JP6310363B2 - Information processing apparatus, information processing method, and information processing program - Google Patents

Description

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

  2. Description of the Related Art Conventionally, there is known an interactive agent system that outputs a message corresponding to a message to the user terminal in response to receiving the message from the user terminal. In such an interactive agent system, there is provided a technique for outputting various messages in response to a message received from a user terminal and outputting a message suitable for the user of the user terminal.

  Also, in the above interactive agent system, when a predetermined keyword is included in a message from the user terminal, a technique for outputting an advertisement corresponding to the keyword to the user terminal as a message is known.

JP 2013-122012 A JP 2011-146046 A

  However, the above-described prior art cannot always improve the quality of dialogue. Specifically, in the above-described conventional technology, the output of the advertisement is determined according to the rules relating to the dialog set in advance. Therefore, when the flow of the dialog does not match the actual situation of the conversation, the quality of the dialog deteriorates. In some cases, it may not be possible to output highly appealing advertisements.

  An object of the present application is to provide an information processing apparatus, an information processing method, and an information processing program capable of improving the quality of dialogue.

  An information processing apparatus according to the present application relates to an acquisition unit that acquires the number of times that information displayed on a screen is specified by a user, and the information specified by the user based on the number of times acquired by the acquisition unit And an update unit for updating the evaluation relating to the voice dialogue to be performed.

  According to one aspect of the embodiment, there is an effect that the quality of dialogue can be improved.

FIG. 1 is a diagram illustrating an example of an automatic dialogue process according to the embodiment. FIG. 2 is a diagram illustrating an example of a determination tree according to the embodiment. FIG. 3 is a diagram illustrating a configuration example of the interactive agent system according to the embodiment. FIG. 4 is a diagram illustrating a configuration example of the information processing apparatus according to the embodiment. FIG. 5 is a diagram illustrating an example of the dialogue information storage unit according to the embodiment. FIG. 6 is a diagram illustrating an example of the advertisement information storage unit according to the embodiment. FIG. 7 is a diagram illustrating an example of a user terminal according to the embodiment. FIG. 8 is a diagram illustrating an example of a determination tree according to the embodiment. FIG. 9 is a sequence diagram showing an automatic dialogue processing procedure according to the embodiment. FIG. 10 is a diagram illustrating an example of a determination tree according to the modification. FIG. 11 is a diagram illustrating an example of a determination tree according to a modification. FIG. 12 is a hardware configuration diagram illustrating an example of a computer that implements the functions of the information processing apparatus.

  Hereinafter, a mode for realizing an information processing apparatus, an information processing method, and an information processing program according to the present application (hereinafter referred to as “embodiment”) will be described in detail with reference to the drawings. Note that the information processing apparatus, the information processing method, and the information processing program according to the present application are not limited by this embodiment. Moreover, in the following each embodiment, the same site | part is attached | subjected the same code | symbol and the overlapping description is abbreviate | omitted.

  First, an example of an automatic dialogue process according to the embodiment will be described with reference to FIG. FIG. 1 is a diagram illustrating an example of an automatic dialogue process according to the embodiment. The response generation server 100 illustrated in FIG. 1 is an information processing apparatus that provides a voice service that outputs a response message based on response information with respect to utterance information about an utterance input by a user as an input message. Specifically, the response generation server 100 selects response information to be output with respect to the utterance information from the user using the determination tree as the determination information that is information regarding the rules used in the interactive agent system 1.

[1-1. (Judgment tree)
Here, the determination tree used by the response generation server 100 will be described with reference to FIG. FIG. 2 is a diagram illustrating an example of a determination tree according to the embodiment. In other words, the dialog agent system 1 realizes a dialog with the user by using a decision tree as shown in FIG. Specifically, the response generation server 100 determines response information to be returned to the user with respect to the utterance information that is the user's utterance based on the determination tree as illustrated in FIG.

  In the example shown in FIG. 2, the broken line block indicates a detection node, and the solid line block indicates an operation node. Further, node IDs N101 to N113 are assigned to these blocks, respectively. For example, a broken line block ““ curry ”detection” with a node ID “N101” indicates a detection node, and a solid line block ““ is delicious ”with a node ID“ N102 ”indicates an operation node. The detection node defines a processing procedure corresponding to the utterance information from the user, and the operation node defines a processing procedure corresponding to the response information. Therefore, the decision tree shown in FIG. 2 includes a detection node and an operation node.

  Further, the nodes shown in FIG. 2 are associated by being connected by an arrow line. Specifically, the node on the starting point side of the arrow line is associated as the connection source node, and the node on the arrow side is associated as the connection destination node. For example, the arrow line that associates the node with the node ID “N104” and the node with the node ID “N106” indicates that the connection source node is the detection node with the node ID “N104” and the connection destination node is the operation with the node ID “N106”. Indicates a node. The node with the node ID “N104” is also associated with the operation node with the node ID “N107” as the connection source node. Here, the response generation server 100 detects a keyword from the user's utterance, and transitions to the detection node if the detection node corresponding to the keyword is associated as the connection destination node. Specifically, the response generation server 100 sets the response information ““ Where curry? ”Corresponding to the operation node having the node ID“ N103 ”. Roppongi? If “Roppongi” is included in the user's utterance after responding to the user, it is determined that “Roppongi” detection corresponding to the node ID “N104” has been performed, and the connection destination node A transition is made to an operation node of either a certain node ID “N106” or a node ID “N107”. Then, response information of the corresponding operation node is returned to the user. Specifically, when transitioning to the node ID “N106”, the response generation server 100 responds to the user with the response information ““ I like curry at store A ”” corresponding to the operation node with the node ID “N106”. . In this way, the dialog agent system 1 advances the dialog with the user.

[1-2. Node and transition probability
Here, the operation nodes in the determination tree shown in FIG. 2 include an advertisement node that is an operation node related to an advertisement and other operation nodes. For example, the operation node having the node ID “N102” is not an advertisement node because it is response information that does not include a specific store name or product name ““ It's delicious ””. On the other hand, the operation node of the node ID “N106” is an advertisement node because it is response information including a specific store name “I like curry at store A”. The dialogue agent system 1 provides the user with an advertisement in a natural conversation by interweaving such an advertisement node with other operation nodes. Here, the response generation server 100 accepts registration of the contents of the advertisement node, various information related to the advertisement node, the connection relationship between the advertisement node and the operation node, etc. from the terminal device used by the advertiser or the advertising agency. Also good. When accepting such registration, the response generation server 100 sets a new operation node in order to guide the dialog with the user from the dialog corresponding to the registered operation node to the dialog corresponding to the advertisement node to be registered. The registration may be accepted together with the registration of the advertisement node.

  Each node in the determination tree is associated with each other including a transition probability. In the example shown in FIG. 2, a transition probability is given above the arrow line of the determination tree, and the transition from each detection node to the operation node is performed according to the transition probability. For example, since the transition probabilities of the operation nodes with the node ID “N106” and the node ID “N107” shown in FIG. 2 are both 0.5, the detection node with the node ID “N104” makes a transition with a probability of 50%. . Specifically, the response generation server 100 sends the response information ““ Where curry? ” Roppongi? If “Roppongi” is included in the utterance information from the user after responding “” to the user, it is determined that “Detection of“ Roppongi ”” has been performed in N104, and transition to N106 or N107 with a probability of 50%. To do. That is, in the example illustrated in FIG. 2, the transition from the operation node to the detection node is performed based on the utterance information from the user, and the transition from the detection node to the operation node is performed based on the transition probability.

[1-3. (Automatic dialogue processing)
Returning to FIG. 1, the automatic dialogue processing according to the embodiment will be described. As shown in FIG. 1, the response generation server 100 receives utterance information related to the user's utterance as an input message from the user terminal 10 (step S11). An outline of processing in which the dialogue agent system 1 provides voice services to the user, such as generation of speech information in the user terminal 10, will be described later. Subsequently, the response generation server 100 selects response information corresponding to the acquired utterance information from the determination tree, and transmits the selected response information to the user terminal 10 (step S12). At this time, if there is related information related to the selected response information, the response generation server 100 also transmits the related information. Note that as the related information, various information related to the advertisement, such as information related to the response information related to the advertisement, for example, the URL of the advertiser's home page or information related to the campaign performed by the advertiser, is used. The related information may be registered by the advertiser together with the response information, or may be registered after the response information is registered. When the response information is output by the user terminal 10 and the user terminal 10 receives the related information, the related information is displayed on the user terminal 10 so as to be designated. In addition, when selecting the response information corresponding to the utterance information, the response generation server 100 selects the response information based on the transition probability associated with the transition from the detection node to the operation node in the determination tree.

  Next, when the related information displayed so as to be specifiable on the user terminal 10 is designated by the user, designation information related to designation of the related information by the user is transmitted to the advertisement log server 50 (step S13). The advertisement log server 50 that has acquired the designation information transmits a Web page corresponding to the related information to the user terminal 10 (step S14), and the user terminal 10 displays the acquired Web page. At this time, the advertisement log server 50 counts the number of display times (number of impressions) of the Web page related information displayed based on the specified information, that is, the number of times the advertisement is specified by the user. Here, the advertisement log server 50 is an information processing apparatus such as a Web server, for example. Note that the advertisement log server 50 may only count the number of display times of the Web page related information displayed based on the specified information. In this case, the user terminal 10 transmits the specified information to the corresponding Web server or the like, and displays the Web page acquired from the Web server or the like. Then, the advertisement log server 50 acquires information related to the number of display times from a Web server or the like at a predetermined interval such as every other day.

  Next, the advertisement log server 50 transmits, to the response generation server 100, log information including the number of display times of the Web page related information, that is, the number of times that the advertisement is designated by the user, every predetermined period, for example, every other day. (Step S15). The response generation server 100 that has acquired the log information updates the transition probability based on the log information (step S16).

  As described above, the response generation server 100 according to the embodiment updates the transition probability, which is the probability based on the evaluation regarding the voice conversation, using the log information including the display count of the Web page displayed based on the designation information. The Web page displayed in the dialog can also be highly evaluated for the conversation flow that the user wants to browse. In other words, the response generation server 100 can improve the effectiveness of the advertisement by automatically evaluating the dialog based on the decision tree and facilitating the generation of a highly evaluated dialog flow.

[1-4. Dialogue agent system configuration]
As illustrated in FIG. 3, the dialog agent system 1 according to the present embodiment includes a user terminal 10, a speech recognition server 20, a speech synthesis server 30, a service API (Application Program Interface) server 40, an advertisement log A server 50 and a response generation server 100 are included. The user terminal 10, the speech recognition server 20, the speech synthesis server 30, the service API server 40, the advertisement log server 50, and the response generation server 100 perform information transmission / reception via the communication network 60.

  From here, the outline | summary of the process in which the dialogue agent system 1 provides a voice service to a user is demonstrated. The user terminal 10 is an information processing apparatus such as a mobile phone, a smartphone, a PDA (Personal Digital Assistant), a tablet PC, a notebook PC, or a desktop PC. When the user terminal 10 detects the user's speech after the application is started, the user terminal 10 transmits speech data of the speech to the speech recognition server 20.

  When the speech recognition server 20 receives the speech data from the user terminal 10, the speech recognition server 20 converts the speech data into text data, and transmits the speech text data to the user terminal 10. The user terminal 10 that has received the utterance text data from the speech recognition server 20 transmits the utterance text data to the response generation server 100 as utterance information.

  As described above, the response generation server 100 transmits response information to the user terminal 10 according to the utterance information transmitted from the user terminal 10. Specifically, when the response generation server 100 receives the utterance information from the user terminal 10, the response generation server 100 selects the response information based on the determination tree. Then, the response generation server 100 transmits the text data of the response information to the speech synthesis server 30. The speech synthesis server 30 transmits an intermediate notation of response information generated by performing speech synthesis processing for synthesizing speech from text data of response information to the response generation server 100. The response generation server 100 transmits response information including intermediate notation for voice output and text data for display to the user terminal 10. In addition, when there is related information related to the response information, the response generation server 100 transmits the related information to the user terminal 10. The response generation server 100 acquires from the service API server 40 as necessary, and transmits it to the user terminal 10.

  The user terminal 10 outputs the response voice using the intermediate notation for voice output included in the received response information, and displays the response contents using the display text data included in the response information. . Further, when the user terminal 10 receives the related information, the user terminal 10 displays the received related information so that it can be designated. In this way, the dialogue agent system 1 realizes a voice service that makes an appropriate response to the user's utterance.

[2. Configuration of information processing apparatus]
Next, the configuration of the response generation server 100 according to the embodiment will be described with reference to FIG. FIG. 4 is a diagram illustrating a configuration example of the response generation server 100 according to the embodiment. As illustrated in FIG. 4, the response generation server 100 includes a communication unit 110, a storage unit 120, and a control unit 130. The response generation server 100 includes an input unit (for example, a keyboard and a mouse) that receives various operations from an administrator of the response generation server 100 and a display unit (for example, a liquid crystal display) for displaying various types of information. You may have.

(Communication unit 110)
The communication unit 110 is realized by a NIC (Network Interface Card), for example. The communication unit 110 is connected to the communication network 60 by wire or wireless. The communication unit 110 transmits and receives information to and from the advertisement log server 50 and the user terminal 10 via the communication network 60.

(Storage unit 120)
The storage unit 120 is realized by, for example, a semiconductor memory device such as a RAM (Random Access Memory) or a flash memory, or a storage device such as a hard disk or an optical disk. As shown in FIG. 4, the storage unit 120 according to the embodiment includes a dialogue information storage unit 121 and an advertisement information storage unit 122.

(Dialogue information storage unit 121)
The dialogue information storage unit 121 according to the embodiment stores information related to dialogue. For example, the dialogue information storage unit 121 stores information about the determination tree as shown in FIG. 2 as information about the dialogue.

  FIG. 5 shows an example of the dialogue information storage unit 121 according to the embodiment. As illustrated in FIG. 5, the dialogue information storage unit 121 includes items such as “node ID”, “transition destination node ID”, “transition probability”, and “detection word”.

  “Node ID” corresponds to the node ID of the determination tree shown in FIG. 2 and indicates identification information for identifying each node. The “transition destination node ID” indicates identification information for identifying a node that makes a transition from the node having the node ID. Specifically, the “transition destination node ID” indicates identification information for identifying the node corresponding to the connection destination node when the node ID is the connection source node in the determination tree illustrated in FIG. “Transition probability” indicates the probability of transition from the node of the node ID to each node of the transition destination node ID. “Detection word” indicates a keyword for making a transition from the node having the node ID to each node having the transition destination node ID.

  The example illustrated in FIG. 5 corresponds to the determination tree illustrated in FIG. 2 and indicates that there are three transition destination node IDs N102, N103, and N106 that transition from N101. In FIG. 5, the transition probability from N101 to N102 is “0.33”, the transition probability from N101 to N103 is “0.33”, and the transition probability from N101 to N106 is “0.33”. It is shown that. Here, as shown in the determination tree of FIG. 2, there is no transition destination node that transitions from N102, and therefore the transition destination node ID of the corresponding location is “−”.

  Further, in the example shown in FIG. 5, it is shown that there are two transition destination node IDs, N104 and N105, which transition from N103. Since N103 is an operation node, the transition to the transition destination node ID is performed by detecting a keyword included in the user's utterance information, so that the “detection word” for transition of each transition destination node ID transitions. Stored in association with the destination node ID. Specifically, since the transition from N103 to N104 is performed by detecting the detection word “Roppongi”, the detection word in the column where “node ID” is N103 and “transition destination node ID” is N104 is “Roppongi”. " Similarly, the detection word in the column where “node ID” is N103 and “transition destination node ID” is N105 is “other”. As described above, the dialogue information storage unit 121 stores the transition destination node ID, the transition probability, and the like in association with the node ID. Further, the dialogue information storage unit 121 may store response information in association with the node ID. Then, the dialogue information storage unit 121 may store an advertisement ID described later in association with the node ID of the node related to the advertisement, and refer to response information associated with the corresponding advertisement ID. Note that the dialogue information storage unit 121 may store any format as long as it can store information about the determination tree as shown in FIG. For example, the dialogue information storage unit 121 may store the detection node and the operation node separately.

(Advertisement information storage unit 122)
The advertisement information storage unit 122 according to the embodiment stores information related to advertisements. For example, the advertisement information storage unit 122 stores response information regarding advertisement nodes, related information, and a list as information regarding advertisements.

  FIG. 6 illustrates an example of the advertisement information storage unit 122 according to the embodiment. As illustrated in FIG. 6, the advertisement information storage unit 122 includes items such as “response information ID”, “response information”, “node ID”, “related information ID”, and “related information”.

  “Response information ID” indicates identification information for identifying response information related to the advertisement. “Response information” indicates response information regarding the advertisement identified by the response information ID. “Node ID” indicates identification information for identifying the node corresponding to the response information ID in the determination tree shown in FIG. “Related information ID” indicates identification information for identifying related information related to the response information. “Related information” indicates related information identified by the related information ID.

  The example shown in FIG. 6 has shown the information regarding the advertisement output by the determination tree shown in FIG. That is, the example shown in FIG. 6 corresponds to the node related to the advertisement in the determination tree shown in FIG. 2, and the response information related to the four nodes N106, N107, N111, and N112 which are the nodes related to the advertisement, the related information, etc. Show. Specifically, in the example illustrated in FIG. 6, the response information corresponding to the response information ID “AI1” is “I like curry at store A”, and the node ID corresponding to AI1 is “N106”. Show. Further, the related information ID corresponding to AI1 is “RI11”, and the related information corresponding to AI1 is “www.a-curry. As described above, the advertisement information storage unit 122 stores the response information and the related information in association with each other. Note that the advertisement information storage unit 122 may store a plurality of pieces of related information in association with one response information. For example, the advertisement information storage unit 122 may store, for example, related information of RI12 in addition to RI11 as related information corresponding to AI1. Moreover, you may memorize | store the frequency | count of transmission included in the log information acquired from the advertising log server 50 mentioned later for every relevant information, and the advertisement log server 50 mentioned later.

  The control unit 130 includes, for example, a CPU (Central Processing Unit), an MPU (Micro Processing Unit), or the like that stores various programs (corresponding to an example of an information processing program) stored in a storage device inside the response generation server 100 in the RAM. This is realized by being executed as a work area. The control unit 130 is realized by an integrated circuit such as an application specific integrated circuit (ASIC) or a field programmable gate array (FPGA).

  As shown in FIG. 4, the control unit 130 includes an acquisition unit 131, a selection unit 132, a transmission unit 133, and an update unit 134, and realizes or executes an automatic dialog function and operation described below. . Note that the internal configuration of the control unit 130 is not limited to the configuration illustrated in FIG. 4, and may be other configurations as long as the configuration performs an automatic dialogue described later. In addition, the connection relationship between the processing units included in the control unit 130 is not limited to the connection relationship illustrated in FIG. 4, and may be another connection relationship.

(Acquisition part 131)
The acquisition unit 131 acquires the number of times the advertisement is designated by the user. Specifically, the acquisition unit 131 acquires, from the advertisement log server 50, log information including the display count of the Web page displayed based on the specified information, that is, the number of times the advertisement is specified by the user. Note that the log information acquired by the acquisition unit 131 may include the related information ID and the number of display times associated with each related information ID. Further, the acquisition unit 131 may acquire log information from the advertisement log server 50 every predetermined period such as every other day. Utterance information related to the user's utterance is acquired from the user terminal 10 as an input message. In addition, the acquisition unit 131 acquires, from the speech synthesis server 30, intermediate notation of response information generated by the speech synthesis server 30 performing speech synthesis processing for synthesizing speech from text data of response information.

(Selection unit 132)
The selection unit 132 selects response information corresponding to the utterance information acquired by the acquisition unit 131 based on the determination tree. Specifically, when selecting response information corresponding to utterance information, the selection unit 132 selects response information based on the transition probability associated with the transition from the detection node to the action node in the determination tree.

(Transmitter 133)
The transmission unit 133 transmits, as response information selected by the selection unit 132, intermediate notation for voice output, text data for display, and the like to the user terminal 10. The user terminal 10 that has received the response information outputs the response voice using the intermediate notation for voice output and displays the response content using the display text data included in the response information.

  In addition, the transmission unit 133 transmits the text data of the response information selected by the selection unit 132 to the speech synthesis server 30. The speech synthesis server 30 that has received the text data of the response information performs speech synthesis processing for synthesizing speech and generates an intermediate description of the response information.

  When there is related information related to the response information selected by the selection unit 132, the transmission unit 133 transmits related information that can be specified to the user terminal 10. Then, the user terminal 10 displays the received related information so that it can be designated.

  Here, FIG. 7 shows an example of a dialog display page displayed on the user terminal 10 that has received the response information from the response generation server 100. The user terminal 10 displays the content of the dialogue with the dialogue agent system 1 as character information on the dialogue display page W10. The character information posted in the right balloon in the dialog display page W10 corresponds to the user's speech, and the character information posted in the left balloon in the dialog display page W10 is the user based on the response information received from the response generation server 100. This corresponds to the output of the terminal 10. That is, the user terminal 10 outputs a sound based on the response information when the left balloon in the dialogue display page W10 is displayed.

  Next, the correspondence between the dialogue display page shown in FIG. 7 and the determination tree shown in FIG. 2 will be described based on the following processing. First, when receiving the user's utterance “delicious curry” from the user, the user terminal 10 displays “delicious curry” in the dialogue display page W10 of FIG. At this time, the acquisition unit 131 acquires utterance information related to the user's utterance “delicious curry” from the user terminal 10. Then, the selection unit 132 determines that N101 “curry” detection in the determination tree of FIG. 2 has been performed, based on the transition probability from the three nodes N102, N103, and N106 connected to N101. Select the transition node. Specifically, in the determination tree of FIG. 2, each node of N102, N103, and N106 is selected with a probability of 33%. Here, if the selection unit 132 selects N103, the transmission unit 133 transmits response information corresponding to N103 to the user terminal 10. Then, the user terminal 10 displays “where is curry? Roppongi?” Next to “delicious curry” which is the user's utterance in the dialogue display page W10 of FIG. Output audio.

  Subsequently, when the user terminal 10 accepts the user's utterance “NO, etc.” from the user, the user terminal 10 continues to “Where curry? Roppongi?” In the dialogue display page W10 of FIG. Is displayed. At this time, the acquisition unit 131 acquires utterance information regarding the user's utterance “No, other” from the user terminal 10. Then, the selection unit 132 assumes that N105 ““ other ”detection” in the determination tree of FIG. 2 has been performed, and makes a transition based on the transition probability from the two nodes N105 and N113 of the N105. Select. Specifically, in the determination tree of FIG. 2, each of the nodes N108 and N113 is selected with a probability of 50%. Here, if the selection unit 132 selects N108, the transmission unit 133 transmits response information corresponding to N108 to the user terminal 10. Then, the user terminal 10 displays “What kind of curry do you like?” Following “No, other” as the user's utterance in the dialogue display page W10 of FIG. 7 and “What kind of curry do you like?” Is output.

  Further, when the user terminal 10 accepts the user's utterance “Kama Curry” from the user, “What kind of curry do you like?” In the dialogue display page W10 of FIG. 7 is displayed. To do. At this time, the acquisition unit 131 acquires utterance information related to the user's utterance “Kama Curry” from the user terminal 10. Then, the selection unit 132 assumes that N109 “Kima” detection in the determination tree of FIG. 2 has been performed, and makes a transition based on the transition probability from the two nodes N109 and N112 of N109. Select. Specifically, in the determination tree of FIG. 2, each of the nodes N111 and N112 is selected with a probability of 50%. Here, if the selection unit 132 selects N111, the transmission unit 133 transmits response information corresponding to N111 to the user terminal 10. Then, the user terminal 10 displays “I like Kema curry at store C” in the dialogue display page W10 in FIG. “I like Kema ’s Kema curry”.

  Here, since the response information related to the advertisement node “I like the key C curry at store C” is output by the user terminal 10, the transmission unit 133 can specify the response information selected by the selection unit 132. The related information is transmitted to the user terminal 10. Then, the user terminal 10 displays the received related information so that it can be designated. Specifically, the user terminal 10 that has received the related information includes the URL (Uniform Resource Locator) of the Web page of the store C, which is the related information, in the response R10, which is the response information “I like Kama ’s keyma curry”. ) Can be specified. “Specifiable” here means that a hyperlink is set, and the user touches the surface on which the related information of the user terminal 10 is displayed with a finger, so that the Web page of the specified store C is displayed. Is mentioned. The display of the related information is not limited to the above example, and may be displayed at a position away from the response information, for example, displayed at the bottom of the screen.

  Then, when the user designates the URL of the Web page of store C, which is related information displayed in the interactive display page W10 of the user terminal 10, for example, the Web page W20 of store C as shown in FIG. Is displayed. Specifically, when the user designates the URL of the Web page of the store C, which is related information displayed in the interactive display page W10 of the user terminal 10, designation information related to designation of the related information is transmitted to the advertisement log server 50. The Then, the advertisement log server 50 that has received the specified information acquires a Web page corresponding to the specified related information by an appropriate means, and transmits it to the user terminal 10. At this time, the advertisement log server 50 stores the display count of the Web page displayed based on the designation information. Thereafter, the user terminal 10 displays the acquired web page.

(Update unit 134)
The update unit 134 updates the evaluation related to the voice conversation related to the advertisement specified by the user based on the number of times the advertisement is specified by the user included in the log information acquired by the acquisition unit 131. Specifically, the update unit 134 determines the number of display times of the Web page corresponding to the related information acquired from the advertisement log server 50 by the acquisition unit 131, that is, log information including the number of times the advertisement is designated by the user. The transition probability, which is the probability based on the evaluation regarding the voice conversation, is updated. For example, the update unit 134 updates the transition probability so that the probability of transition to a node of response information related to the related information increases as the number of display times increases. As shown in the example of FIG. 7, when a lot of related information related to N111 “I like Kama ’s Kima curry” is specified and the Web page corresponding to the related information has a display count exceeding a predetermined threshold, The transition probability is updated so that the probability that the flow of dialogue is directed to N111 is increased.

  In addition, the update unit 134 updates, for example, the transition probability of transition to a node that has transitioned to the selected node. In the determination tree shown in FIG. 2, for example, when nodes are selected in the order of N101-N103-N105-N108-N109-N111, the update unit 134 transitions from N109 to N111, and transitions from N05 to N108. The transition probability and the transition probability for transition from N101 to N103 are updated high. Specifically, in the example illustrated in FIG. 8, the updating unit 134 sets the transition probability of transition from N109 to N111 from “0.5” to “0.75”, and the transition probability of transition from N105 to N108 as “0”. .5 ”to“ 0.75 ”and the transition probability of transition from N101 to N103 is updated high from“ 0.33 ”to“ 0.5 ”. As the transition probability is updated, the transition probability to another node that makes a transition from the same detection node is also updated. Specifically, the transition probability of transition from N109 to N112 is changed from “0.5” to “0.25”, the transition probability of transition from N105 to N113 is changed from “0.5” to “0.25”, and The transition probabilities of transition from N101 to N102 and N106 are updated low from “0.33” to “0.25”, respectively. As a result, the response generation server 100 can automatically evaluate the dialogue based on the determination tree, and improve the effectiveness of the advertisement by facilitating the generation of a highly interactive dialogue flow. When there are a plurality of other transition nodes, such as N101, the transition probability may be updated in accordance with the transition probability ratio of each node.

[3. (Automatic dialogue procedure)
Next, the procedure of the automatic dialogue process including the transition probability update process according to the embodiment will be described with reference to FIG. FIG. 9 is a sequence diagram illustrating an automatic dialog processing procedure including a transition probability update process according to the embodiment.

  As illustrated in FIG. 9, the response generation server 100 receives utterance information related to the user's utterance as an input message from the user terminal 10 (step S <b> 101). Then, the response generation server 100 selects response information based on the transition probability associated with the transition from the detection node to the operation node in the determination tree, and transmits the selected response information to the user terminal 10 (step S102). At this time, if there is related information related to the selected response information, the response generation server 100 transmits the related information to the user terminal 10 (step S103).

  Then, the response information is output by the user terminal 10, and the related information is displayed on the user terminal 10 so that it can be designated (step S104). Thereafter, when the related information displayed so as to be specifiable on the user terminal 10 is designated by the user, designation information related to designation of the related information by the user is transmitted to the advertisement log server 50 (step S105).

  Subsequently, the advertisement log server 50 that has acquired the designation information transmits a Web page corresponding to the related information to the user terminal 10 (Step S106), and the user terminal 10 displays the acquired Web (Step S107). At this time, the advertisement log server 50 counts the number of display times of the Web page displayed based on the designation information.

  Next, the response generation server 100 receives log information including the display count of the Web page displayed based on the specified information every predetermined period, for example, every other day, from the advertisement log server 50 (step S108). The response generation server 100 that has acquired the log information updates the transition probability based on the log information (step S109).

[4. (Modification)
The interactive agent system 1 according to the above-described embodiment may be implemented in various different forms other than the above-described embodiment. Therefore, in the following, an embodiment in which the above-described dialog agent system 1 is modified will be described.

[4-1. Update by specified number of times / displayed number of times)
In the above embodiment, the response generation server 100 updates the transition probability based only on the number of times the related information displayed on the user terminal 10 is specified. However, the response generation server may update the transition probability based on the ratio of the designated number of times that the related information is displayed. As a result, the transition probability is updated based on the probability that the related information is specified, not the number of times that the related information is specified, so that it is easier to generate a flow of conversation with a high probability that the related information is specified. Thus, the effect of the advertisement can be improved.

[4-2. (Scope of update)
In the above embodiment, the response generation server 100 updates the transition probability up to N101 at the top of the determination tree shown in FIG. However, the response generation server may update the transition probability regarding the nodes up to a predetermined range.

  Here, FIG. 10 shows an example of the determination tree after the transition probability of the determination tree shown in FIG. 2 is updated. In the example illustrated in FIG. 10, when a large amount of related information related to N111 is specified and the Web page corresponding to the related information has a display count that exceeds a predetermined threshold, the response generation server displays the determination tree illustrated in FIG. 2. The transition probability is updated so that the flow of dialogue is directed to N111.

  For example, the response generation server may update the transition probabilities of the top two detection nodes. Specifically, the response generation server 100 changes the transition probability of transition from the one higher detection node N109 to N111 from “0.5” to “0.75”, and the two higher detection nodes N105 to N108. The transition probability to be updated is increased from “0.5” to “0.75”. As the transition probability is updated, the transition probability to another node that makes a transition from the same detection node is also updated. Specifically, the transition probability of transition from N109 to N112 is updated from “0.5” to “0.25”, and the transition probability of transition from N105 to N113 is updated from “0.5” to “0.25”. To do. As a result, the response generation server 100 can automatically evaluate the dialogue based on the determination tree, and improve the effectiveness of the advertisement by facilitating the generation of a highly interactive dialogue flow. Here, unlike the example illustrated in FIG. 8, the example illustrated in FIG. 10 does not update the transition probability of transition from the three higher detection nodes N101 to N103. Thereby, it is possible to improve the effectiveness of the advertisement while suppressing a reduction in the diversity of dialogue.

[4-3. (Renewal size)
Further, the response generation server may update the transition probability related to the node according to the number of transitions until the Web page transitions to the node related to the response information corresponding to the related information, which is the display count exceeding the predetermined threshold.

  Here, FIG. 11 shows an example of the determination tree after updating the transition probability of the determination tree shown in FIG. In the example shown in FIG. 11, when a lot of related information related to N111 is specified and the Web page corresponding to the related information has a display count exceeding a predetermined threshold, the response generation server displays the determination tree shown in FIG. 2. The transition probability is updated so that the flow of dialogue is directed to N111.

  For example, the response generation server updates the transition probability for transition from N109 to N111, the transition probability for transition from N05 to N108, and the transition probability for transition from N101 to N103, for example. Specifically, in the example illustrated in FIG. 11, the response generation server updates the transition probability of transition from N109 to N111 high from “0.5” to “0.75”. That is, in the response generation server, the transition probability of transition from N109 to N111 is 1.5 times, and the fluctuation rate is 150%. Then, the transition probability of transition from N105 to N108 is updated from “0.5” to “0.6”. That is, in the response generation server, the transition probability of transition from N105 to N108 is 1.2 times, and the fluctuation rate is 120%. Further, the transition probability of transition from N101 to N103 is updated to a high value from “0.33” to “0.35”. That is, in the response generation server, the transition probability of transition from N101 to N103 is 1.05 times, and the variation rate is 105%. Thereby, the response generation server 100 can improve the effect of the advertisement while suppressing a reduction in the diversity of the dialogue. Further, the response generation server may calculate the variation rate of the transition probability of each detection node using a predetermined propagation function. In this case, if the variation rate of the transition probability of the detection node that makes a direct transition to N111 in the above example is the predetermined value x, the response generation server sets the variation rate of the upper detection node to y (x> y) Further, a propagation function may be used in which the variation rate of the higher detection node is z (y> z), and the variation rate decreases as the number of transitions increases from the first node.

[4-4. Judgment information)
In the above embodiment, a determination tree as shown in FIG. 2 is used as the determination information. However, the determination information is not limited to a tree shape such as a determination tree, and may be, for example, a node map shape. When such node-map determination information is used, the root node (corresponding to N101 in the example shown in FIG. 2) is the node at the start of the conversation.

[4-5. (Target of evaluation)
Moreover, in the said embodiment, the response production | generation server 100 updates a transition probability only based on the frequency | count that the relevant information displayed on the user terminal 10 was designated. However, the response generation server updates the transition probability based on so-called trendy words such as words according to the season, buzzwords used in TV and SNS (Social Networking Service), and search query top ranking words. May be. For example, the response generation server 100 may update the transition probability based on the number of times an advertisement related to the content of the voice conversation is designated within a predetermined time by the user using the conversation agent system 1. In addition, the response generation server 100 may update the transition probability based on the number of times an advertisement related to the content of the voice conversation is designated within a predetermined time for a user who is in a login state. In the above-described embodiment, the response generation server 100 updates the transition probability according to the number of times related information displayed on the user terminal 10 is specified. However, the embodiment is not limited to this. For example, the response generation server 100 changes according to the designated number of related information displayed on an arbitrary display device operating in cooperation with the user terminal 10. The probability may be updated. Also, the response generation server 10 outputs the related information by voice in the dialogue with the user, and when the user makes a response to specify the related information, the response generation server 10 assumes that the related information is selected, Updates may be made. In addition, the response generation server 10 selects whether the related information is selected (for example, whether the displayed related information is selected, whether the related information output by voice is selected, or whether the related information is selected by voice) Depending on whether it is selected by the operation, weighting may be performed when the transition probability is updated. As described above, the response generation server 100 may not be based on the screen display of the related information.

[4-6. Device configuration〕
Further, the response generation server 100 illustrated in FIG. 4 may not include the dialogue information storage unit 121 and the advertisement information storage unit 122. Specifically, the response generation server 100 may be connected to a database server that holds the dialogue information storage unit 121 and the advertisement information storage unit 122. In the above embodiment, the interactive agent system 1 includes the response generation server 100 and the advertisement log server 50. However, the response generation server 100 and the advertisement log server 50 are formed as one device. May be.

[4-7. effect〕
As described above, the response generation server 100 according to the embodiment includes the acquisition unit 131 and the update unit 134. The acquisition unit 131 acquires the number of times the information displayed on the screen is designated by the user. Based on the number of times acquired by the acquisition unit 131, the update unit 134 updates the evaluation regarding the voice conversation related to the information specified by the user. For example, the acquisition unit 131 acquires the number of times an advertisement is specified. Then, the update unit 134 updates the evaluation related to the voice conversation related to the advertisement designated by the user based on the number of times acquired by the acquisition unit 131.

  Thereby, the response generation server 100 according to the embodiment can improve the quality of the conversation. For example, since the response generation server 100 updates the evaluation regarding the voice conversation according to the number of times the advertisement is designated, it is possible to perform many conversations that can improve the effect of the advertisement.

  In addition, the response generation server 100 according to the embodiment includes a transmission unit 133. The transmission unit 133 is a probability that is selected as a response to the utterance information related to the user's utterance using the voice dialogue, and can be specified related to the response information selected based on the transition probability that is a probability based on the evaluation concerning the voice dialogue. Related information is transmitted to the user terminal 10. The update unit 134 updates the transition probability corresponding to the selected response information according to the number of times the related information transmitted by the transmission unit 133 is designated by the user.

  Accordingly, the response generation server 100 according to the embodiment updates the transition probability corresponding to the response information selected according to the number of times the related information related to the response information is specified, thereby improving the advertising effect. Can do.

  Further, in the response generation server 100 according to the embodiment, the update unit 134 selects the response selected in accordance with the ratio of the number of times the related information is designated out of the number of times the related information related to the selected response information is displayed. The transition probability corresponding to the information is updated.

  Thereby, since the response generation server 100 according to the embodiment updates the transition probability based on the probability that the related information is specified, not the number of times the related information is specified, the probability that the related information is specified By making it easier to generate a high flow of dialogue, the effectiveness of advertising can be improved.

  Further, in the response generation server 100 according to the embodiment, the transmission unit 133 relates to response information corresponding to the node selected based on the determination information in which the nodes related to the utterance information or the response information are associated with the transition probability. Related information that can be specified is transmitted to the user terminal. The update unit 134 updates the transition probability of transition to the selected node according to the number of times the related information transmitted by the transmission unit 133 is designated by the user.

  Accordingly, the response generation server 100 according to the embodiment updates the transition probability corresponding to the response information selected according to the number of times the related information related to the response information is specified, thereby improving the advertising effect. Can do.

  Further, in the response generation server 100 according to the embodiment, the update unit 134 transitions to a node in a predetermined range including the selected node according to the number of times the related information transmitted by the transmission unit 133 is designated by the user. Update transition probabilities.

  Thereby, the response generation server 100 according to the embodiment can improve the effect of the advertisement while suppressing the diversity of the conversation from being lowered.

  Further, in the response generation server 100 according to the embodiment, the update unit 134 transitions to a node that has transitioned to the selected node according to the number of times related information transmitted by the transmission unit 133 is designated by the user. Update transition probabilities.

  Thereby, the response generation server 100 according to the embodiment can improve the effect of the advertisement while suppressing the diversity of the conversation from being lowered.

  Further, in the response generation server 100 according to the embodiment, the updating unit 134 updates the transition probability of transitioning to the node that has transitioned to the selected node according to the number of transitions until transition to the selected node. Change the percentage.

  Thereby, the response generation server 100 according to the embodiment can improve the effect of the advertisement while suppressing the diversity of the conversation from being lowered.

[4-7. program〕
The information processing apparatus according to each embodiment described above is realized by a computer 1000 configured as shown in FIG. 12, for example. FIG. 12 is a hardware configuration diagram illustrating an example of a computer 1000 that implements the functions of each information processing apparatus. The computer 1000 includes a CPU 1100, RAM 1200, ROM 1300, HDD 1400, communication interface (I / F) 1500, input / output interface (I / F) 1600, and media interface (I / F) 1700.

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

  The HDD 1400 stores programs executed by the CPU 1100, data used by the programs, and the like. The communication interface 1500 receives data from other devices via the communication network 60 and sends the data to the CPU 1100, and transmits data generated by the CPU 1100 to other devices via the communication network 60.

  The CPU 1100 controls an output device such as a display and a printer and an input device such as a keyboard and a mouse via the input / output interface 1600. The CPU 1100 acquires data from the input device via the input / output interface 1600. In addition, 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 it to the CPU 1100 via the RAM 1200. The CPU 1100 loads such a 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 an MO (Magneto-Optical disk), a tape medium, a magnetic recording medium, or a semiconductor memory. Etc.

  For example, when the computer 1000 functions as the response generation server 100 according to the embodiment, the CPU 1100 of the computer 1000 implements the function of the control unit 130 by executing a program loaded on the RAM 1200. The HDD 1400 stores data in the storage unit 120 in the information processing apparatus according to the embodiment. The CPU 1100 of the computer 1000 reads these programs from the recording medium 1800 and executes them, but as another example, these programs may be acquired from other devices via the communication network 60.

[4-8. Others]
In addition, among the processes described in the above embodiments, all or a part of the processes described as being automatically performed can be manually performed, or the processes described as being manually performed All or a part of the above can be automatically performed by a known method. In addition, the processing procedures, specific names, and information including various data and parameters shown in the document and drawings can be arbitrarily changed unless otherwise specified. For example, the various types of information illustrated in each drawing is not limited to the illustrated information.

  Further, each component of each illustrated apparatus is functionally conceptual, and does not necessarily need to be physically configured as illustrated. In other words, the specific form of distribution / integration of each device is not limited to that shown in the figure, and all or a part thereof may be functionally or physically distributed or arbitrarily distributed in arbitrary units according to various loads or usage conditions. Can be integrated and configured. For example, the acquisition unit 131 and the transmission unit 133 illustrated in FIG. 4 may be integrated. In addition, the above-described voice recognition process for recognizing the voice is realized by the user terminal 10, and each function described above is realized by transmitting the result of the voice recognition process executed by the user terminal 10 to the response generation server 100. May be.

  In the above-described embodiment, an example has been described in which the quality of a conversation in which the advertisement is displayed is evaluated according to the number of times the advertisement is selected. However, the embodiment is not limited to this. In other words, the response generation server 100 has information that is displayed by the user terminal 10 in the dialogue, a function that evaluates “Like”, for example, and a function that can be shared with other users, such as retweets. The dialogue may be evaluated according to the number of times evaluation or sharing is performed by such a function. In addition, for example, the response generation server 100 includes other information displayed on the user terminal 10, for example, an advertisement such as a banner advertisement displayed together with a web page, information posted on a social network service or a blog, The quality of the dialogue can be evaluated according to the evaluation of the information related to the dialogue service provided by the response generation server 100.

  Further, the response generation server 100 may evaluate the quality of the dialog using a log associated with the user who has performed the dialog. For example, the response generation server 100 logs (ie, the user's location, the content viewed by the user, the app acquired by the user, the product purchased by the user, the user's attributes, the operation performed by the user, etc.) Life log). Then, the response generation server 100 is used when the user visits a store related to the advertisement output in the dialogue, when the user purchases a product or an application related to the advertisement, when the user views content related to the advertisement, or the like. May update the transition probability as the quality of the dialogue including the advertisement is good.

  In the case of executing such processing, for example, the acquisition unit 131 acquires the life log of the user who performed the conversation from the user terminal 10 or a log collection server that collects the life log. Then, the update unit 134 updates the transition probability according to the action performed by the user after the conversation. For example, the update unit 134 responds to such advertisement when a product related to the advertisement output during the conversation is viewed or purchased, or when a user visits a store related to the advertisement. Increase the transition probability to the advertising node. That is, the update unit 134 updates the evaluation regarding the dialogue based on the user's behavior history after the dialogue. In addition, the acquisition unit 131 and the update unit 134 may execute any other process described above. In addition, the update unit 134 evaluates not only the action history after the voice conversation, but also, for example, according to the comparison result between the action history before the voice conversation and the action history after the voice conversation. It may be updated.

  In this way, the response generation server 100 acquires the user's action history, and updates the evaluation related to the conversation performed with the user based on the acquired action history. For this reason, the response generation server 100 can improve the quality of the conversation.

  Moreover, each embodiment mentioned above can be combined suitably in the range which does not contradict a process content.

  As described above, some of the embodiments of the present application have been described in detail with reference to the drawings. However, these are merely examples, and various modifications based on the knowledge of those skilled in the art, including the aspects described in the disclosure line of the invention. It is possible to implement the present invention in other forms with improvements.

  In addition, the “section (module, unit)” described above can be read as “means” or “circuit”. For example, the acquisition unit can be read as acquisition means or an acquisition circuit.

DESCRIPTION OF SYMBOLS 1 Dialogue agent system 10 User terminal 20 Speech recognition server 30 Speech synthesis server 40 Service API server 50 Advertisement log server 100 Response generation server 120 Storage part 131 Acquisition part 132 Selection part 133 Transmission part 134 Update part

Claims (9)

An acquisition unit for acquiring the number of times the advertisement displayed on the screen is designated by the user;
An update unit that updates an evaluation related to the voice dialogue related to the advertisement specified by the user based on the number of times acquired by the acquisition unit;
Specifyable related information related to response information selected based on transition probability, which is a probability selected as a response to utterance information related to a user's utterance using the voice dialogue A transmission unit for transmitting to the user terminal;
Equipped with a,
The update unit
Depending on the number of times that the relevant information transmitted by the transmission unit is designated by the user, the information processing apparatus, characterized in that to update the transition probability corresponding to the selected response information. The update unit
Updating a transition probability corresponding to the selected response information according to a ratio of the number of times the related information is designated out of the number of times the related information related to the selected response information is displayed;
The information processing apparatus according to claim 1 . The transmitter is
Transmitting, to the user terminal, related information that can be specified related to the response information corresponding to the node selected based on the determination information associated with the transition probability between the nodes related to the utterance information or the response information,
The update unit
Updating the transition probability of transitioning to the selected node according to the number of times the related information transmitted by the transmission unit is designated by the user;
The information processing apparatus according to claim 1 or claim 2, characterized in that. The update unit
Update the transition probability of transitioning to a predetermined range of nodes including the selected node according to the number of times the related information transmitted by the transmission unit is designated by the user.
The information processing apparatus according to claim 3 . The update unit
According to the number of times the related information transmitted by the transmission unit is designated by the user, the transition probability of transitioning to the node that has transitioned to the selected node is updated.
The information processing apparatus according to claim 4 . The update unit
According to the number of transitions until the transition to the selected node, change the rate of updating the transition probability to transition to the node that has transitioned to the selected node,
The information processing apparatus according to claim 4 or claim 5, characterized in that. An acquisition unit for acquiring the action history of the user who performed the voice dialogue;
An update unit that updates the evaluation related to the voice dialogue based on the action history acquired by the acquisition unit;
Specifyable related information related to response information selected based on transition probability, which is a probability selected as a response to utterance information related to a user's utterance using the voice dialogue A transmission unit for transmitting to the user terminal;
Equipped with a,
The update unit
Depending on the number of times that the relevant information transmitted by the transmission unit is designated by the user, the information processing apparatus, characterized in that to update the transition probability corresponding to the selected response information. An information processing method executed by a computer,
An acquisition step of acquiring the number of times the advertisement displayed on the screen is designated by the user;
An update step of updating an evaluation of a voice dialogue related to the advertisement designated by the user based on the number of times acquired by the acquisition step;
Specifyable related information related to response information selected based on transition probability, which is a probability selected as a response to utterance information related to a user's utterance using the voice dialogue Transmitting to the user terminal,
Only including,
The update process includes:
An information processing method, wherein the related information transmitted by the transmission step is in accordance with the number of times specified by the user, to update the transition probability corresponding to the selected response information. An acquisition procedure to get the number of times the ad displayed on the screen is specified by the user,
An update procedure for updating an evaluation of a voice dialogue related to the advertisement specified by the user based on the number of times acquired by the acquisition procedure ;
Specifyable related information related to response information selected based on transition probability, which is a probability selected as a response to utterance information related to a user's utterance using the voice dialogue Sending procedure to send to the user terminal,
To the computer ,
The update procedure is:
The related information transmitted by the transmission procedure in accordance with the number of times specified by the user, the information processing program characterized by to update the transition probability corresponding to the selected response information.

Images