an answer that matches the human sense as a response.

Although some of the embodiments of the present application have been described in detail with reference to the drawings, these are examples, and various modifications are made based on the knowledge of those skilled in the art, including the embodiments described in the disclosure column of the invention. It is possible to carry out the present invention in other modified forms.

Further, the above-mentioned "section, module, unit" can be read as "means" or "circuit". For example, the specific unit can be read as a specific means or a specific circuit.

1 ... Information processing system 10, 10 ₁ , 10 ₂ , 10 ₃ , 10 ₄ ... Terminal device 11, 21, 31 ... Communication unit 12 ... Input unit 13, 22, 32 ... Storage unit 14 ... Output unit 15, 23, 33 ... Control unit 20 ... Service providing server 30 ... Information processing device 321 ... Conversation information database 322 ... Response candidate information database 323 ... Model 331 ... Acquisition unit 332 ... Generation unit 333 ... Discrimination unit 334 ... Specific unit 335 ... Search unit 336 ... Output Control unit SB ... Search button L ... Link M ... Message T ... Text box

Claims (7)

The acquisition part that acquires the search query, and
A specific part that identifies the response of the conversation corresponding to the content indicated by the search query, and
A search unit that executes a search using the response of the conversation as a search query,
It is equipped with an output control unit that outputs information based on the response of the conversation as a search result.
The information based on the response of the conversation is at least one of the identified response, a link to a page containing the response, and the result of searching the response as a search query.
The specific unit identifies a plurality of pages containing the response of the conversation,
The output control unit outputs a plurality of links to the page including the response of the conversation as the search result.
An information processing device characterized by this. Further, a discriminator for determining whether the search query is a natural sentence is provided.
When the discriminating unit determines that the search query is a natural sentence, the specific unit identifies the response of the conversation corresponding to the content indicated by the search query.
When the specific unit specifies a response to the conversation, the output control unit outputs information based on the response as a search result.
The information processing apparatus according to claim 1. The specific unit identifies the response of the conversation corresponding to the content indicated by the search query based on the model of learning the conversation using the network between users.
The information processing apparatus according to claim 1 or 2 . The specific unit identifies the response of the conversation corresponding to the content indicated by the search query based on the model learned from the user's post and the posts of other users to the post.
The information processing apparatus according to any one of claims 1 to 3 , wherein the information processing apparatus is characterized by the above. The identification unit identifies the response of the conversation corresponding to the content indicated by the search query based on the model learned from the question in the user's predetermined support service and the answer of the operator to the question.
The information processing apparatus according to any one of claims 1 to 4 , wherein the information processing apparatus is characterized by the above. It is an information processing method executed by an information processing device.
The acquisition process to acquire the search query and
A specific process that identifies the response of the conversation corresponding to the content indicated by the search query, and
A search process that executes a search using the response of the conversation as a search query,
Includes an output control step that outputs information based on the response of the conversation as a search result.
The information based on the response of the conversation is at least one of the identified response, a link to a page containing the response, and the result of searching the response as a search query.
In the specific step, a plurality of pages containing the response of the conversation are specified.
In the output control step, a plurality of links to the page including the response of the conversation are output as the search result.
An information processing method characterized by that. The acquisition procedure to acquire the search query and
Specific steps to identify the response of the conversation corresponding to the content indicated by the search query, and
A search procedure for executing a search using the response of the conversation as a search query, and
Let the computer execute the output control procedure that outputs the information based on the response of the conversation as the search result.
The information based on the response of the conversation is at least one of the identified response, a link to a page containing the response, and the result of searching the response as a search query.
In the specific procedure, a plurality of pages containing the response of the conversation are specified.
In the output control procedure, a plurality of links to the page including the response of the conversation are output as the search result.
An information processing program characterized by this.

Images