Relation visualization device, relation visualization method, and program

Abstract

To enable visualization of a relation between users based on their utterances.SOLUTION: A relation visualization device includes: an extraction unit that extracts an utterance message classified into a predetermined category from among utterance messages included in conversation data between users participating in a conversation; and a graph generation unit that generates a graph in which a relation between nodes representing a user who sent the extracted utterance messages and nodes representing a user who received the extracted utterance messages is visualized using a directional edge.SELECTED DRAWING: Figure 1

Description

Technical Field

[0001] The present disclosure relates to a relationship visualization device, a relationship visualization method, and a program.

Background Art

[0002] As a related art, Patent Document 1 discloses an organizational communication analysis support device. The organizational communication analysis support device supports the analysis of communication logs using information systems used for organizational communication, such as electronic bulletin boards and mailing lists. Patent Document 1 describes performing network analysis, which is one of the analysis methods of mathematical social science, on message logs. Information for network analysis is obtained from relationships between utterances or relationships between speakers.

[0003] The organizational communication analysis support device creates relationship information between speakers in network analysis. The relationship information between speakers includes information indicating the relationship of how much one speaker has spoken to another speaker. The relationship information includes the user ID of the source, the user ID of the destination of the utterance, and the frequency of the utterance directed from the source to the destination. This relationship information can be automatically calculated from the message log.

[0004] The organizational communication analysis support device calculates, for example, the ratio of the number of one-way speech relationships to the total number of theoretically possible speech relationships as "density". A one-way speech relationship is a relationship where there is at least one speech directed from one of two speakers to the other. Also, when there is an exchange of two-way speech between speakers, the organizational communication analysis support device counts the relationship between those speakers and calculates the ratio of the count result to the total number of theoretically possible relationships as "direct connectivity".

[0005] The organizational communication analysis support device can display a network diagram of the relationships between speakers as an analysis result. The network diagram includes nodes representing the speakers (i.e., participating users) of a mailing list (ML). When there is a speech from a first participating user to a second participating user on the ML, the organizational communication analysis support device displays, in the network diagram, an edge with an arrow from the first participating user to the second participating user. When there is a two-way exchange of speech between the first participating user and the second participating user, the organizational communication analysis support device displays an edge with a two-way arrow between the first participating user and the second participating user.

Prior Art Documents

Patent Documents

[0006]

Patent Document 1

Summary of the Invention

Problems to be Solved by the Invention

[0007] By looking at the network diagram created in Patent Document 1, the analyst can know which participating users have exchanged speech. However, in the network diagram created in Patent Document 1, although the direction of the speech during a certain period is known, the content of the speech is not known. For example, when a user makes a one-way speech to another user, it is impossible to read from the network diagram whether the user is simply communicating a matter to be contacted or giving some kind of order. Therefore, the network diagram generated in Patent Document 1 is not suitable for analyzing the relationships between users, for example, the organizational structure of the organization to which the users belong.

[0008] One of the objects of the present disclosure is to provide a relationship visualization device, a relationship visualization method, and a program capable of visualizing the relationships between users according to the content of the speech.

Means for Solving the Problems

[0009] The program according to the first aspect of the present disclosure causes a computer to execute a process including extracting, from among speech messages included in conversation data between users participating in a conversation, speech messages classified into a predetermined classification, and generating a graph in which the relationship between a node representing the user who is the source of the extracted speech message and a node representing the user who is the destination of the extracted speech message is visualized using a directed edge.

[0010] The relationship visualization method according to the second aspect of the present disclosure includes extracting, from among speech messages included in conversation data between users participating in a conversation, speech messages classified into a predetermined classification, and generating a graph in which the relationship between a node representing the user who is the source of the extracted speech message and a node representing the user who is the destination of the extracted speech message is visualized using a directed edge.

[0011] The relationship visualization device according to the third aspect of the present disclosure includes an extraction unit that extracts, from among speech messages included in conversation data between users participating in a conversation, speech messages classified into a predetermined classification, and a graph generation unit that generates a graph in which the relationship between a node representing the user who is the source of the extracted speech message and a node representing the user who is the destination of the extracted speech message is visualized using a directed edge.

Advantages of the Invention

[0012] The relationship visualization device, relationship visualization method, and program according to the present disclosure can visualize the relationship between users according to the speech content.

Brief Description of the Drawings

[0013]

Figure 1

Figure 2

Figure 3

Figure 4

Figure 5

Figure 6

Figure 7

Figure 8

Mode for Carrying Out the Invention

[0014] Prior to the description of the embodiments of the present disclosure, the outline of the present disclosure will be described. FIG. 1 shows a schematic configuration example of a relationship visualization device according to the present disclosure. The relationship visualization device 10 includes an extraction unit 11 and a graph generation unit 12. The extraction unit 11 analyzes conversation data between users participating in the conversation. The extraction unit 11 extracts the speech messages classified into a predetermined classification from the speech messages included in the conversation data.

[0015] The graph generation unit 12 generates a graph having users participating in the conversation as nodes. The graph generation unit 12 generates a graph in which the relationship between the node representing the user who is the source of the speech message extracted by the extraction unit 11 and the node representing the user who is the destination of the extracted speech message is visualized using a directed edge.

[0016] In the present disclosure, the graph generation unit 12 generates a graph in which a node representing a user who is the source of a speech message classified into a predetermined classification and extracted by the extraction unit 11 and a node representing a user who is the destination of the speech message are connected by a directed edge. In other words, the graph generation unit 12 visualizes the relationship between a user who made a speech classified into a predetermined classification and a user who received the speech using a directed graph. The relationship visualization device 10 according to the present disclosure can visualize the relationship between users according to the content of the speech by appropriately setting a predetermined classification according to the relationship between the users to be analyzed.

[0017] Hereinafter, embodiments of the present disclosure will be described in detail with reference to the drawings. Note that the following description and drawings are appropriately omitted and simplified for clarity of explanation. Also, in each drawing, the same elements and similar elements are denoted by the same reference numerals, and redundant descriptions are omitted as necessary.

[0018] An embodiment will be described. FIG. 2 shows a schematic configuration of a relationship visualization device according to the present disclosure. The relationship visualization device 100 includes a conversation history acquisition unit 101, a classification unit 102, a node extraction unit 103, a classification result storage unit 104, an extraction unit 105, and a graph generation unit 106. Physically, the relationship visualization device 100 can be configured as, for example, an information processing device having one or more processors and one or more memories, or a server device. At least a part of the functions of each unit in the relationship visualization device 100 can be realized by one or more processors operating according to a program read from one or more memories. The relationship visualization device 100 corresponds to the relationship visualization device 10 shown in FIG. 1.

[0019] In this embodiment, for example, a messaging application is installed on a portable communication device such as a smartphone used by a user. The user uses the messaging application to conduct a two-party conversation or a group chat. The user may also send a message to another user using email. For example, the user belongs to an organization and has a conversation with another user belonging to the same organization using the messaging application.

[0020] The conversation history acquisition unit 101 acquires conversation history data including conversation data between users participating in a conversation. The conversation history acquisition unit 101 acquires, for example, conversation history data stored in the portable communication device, that is, data indicating the history of conversations, from the portable device used by the user. Alternatively, the conversation history acquisition unit 101 may acquire conversation history data from the server of a service provider that provides a messaging service. The conversation history acquisition unit 101 acquires conversation history data from, for example, the portable communication devices of a plurality of users.

[0021] The classification unit 102 classifies the words spoken by the speaker in the conversation between users, that is, the speech messages, into a plurality of classifications. The plurality of classifications include, for example, fact, greeting, question, command, instruction, and threat. The classification unit 102 may classify the messages by natural language processing (NLP) using artificial intelligence (AI), for example.

[0022] The node extraction unit 103 extracts the user who is the speaker and the user who is the recipient of the speech from the conversation history data. The user who is the speaker is also called the source user of the speech message. The user who is the recipient is also called the destination user of the speech message. The node extraction unit 103 extracts, for example, for each pair of users participating in the conversation, one of the user pairs as the source user of the speech message and the other as the destination user of the speech message.

[0023] When a single user sends a message to multiple users, for example, in a group chat, the node extraction unit 103 may extract each of the multiple users as the destination user of the speech message. Alternatively, when a single user sends a message to multiple users, the node extraction unit 103 may extract, as the destination user of the speech message, the user who replied to the sent message among the multiple users. Further, the node extraction unit 103 may identify the destination user of the speech message based on the relationship between the sent speech message and the content of the reply.

[0024] Also, when a single user sends a message to multiple users, the node extraction unit 103 may extract, as the destination user of the speech message, the user who showed a reaction among the multiple users. For example, the node extraction unit 103 may identify, as the user who reacted to the speech message, the user who read the speech message or the user who performed a predetermined operation such as pressing a stamp on the speech message. The node extraction unit 103 may extract, as the destination user of the speech message, for example, the user who reacted within a predetermined time from the transmission of the speech message.

[0025] The classification result storage unit 104 stores, for example, for each speech, the classification result of the speech classified by the classification unit 102 and the message sender and message recipient extracted by the node extraction unit 103. The classification result storage unit 104 is configured as a storage device such as a hard disk device or a Solid State Drive (SSD), for example. The classification result storage unit 104 does not necessarily have to be included in the relationship visualization device 100. The classification result storage unit 104 may be configured as cloud storage, for example.

[0026] The extraction unit 105 extracts speech messages classified into a predetermined classification from among the speech messages included in the conversation data between users. In the present embodiment, the extraction unit 105 extracts speech messages classified into commands, instructions, and threats from the classification result storage unit 104. The extraction unit 105 corresponds to the extraction unit 11 shown in FIG. 1.

[0027] The graph generation unit 106 generates a graph including nodes representing each of a plurality of users who participated in the conversation. In the graph generated by the graph generation unit 106, a node representing the user who is the source of the uttered message and a node representing the user who is the destination of the extracted uttered message are connected using a directed edge. The graph generated by the graph generation unit 106 is also called a directed graph. The graph generation unit 106 displays the generated directed graph on the display screen of the display device 150. The graph generation unit 106 corresponds to the graph generation unit 12 shown in FIG. 1.

[0028] The graph generation unit 106 includes a relationship determination unit 107. When there is an uttered message classified into a predetermined classification in a conversation between a user pair including a first user and a second user, the relationship determination unit 107 determines the upper and lower relationship of the positions between the first user and the second user. The relationship determination unit 107 determines the upper limit relationship according to whether the first user is the user who is the source of an uttered message classified into a predetermined classification such as an order or a threat, or whether the second user is the user who is the source of an uttered message classified into a predetermined classification such as an order or a threat.

[0029] The relationship determination unit 107 may count the number of utterances of the first user and the number of utterances of the second user for the uttered messages classified into a predetermined classification in the conversation between the user pair. That is, the relationship determination unit 107 may count the number of uttered messages classified into a predetermined classification in which the first user is the source user and the number of uttered messages classified into a predetermined classification in which the second user is the source user. The relationship determination unit 107 compares the number of utterances of the first user and the number of utterances of the second user. The relationship determination unit 107 may determine which of the first user and the second user is the user with the higher position based on the result of the comparison.

[0030] For example, the relationship determination unit 107 may determine that, among the first user and the second user, the user with a larger number of utterance messages classified into a predetermined classification in the conversation between the user pair is the user with a higher position. In the present embodiment, the extraction unit 105 extracts utterance messages classified into commands, instructions, and threats. In this case, the relationship determination unit 107 may determine, for example, which of the first user and the second user is the user with a stronger position of giving commands and which is the user with a weaker position of being commanded.

[0031] FIG. 3 shows a first example of a conversation between a user pair, that is, the utterance messages transmitted and received. The classification unit 102 classifies each utterance message into a plurality of classifications. Further, the node extraction unit 103 extracts or identifies, for each utterance message, the user who is the source of the utterance message and the user who is the destination of the utterance message. In the example shown in FIG. 3, user A, who is one of the user pair, has transmitted utterance messages classified into "threat" and "command" twice. On the other hand, user B, who is the other user of the user pair, has transmitted an utterance message classified into "instruction" once. In this case, the relationship determination unit 107 determines that user A is a user with a higher position than user B. In that case, in the graph generated by the graph generation unit 106, the node representing user A and the node representing user B are connected using an arrow pointing from the node representing user A to the node representing user B.

[0032] FIG. 4 shows a second example of a conversation between a user pair. In this example, user A has transmitted utterance messages classified into "threat" and "command" twice. On the other hand, user B has transmitted utterance messages classified into "instruction" and "threat" twice. When the number of utterance messages belonging to a predetermined classification is the same between the user pair as in this example, the relationship determination unit 107 may determine that user A and user B have an equal relationship. In that case, in the graph generated by the graph generation unit 106, the node representing user A and the node representing user B may be connected using a bidirectional arrow.

[0033] Note that even if the number of utterances of User A and the number of utterances of User B are not the same, when the difference in the number of utterances is within a predetermined value, the relationship determination unit 107 may regard the number of utterances as the same and determine that User A and User B are in an equal relationship. Alternatively, when the ratio of the number of utterances of one user to the total number of utterances of User A and User B is equal to or less than a predetermined ratio, the relationship determination unit 107 may determine that User A and User B are in an equal relationship. The predetermined ratio is set to, for example, 60%. For example, when there are 20 predetermined classification utterance messages between User A and User B, if the number of utterances of User A exceeds 12 times, the relationship determination unit 107 determines that User A has a higher position than User B. If neither User A nor User B exceeds 12 utterances, the relationship determination unit 107 may determine that User A and User B are equal.

[0034] Figure 5 shows a third example of a conversation between user pairs. In this example, User A has sent one utterance message classified as "greeting". User B has also sent one utterance message classified as "greeting". When no utterance messages belonging to a predetermined classification are transmitted or received between user pairs as in this example, the relationship determination unit 107 may determine that no hierarchical relationship occurs between User A and User B. In that case, the graph generation unit 106 does not connect the node representing User A and the node representing User B in the generated graph. In other words, no link is generated between the node representing User A and the node representing User B in the generated graph.

[0035] The auxiliary information storage unit 130 stores auxiliary information that can be used for the determination of the hierarchical relationship in the relationship determination unit 107. The auxiliary information includes, for example, relationship information among at least some of the users participating in the conversation. The auxiliary information may include information regarding the positions of the users participating in the conversation in the organization to which they belong. The auxiliary information may include, for example, in a group chat, information indicating the user who invited other users to the group chat and the user who was invited by other users. The auxiliary information can be created, for example, by collecting information publicly disclosed on a social networking service (SNS) and analyzing the collected information. The auxiliary information may be extracted from information described in natural language regarding the organization to which the users participating in the conversation belong. The auxiliary information may be generated manually.

[0036] The relationship determination unit 107 may weight the number of times of sending speech messages classified into a predetermined classification according to the type of auxiliary information related to each user for each user. For example, the relationship determination unit 107 uses the auxiliary information to determine whether there is an organizational hierarchical relationship between user A and user B. For example, assume that in the auxiliary information, it is shown that user A is the supervisor of user B. In that case, the number of speech messages belonging to a predetermined classification sent by user A to user B may be weighted with a weight greater than 1. For example, when the weight is 10, if the relationship determination unit 107 determines that user A has sent a speech message such as a threat to user B, the number of speeches is set to 10. Alternatively, the relationship determination unit 107 may weight the number of speeches of the user invited to the group chat with a weight greater than 1. The weight may be set for each classification. The relationship determination unit 107 may determine the relationship between users based on the product of the number of speeches for each classification and the weight for each classification.

[0037] The graph generated by the graph generation unit 106 may be hierarchically structured into multiple layers. For example, the nodes belonging to the topmost layer are displayed at the top of the graph. Also, the nodes belonging to the bottommost layer are displayed at the bottom of the graph. The graph generation unit 106 determines the layer to which the node representing each user belongs according to the upper limit relationship determined by the relationship determination unit 107. In the graph generated by the graph generation unit 106, among the user pairs, the layer to which the user determined to be in the upper position belongs is set to a layer higher than the layer to which the user in the lower position belongs. The graph generation unit 106 may display the nodes in the upper layer in a larger size than the nodes in the lower layer. The graph generation unit 106 may change the shape or display color of the nodes between the upper layer nodes and the lower layer nodes, and display the upper layer nodes and the lower layer nodes in a distinguishable manner.

[0038] The graph generation unit 106 may calculate, for each user, the number of users whose position is directly or indirectly lower than that of the user by tracing the relationship between user pairs. Suppose the first user is determined to be a user higher in position than the second user, and the second user is determined to be a user higher in position than the third user. In that case, the graph generation unit 106 calculates that the first user has two users in the lower position. For a certain user, the users whose position is directly or indirectly lower than that of the user are also called the users under a certain user.

[0039] The graph generation unit 106 may calculate the number of subordinate users for each user and determine the layer to which each user belongs according to the number of subordinate users. In other words, the graph generation unit 106 may calculate the number of reachable nodes by tracing the directed edges and determine the layer to which each user belongs according to the number of reachable nodes. In the directed graph, for example, the graph generation unit 106 may display the user with the most subordinate users in the topmost layer and the user with no subordinate users in the bottommost layer.

[0040] Next, the operation procedure will be described. FIG. 6 shows the operation procedure of the relationship visualization device 100. The operation procedure of the relationship visualization device 100 corresponds to the relationship visualization method. A plurality of users submit an electronic device such as a smartphone to a user who operates the relationship visualization device 100 (hereinafter also referred to as an analyst). The analyst extracts conversation history data from the submitted electronic device. The analyst inputs the extracted conversation history data into the relationship visualization device 100. The conversation history data may be obtained from a server of a service provider that provides a messaging service and input into the relationship visualization device 100.

[0041] The conversation history acquisition unit 101 acquires conversation history data (step S1). The classification unit 102 classifies the speech messages included in the conversation history data into a plurality of classifications (step S2). The node extraction unit 103 extracts the user who is the sender of the speech message and the user who is the recipient (step S3). The classification unit 102 stores the classification result of the speech message in the classification result storage unit 104. The classification unit 102 may store only the classification results of the speech messages classified into a predetermined classification in the classification result storage unit 104. Step S2 and step S3 may be performed in parallel.

[0042] The extraction unit 105 extracts the speech messages classified into a predetermined classification from the classification result storage unit 104 (step S4). In step S4, the extraction unit 105 extracts, for example, speech messages classified as any of instruction, command, and threat. The relationship determination unit 107 determines the hierarchical relationship of positions between user pairs between whom speech messages classified into a predetermined classification are transmitted and received (step S5). In step S5, the relationship determination unit 107 determines that, for example, the user who transmitted the speech message classified as any of instruction, command, and threat has a higher position than the destination user. In step S5, the relationship determination unit 107 may extract information regarding the hierarchical relationship between users as auxiliary information from the auxiliary information storage unit 130 and use it for determining the hierarchical relationship of positions.

[0043] The graph generation unit 106 generates a directed graph including nodes representing each of a plurality of users who participated in the conversation (step S6). In step S6, the graph generation unit 106 generates a directed graph hierarchically divided into a plurality of layers. For example, the graph generation unit 106 summarizes the relationships between user pairs for all users, and calculates the number of users directly or indirectly subordinate to each user. The graph generation unit 106 displays the nodes representing each user in a layer corresponding to the number of subordinate users when displaying the graph.

[0044] FIG. 7 shows an example of the displayed directed graph. Here, it is assumed that the directed graph is used in criminal investigations. The extraction unit 105 extracts speech messages classified into instructions, orders, and threats. The user who is the source of the speech message classified into instructions, orders, and threats is considered to be in a position to give orders to the destination user. In this example, the directed graph is also called an order graph.

[0045] For example, in a conversation between user "Yamagami" and user "Yamanaka", assume that Yamagami sent a speech message classified into instructions, orders, and threats to Yamanaka. In that case, the relationship determination unit 107 determines that Yamagami is a user in a higher position than Yamanaka. Also, in a conversation between user "Yamanaka" and user "Yamaya", assume that Yamanaka sent a speech message classified into instructions, orders, and threats to Yamaya. In that case, the relationship determination unit 107 determines that Yamanaka is a user in a higher position than Yamaya. The relationship determination unit 107 determines, for each user pair, the user in the higher position, that is, the user on the side giving the order, and the user in the lower position, that is, the user on the side receiving the order.

[0046] The graph generation unit 106 summarizes the hierarchical relationships between user pairs and calculates the number of subordinate users for each user. For example, in a directed graph, the graph generation unit 106 calculates the number of nodes that can be traced by an arrow. The graph generation unit 106 determines the level to which each user belongs, that is, the height in the directed graph, according to the number of nodes that can be traced by the arrow. The graph generation unit 106 draws an arrow from the node on the instructing side to the node on the instructed side. When the relationship between the user on the instructing side and the user on the instructed side loops, the graph generation unit 106 determines that the levels to which the users forming the loop belong are the same level.

[0047] In the example shown in FIG. 7, the number of nodes that can be traced by an arrow from Yamashita is the largest, and Yamashita is displayed at the top level. The number of nodes that can be traced by an arrow from Yamanaka is four. Also, the hierarchical relationships among Kawashima, Kawana, Kawanose, and Kawada form a loop, and the number of nodes that can be traced is four. In this case, the group of Yamanaka and Kawashima, Kawana, Kawanose, and Kawada is displayed in the second level from the top. For the group of Yamanaka and Kawashima, Kawana, Kawanose, and Kawada, if the heights of these four nodes are aligned, the connections within the loop become difficult to see. Therefore, it is advisable to display them with slightly shifted heights from each other.

[0048] In the example shown in FIG. 7, the number of nodes that can be traced by an arrow from Yamatani is one. Also, there are no nodes that can be traced by an arrow from Yamashita, Yamada, and Yamakawa. In this case, Yamatani is displayed in the third level from the top, and Yamashita, Yamada, and Yamakawa are displayed in the bottommost level. In the supplementary information, if the relationship indicating that Yamatani and Yamashita are colleagues is shown, Yamashita is displayed side by side with Yamatani in the third level. The positions of the nodes within the same level may be arbitrarily changed by the analyst operating the relationship visualization device 100. For example, the positions of the nodes within the group of Kawashima, Kawana, Kawanose, and Kawada may be arbitrarily changed by the analyst.

[0049] In FIG. 7, the arrows indicate the relationship between the user who gives an order and the user who receives the order. By referring to the directed graph shown in FIG. 7, the analyst can understand the command hierarchy. For example, suppose Yamada performs an action upon receiving an order from someone. In that case, even if Taniguchi gives an order directly to Yamada, the analyst can determine that Taniguchi may have received the order from Yamanaka, and Yamanaka may have received the order from Yamanokami. In this way, the analyst can determine from the command graph shown in FIG. 7 that Yamanokami is located at the top of the command hierarchy.

[0050] In this embodiment, the extraction unit 105 extracts the speech messages classified into a predetermined classification. The graph generation unit 106 generates a directed graph in which the user of the source of the speech message and the destination user are connected by a directed edge. In this embodiment, the relationship between the users who send and receive the speech messages classified into a predetermined classification is visualized using a directed graph. The analyst operating the relationship visualization device 100 can understand the relationship between users regarding the speech messages classified into a predetermined classification by referring to the displayed directed graph.

[0051] In this embodiment, the relationship determination unit 107 determines the hierarchical relationship of positions between user pairs who send and receive the speech messages classified into a predetermined classification. The graph generation unit 106 determines the hierarchy of each user in the directed graph with a hierarchical structure based on the determined hierarchical relationship of positions. In this case, the analyst can understand the hierarchical relationship among three or more users, rather than for each user pair.

[0052] This embodiment focuses on the human relationship with the classification items suggesting the instructing side and the instructed side as the relationship. In this embodiment, "instructing" and "threatening" are also regarded as a kind of instructing relationship and are treated similarly. The relationship determination unit 107 determines the person who gives an instruction as the superior and the person who receives the instruction as the inferior. In the case where instructions are mutually given between users, the relationship determination unit 107 determines the user who gives more instructions as the superior. The graph generation unit 106 creates a directed edge from the "person who gives an instruction" to the "person who receives the instruction". In the directed graph, the graph generation unit 106 displays the node with a larger number of reachable people, that is, the number of nodes, higher by following the arrow. By displaying such a directed graph, the relationship visualization device 100 can intuitively and easily display the hierarchical relationship between users. In the directed graph, by tracing the relationship between users in the direction opposite to the arrow direction from the lower layer, that is, the terminal-side node, the analyst can identify the originating user who indirectly gives an instruction to the terminal-side user.

[0053] In the above embodiment, an example where the predetermined classification is instruction, indication, and threat has been described. However, the present disclosure is not limited to this. The predetermined classification can be appropriately set or selected according to the analyzed relationship. For example, when it is desired to visualize relationships such as bullying, power harassment, or sexual harassment, the extraction unit 105 may use the classification of related speech messages as the predetermined classification.

[0054] Subsequently, the hardware configuration of the relationship visualization device 100 will be described. FIG. 8 shows a configuration example of a computer device that can be used as the relationship visualization device 100. The computer device 500 includes a processor 510 such as a CPU (Central Processing Unit), a storage unit 520, a ROM (Read Only Memory) 530, a RAM (Random Access Memory) 540, a communication interface (IF: Interface) 550, and a user interface 560.

[0055] The communication interface 550 is an interface for connecting the computer device 500 and the communication network via means such as wired communication means or wireless communication means. The user interface 560 includes a display unit such as a display. Further, the user interface 560 includes an input unit such as a keyboard, a mouse, and a touch panel.

[0056] The storage unit 520 is an auxiliary storage device that can hold various types of data. The storage unit 520 can be used as the product information DB 110. The storage unit 520 does not necessarily have to be a part of the computer device 500 and may be an external storage device or a cloud storage connected to the computer device 500 via a network.

[0057] The ROM 530 is a non-volatile storage device. A semiconductor storage device such as a flash memory with relatively small capacity is used for the ROM 530. The programs executed by the CPU 510 can be stored in the storage unit 520 or the ROM 530. The storage unit 520 or the ROM 530 stores programs for realizing the functions of each part of the relationship visualization device 100.

[0058] When the above program is loaded into a computer, it includes a set of instructions (or software code) for causing the computer to perform one or more functions described in the embodiments. The program may be stored in a non-transitory computer-readable medium or a tangible storage medium. By way of example and not limitation, a computer-readable medium or a tangible storage medium includes random-access memory (RAM), read-only memory (ROM), flash memory, solid-state drive (SSD), or other memory technologies, Compact Disc (CD), digital versatile disc (DVD), Blu-ray (registered trademark) disc, or other optical disc storage, magnetic cassette, magnetic tape, magnetic disk storage, or other magnetic storage devices. The program may also be transmitted on a transitory computer-readable medium or a communication medium. By way of example and not limitation, a transitory computer-readable medium or a communication medium includes electrical, optical, acoustic, or other forms of propagated signals.

[0059] RAM 540 is a volatile memory device. Various semiconductor memory devices such as DRAM (Dynamic Random Access Memory) or SRAM (Static Random Access Memory) are used for RAM 540. RAM 540 can be used as an internal buffer for temporarily storing data and the like. The CPU 510 expands and executes a program stored in the storage unit 520 or the ROM 530 in the RAM 540. By executing the program by the CPU 510, the functions of each part in the relationship visualization device 100 can be realized. The CPU 510 may have an internal buffer capable of temporarily storing data and the like.

[0060] Note that in the present disclosure, the relationship visualization device 100 does not necessarily have to be physically configured as a single device. The relationship visualization device 100 may be configured using a plurality of physically separated devices. For example, the relationship visualization device 100 may have a configuration including a first device having a conversation history acquisition unit 101 (see FIG. 2), a classification unit 102, and a node extraction unit 103, and a second device having an extraction unit 105 and a graph generation unit 106.

[0061] As described above, the present disclosure has been described with reference to the embodiments. However, the present disclosure is not limited to the above-described embodiments. Various changes that can be understood by those skilled in the art can be made to the configuration and details of the present disclosure. Each embodiment can be combined with other embodiments as appropriate.

[0062] Each drawing is merely an example for explaining one or more embodiments. Each drawing is not associated with only one specific embodiment, but may be associated with one or more other embodiments. As can be understood by those skilled in the art, various features or steps described with reference to any one drawing can be combined with features or steps shown in one or more other drawings to create, for example, embodiments that are not explicitly illustrated or described. Not all of the features or steps shown in any one drawing for explaining exemplary embodiments are necessarily essential, and some features or steps may be omitted. The order of the steps described in any drawing may be changed as appropriate.

[0063] Some or all of the above-described embodiments may be described as follows in the appended claims, but are not limited thereto.

[0064] [Appended Claim 1] Among the speech messages included in the conversation data between the users participating in the conversation, extract the speech messages classified into a predetermined classification, A program for causing a computer to execute a process having: generating a graph in which the relationship between a node representing a user who is the source of the extracted speech message and a node representing a user who is the destination of the extracted speech message is visualized using a directed edge.

[0065] [Appendix 2] The generating of the graph includes, when there is a speech message classified into the predetermined classification in a conversation between a user pair including a first user and a second user, determining an upper-lower relationship in position between the first user and the second user according to whether the first user is the user who is the source or the second user is the user who is the source. The program according to Appendix 1.

[0066] [Appendix 3] The graph is hierarchically divided into a plurality of layers. In the graph, a layer to which a user with a higher position belongs is set to a layer higher than a layer to which a user with a lower position belongs. The program according to Appendix 2.

[0067] [Appendix 4] The generating of the graph includes, for each user, calculating the number of users whose positions are directly or indirectly lower than that of the user, and determining the layer to which the user belongs according to the calculated number of users whose positions are lower. The program according to Appendix 3.

[0068] [Appendix 5] The determining of the upper-lower relationship includes, in the conversation between the user pair, comparing the number of speech messages classified into the predetermined classification in which the first user is the user who is the source with the number of speech messages classified into the predetermined classification in which the second user is the user who is the source, and determining, based on the result of the comparison, which of the first user and the second user is the user with a higher position. The program according to any one of Appendices 2 to 4.

[0069] [Appendix 6] Determining the above-lower relationship includes determining, among the first user and the second user, that the user with a larger number of speech messages classified into the predetermined classification in the conversation between the user pair is the user with a higher position, according to the program described in Supplementary Note 5.

[0070] [Supplementary Note 7] Determining the above-lower relationship includes determining the above-lower relationship of the positions by further using auxiliary information indicating the organizational upper-lower relationship among at least some of the users participating in the conversation, according to the program described in any one of Supplementary Notes 2 to 6.

[0071] [Supplementary Note 8] Extracting the speech message by the extraction unit includes extracting, as the speech belonging to the predetermined classification, the speech classified into commands, instructions, and threats among the speech messages included in the conversation data, according to the program described in any one of Supplementary Notes 1 to 6.

[0072] [Supplementary Note 9] Among the speech messages included in the conversation data between the users participating in the conversation, extract the speech messages classified into a predetermined classification, A relationship visualization method including generating a graph in which the relationship between the node representing the user who is the source of the extracted speech message and the node representing the user who is the destination of the extracted speech message is visualized using a directed edge.

[0073] [Supplementary Note 10] An extraction unit that extracts speech messages classified into a predetermined classification from among the speech messages included in the conversation data between the users participating in the conversation, A relationship visualization apparatus including a graph generation unit that generates a graph in which the relationship between the node representing the user who is the source of the extracted speech message and the node representing the user who is the destination of the extracted speech message is visualized using a directed edge.

[0074] Some or all of the elements (e.g., configurations and functions) described in Supplementary Notes 2 to 8 that are subordinate to Supplementary Note 1 may be subordinate to Supplementary Notes 9 and 10 in the same subordinate relationship as Supplementary Notes 2 to 8. Some or all of the elements described in any supplementary note may be applied to various hardware, software, recording means for recording software, systems, and methods.

Explanation of Signs

[0075] 10: Relationship Visualization Device 11: Extraction Unit 12: Graph Generation Unit 100: Relationship Visualization Device 101: Conversation History Acquisition Unit 102: Classification Unit 103: Node Extraction Unit 104: Classification Result Storage Unit 105: Extraction Unit 106: Graph Generation Unit 107: Relationship Judgment Unit 130: Auxiliary Information Storage Unit 150: Display Device

Claims

1. Extracting speech messages classified into a predetermined classification from among the speech messages included in the conversation data between users participating in a conversation, and causing a computer to execute a process including generating a graph in which the relationship between a node representing the user who is the source of the extracted speech message and a node representing the user who is the destination of the extracted speech message is visualized using a directed edge.

2. Generating the graph includes determining an upper-lower relationship in position between the first user and the second user according to whether the first user or the second user is the user who is the source of the speech message classified into the predetermined classification in a conversation between a user pair including the first user and the second user. The program according to claim 1.

3. The graph is hierarchically divided into a plurality of layers, and in the graph, the layer to which the user with the higher position belongs is set to a layer higher than the layer of the user with the lower position. The program according to claim 2.

4. Generating the graph includes, for each user, calculating the number of users whose position is lower, either directly or indirectly, with respect to the user, and determining the layer to which the user belongs according to the calculated number of users whose position is lower. The program according to claim 3.

5. Determining the upper-lower relationship includes comparing the number of speech messages classified into the predetermined classification in which the first user is the user who is the source in the conversation between the user pair, and the number of speech messages classified into the predetermined classification in which the second user is the user who is the source, and determining, based on the result of the comparison, which of the first user and the second user is the user with the higher position. The program according to any one of claims 2 to 4.

6. Determining the above - mentioned upper - lower relationship includes determining that, among the first user and the second user, the user with a larger number of speech messages classified into the predetermined classification in the conversation between the user pair is the user with a higher position. The program according to claim 5.

7. Determining the above - mentioned upper - lower relationship includes further using auxiliary information indicating the upper - limit relationship in the organization among at least some of the users participating in the conversation to determine the upper - lower relationship of the positions. The program according to any one of claims 2 to 4.

8. Extracting the speech message by the extraction unit includes extracting, among the speech messages included in the conversation data, the speeches classified into commands, instructions, and threats as the speeches belonging to the predetermined classification. The program according to any one of claims 1 to 4.

9. Among the speech messages included in the conversation data between the users participating in the conversation, extract the speech messages classified into a predetermined classification, A relationship visualization method having: generating a graph in which the relationship between the node representing the user who is the source of the extracted speech message and the node representing the user who is the destination of the extracted speech message is visualized using a directed edge.

10. An extraction unit that extracts speech messages classified into a predetermined classification from among the speech messages included in the conversation data between the users participating in the conversation, A relationship visualization device comprising: a graph generation unit that generates a graph in which the relationship between the node representing the user who is the source of the extracted speech message and the node representing the user who is the destination of the extracted speech message is visualized using a directed edge.

Images