Arithmetic device and support method

JP2023077009A5Active Publication Date: 2024-11-29THE RITSUMEIKAN TRUST
View PDF 0 Cites 0 Cited by

Patent Information

Application Number
JP2021190082
Authority / Receiving Office
JP · JP
Patent Type
Applications
Current Assignee / Owner
Filing Date
2021-11-24
Publication Date
2024-11-29
Estimated Expiration
2041-11-24

AI Technical Summary

Technical Problem

Existing group communication systems fail to effectively share information across multiple groups, limiting the opportunity for cross-group discussions and reducing the diversity of topics and engagement.

Method used

An arithmetic device and method that utilize natural language processing to identify and present feature parameters from other groups, allowing for real-time sharing of information by comparing and extracting feature parameters that differ from the first group's parameters, using co-occurrence networks and frequent word lists to enhance information exchange.

Benefits of technology

Enhances information sharing among groups, promoting active and diverse discussions by introducing new perspectives and maintaining discussion complexity, as demonstrated by increased word complexity and engagement metrics.

✦ Generated by Eureka AI based on patent content.

Smart Images

  • Figure 00000000_0000_ABST
    Figure 00000000_0000_ABST
Patent Text Reader

Abstract

To provide an arithmetic device and a support method which, in order to share information among a plurality of users who constitute a plurality of groups, causes one of the groups to output information so that information of another one of the groups can be used effectively.SOLUTION: An arithmetic device 1 comprises: an input unit which receives input of text data while classifying the text data into a corresponding group; and a processor which executes arithmetic processing using the text data. The processor is configured to: apply natural language processing to the text data classified into a first group to calculate a first feature parameter; compare the first feature parameter with a second feature parameter obtained from text data classified into a second group to extract, as a presentation feature parameter, a feature parameter not corresponding to the first feature parameter among second feature parameters; and output, from an output device, the presentation feature parameter in association with the first group.SELECTED DRAWING: Figure 2
Need to check novelty before this filing date? Find Prior Art

Description

[Technical Field]

[0001] The present disclosure relates to a computing device and a support method. [Background technology]

[0002] There are cases where multiple users in a group wish to share information. Japanese Patent Application Laid-Open Publication No. 2005-267266 (Patent Document 1) discloses an information sharing system between groups, which includes a communication control unit for transmitting and receiving data to and from a mail server, a shared information storage unit for storing address data and shared data for each user terminal in the group, and a shared information management unit for managing this data. The shared information management unit has a shared information modification function for accepting change request data input from outside to modify the content of data stored in the shared information storage unit and modifying the content of the data in the shared information storage unit based on the change request data, a change data generation function for generating change data representing the changes to the data in the shared information storage unit based on the change request data, and a change data transmission function for reading the address data of each user terminal from the shared information storage unit and transmitting the change data to a data distribution server via the communication control unit. This information sharing system between groups allows information sharing even when the users in the group are located inside or outside a firewall, without the need for a dedicated server outside the firewall. [Prior art documents] [Patent documents]

[0003] [Patent Document 1] Japanese Patent Application Laid-Open No. 2005-267266 Summary of the Invention

[0004] One situation in which multiple users who make up a group want to share information is when multiple users are divided into multiple groups, each consisting of one or more users, and discussing something. In this case, sharing information with other groups is expected to add new topics to the discussion within one's own group and make the discussion more lively.

[0005] However, when multiple groups are discussing, there are few opportunities to access information from other groups. As a result, there are cases where information from other groups cannot be shared. There is also a desire to share information from other groups in a way that can be effectively used by one's own group. Therefore, there is a demand for a computing device and a support method that can output information from other groups to a certain group in a way that can be effectively used.

[0006] According to one embodiment, the arithmetic device includes an input unit that distinguishes between groups and accepts input of text data, and a processor that performs arithmetic processing using the text data, and the processor is configured to calculate first feature parameters by performing natural language processing on the text data distinguished into a first group, compare the first feature parameters with second feature parameters obtained from the text data distinguished into a second group, extract feature parameters from the second feature parameters that do not correspond to the first feature parameters as presentation feature parameters, and output the presentation feature parameters from an output device in association with the first group.

[0007] According to one embodiment, the support method is a method for supporting information sharing among multiple users, and includes the steps of: inputting the content of statements made by one or more users constituting each of multiple groups, each of which is made up of one or more users from the multiple users; calculating feature parameters for the information in each of the multiple groups by performing natural language processing on the text representing the content of statements; comparing the feature parameters for one of the multiple groups with the feature parameters for other groups from the multiple groups, extracting feature parameters for presentation that do not correspond to the feature parameters for the one group from the feature parameters for the other groups; and presenting the feature parameters for presentation to one or more users constituting one group.

[0008] Further details will be described in the following embodiments. [Brief explanation of the drawings]

[0009] [Figure 1] FIG. 1 is a schematic diagram illustrating an example of the configuration of a support system according to an embodiment. [Figure 2] FIG. 2 is a schematic block diagram showing the configuration of each of the servers and end devices included in the support system. [Figure 3] FIG. 3 is a diagram for explaining natural language processing and frequent word list generation processing in the server. [Figure 4] FIG. 4 is a diagram for explaining the co-occurrence network generation process performed by the server. [Figure 5] FIG. 5 is a diagram for explaining the suggestion source group determination process and the suggestion word extraction process performed by the server. [Figure 6] FIG. 6 is a diagram showing an example of the flow of a support method for supporting information sharing among a plurality of users in a support system. [Figure 7] FIG. 7 is a schematic diagram of a display screen displayed on the display of the terminal device. [Figure 8]FIG. 8 is a schematic diagram of a display screen displayed on the display of the terminal device. [Figure 9] FIG. 9 shows the results of an experiment conducted by the inventors. [Figure 10] FIG. 10 shows the results of an experiment conducted by the inventors. [Figure 11] FIG. 11 shows the results of an experiment conducted by the inventors. DETAILED DESCRIPTION OF THE INVENTION

[0010] <1. Overview of the Calculation Device and Support Method>

[0011] (1) The arithmetic device according to this embodiment includes an input unit that accepts input of text data by distinguishing between groups, and a processor that executes arithmetic processing using the text data. The processor is configured to calculate first feature parameters by performing natural language processing on the text data classified into a first group, compare the first feature parameters with second feature parameters obtained from the text data classified into a second group, extract feature parameters from the second feature parameters that do not correspond to the first feature parameters as presentation feature parameters, and output the presentation feature parameters from an output device in association with the first group.

[0012] The characteristic parameters are indices that represent the characteristics of the information in the group, and are, for example, words obtained from the text data, conceptual words such as synonyms, and semantic concepts extracted from sentences indicated by the text data.

[0013] Among the second feature parameters, feature parameters that do not correspond to the first feature parameters are extracted as presentation feature parameters, and the presentation feature parameters are output from the output device in association with the first group, so that the presentation feature parameters can be output in a manner that allows the first group to effectively use the information of the second group, thereby enabling information to be shared among multiple users.

[0014] (2) Preferably, the processor is configured to execute the calculation process each time text data is input to the input unit, thereby enabling the presentation feature parameters to be output in real time in association with the first group.

[0015] (3) Preferably, the processor is configured to calculate feature parameters for each of a plurality of groups, including the first group, from text data separated into each of the plurality of groups, and determine a second group from among a plurality of groups other than the first group, based on a first index of each of the plurality of feature parameters. The first index indicates an index representing a difference between the first feature parameter and the second feature parameter. The first index may be, for example, a difference between words used in the first group and words used in the second group.

[0016] (4) Preferably, the processor is configured to extract the presentation feature parameters based on a second index of the second feature parameters that do not correspond to the first feature parameters. The second index is a criterion used to extract the presentation feature parameters to be output in association with the first group from the second feature parameters. As an example, the feature parameters may be words, and the second index may be the frequency of appearance of the words in the second group.

[0017] (5) Preferably, the feature parameters are words obtained from the text data, and outputting the feature parameters for presentation includes presenting the words that are the first feature parameters based on co-occurrence information obtained by natural language processing of the text data, and presenting the words that are obtained from the text data classified into the second group that are the feature parameters for presentation together with the words that are the first feature parameters. This allows information in language, such as discussions, to be shared, for example.

[0018] (6) Preferably, displaying the presented feature parameter together with the word that is the first feature parameter includes presenting the word that is the presented feature parameter separately from the word that is the first feature parameter, thereby making it easier to understand the presented feature parameter.

[0019] (7) Preferably, the processor is configured to execute a calculation process each time text data is input to the input unit, and to present updated information on the feature parameters to be presented each time the feature parameters to be presented are extracted. This allows the feature parameters to be presented in real time, making it easy to understand that the feature parameters to be presented have been updated.

[0020] (8) A support method according to this embodiment is a method for supporting information sharing among multiple users, and includes the steps of: inputting, for multiple groups each consisting of one or more users from the multiple users, the content of each of the one or more users constituting each group; calculating feature parameters for the information in each of the multiple groups by performing natural language processing on the text representing the content of the utterances; comparing the feature parameters for one of the multiple groups with the feature parameters for the other groups; extracting, as presentation feature parameters, feature parameters that do not correspond to the feature parameters for the one group from the feature parameters for the other groups; and presenting the presentation feature parameters to one or more users constituting one group. By presenting the presentation feature parameters to one or more users constituting each group in this manner, information is shared among the multiple users.

[0021] <2. Examples of Calculation Devices and Support Methods>

[0022] 1 is a schematic diagram showing an example of the configuration of a support system 100 according to this embodiment. The support system 100 is a system for supporting information sharing among multiple users. As an example, the information sharing may be information sharing when multiple users are having a discussion.

[0023] The multiple users are divided into multiple groups A to N, each consisting of one or more users, and hold discussions. Each group is provided with one or more terminal devices 5. Preferably, a terminal device 5 is provided for each user.

[0024] Terminal devices 5, each of which is owned by a plurality of users, are connected to the server 1 by a communication line 3, and can communicate with the server 1. The communication line 3 may be a communication line that physically connects the terminal devices 5 and the server 1, or may be a communication line that conceptually represents wireless communication. Furthermore, communication between the terminal devices 5 and the server 1 may be realized via a communication network (not shown), such as the Internet, or via another device, such as a router.

[0025] 2 is a schematic block diagram showing the configuration of each of the server 1 and the terminal device 5 included in the support system 100. Referring to FIG. 2, the server 1 is configured as a computer having a processor 11 and a memory 12. The processor 11 is, for example, a CPU (Central Processing Unit). The memory 12 includes a flash memory, an EEPROM, a ROM, a RAM, etc. Alternatively, the memory 12 may be a primary storage device or a secondary storage device.

[0026] The memory 12 stores a computer program 121 that is executed by the processor 11. The memory 12 has a word database (DB) 122, which is an area for storing words for each group. The word DB 122 will be described later.

[0027] The processor 11 performs arithmetic processing by executing the computer program 121. The arithmetic processing is arithmetic processing using text data input from the terminal device 5, which will be described later.

[0028] The server 1 includes a communication unit 13. The communication unit 13 is, for example, a communication module. The communication unit 13 communicates with the terminal device 5 via the communication line 3. The communication unit 13 transmits instructed information to the terminal device 5 in accordance with a control signal from the processor 11. The server 1 also inputs an electrical signal received from the communication unit 13 to the processor 11.

[0029] The terminal device 5 is, for example, a tablet or a personal computer, and is used as a user interface. With reference to Fig. 2, the terminal device 5 is configured by a general computer having a processor 51 and a memory 52. ​​The processor 51 is, for example, a CPU.

[0030] The memory 52 may be a primary storage device or a secondary storage device. The memory 52 stores a computer program 521 executed by the processor 51. The memory 52 has a group information storage unit 522, which is an area for storing group information. The group information includes identification information for each group.

[0031] The terminal device 5 includes a communication unit 53. The communication unit 53 is, for example, a communication module. The communication unit 53 communicates with the server 1 via the communication line 3. The communication unit 53 transmits instructed information to the server 1 in accordance with a control signal from the processor 51. The communication unit 53 also inputs an electrical signal received from the server 1 to the processor 51.

[0032] The terminal device 5 has a display 54. The display 54 is an example of an output device that provides information to a user by outputting information. The display 54 displays the information instructed in accordance with a control signal from the processor 51.

[0033] The terminal device 5 includes a microphone 55. The microphone 55 is an example of a device that constitutes part of an input device for inputting text data representing the content of a discussion, i.e., the content of a user's remarks, to the processor 11. The microphone 55 acquires voice and inputs a voice signal to the processor 51. The microphone 55 is preferably a pin microphone that can be placed near the user's mouth. This can improve the accuracy of voice recognition, which will be described later.

[0034] The processor 51 performs arithmetic processing by executing a computer program 521 stored in the memory 52. ​​The arithmetic processing includes a text conversion process 511. The text conversion process 511 includes converting an audio signal input from the microphone 55 into text by performing a voice recognition process on the audio signal.

[0035] Processor 51 passes the text data obtained in text conversion process 511 to communication unit 53, which transmits the text data to server 1. For this reason, processor 51 executing text conversion process 511 can be said to constitute a part of an input device for inputting text data representing the content of users' comments, which is the content of the discussion, to processor 11.

[0036] Another example of an input device for inputting text data representing user statements, which are the content of the discussion, into processor 11 may be a text input unit such as a keyboard or touch panel. In this case, users' statements are entered as characters entered by operating the keyboard or touch panel. The text input unit inputs an operation signal indicating a user operation to processor 51. Processor 51 generates text data from the operation signal.

[0037] Processor 51 causes server 1 to transmit the group identification information stored in group information storage unit 522 along with the text data. As a result, text data representing the content of each user's comment is input to processor 11 in association with the identification information of the group to which that user belongs. This enables processor 11 to perform arithmetic processing using the text data.

[0038] The arithmetic processing executed by the processor 51 includes a display processing 512. The display processing 512 includes causing the display 54 to display information based on information transmitted from the server 1. The display on the display 54 will be described later.

[0039] The arithmetic processing executed by the processor 11 includes natural language processing 111. The arithmetic processing executed by the processor 11 also includes networking processing 112, which includes frequent word list generation processing 113. FIG. 3 is a diagram for explaining the natural language processing 111 and the frequent word list generation processing 113. Referring to FIG. 3, the natural language processing 111 includes morphological analysis (step S1). The morphological analysis includes dividing text data input from the terminal device 5 into morphemes. A morpheme refers to the smallest unit having a linguistic meaning.

[0040] In the example of Fig. 3, text 20 "Apples are red and sweet fruits" is input from terminal device 5 together with identification information 21 of group A to which the user of terminal device 5 belongs. This means that text 20, which is text data obtained by converting a speech made by a user belonging to group A into text by terminal device 5 carried by the user, is input to processor 11. In the example of Fig. 3, text 20 is divided into morpheme groups 22: "apple," "is," "red," "te," "amai," "kajitsu," and "desu" (step S1).

[0041] The natural language processing 111 includes removing function words (step S2). Function words are words that are used together with content words such as nouns and verbs to indicate their grammatical functions. Specific examples of function words include auxiliary verbs, articles, demonstratives, prepositions, and conjunctions. In the example of FIG. 3, the function words "wa," "te," and "desu" are removed from the morpheme group 22 (step S2). As a result, the content words "ringo," "akai," "amai," and "desu" are extracted from the text 20, and a word group 23 based on these words is obtained.

[0042] The words extracted from the text data entered over a predetermined period of time in association with a group represent the words used in the discussions in that group. Therefore, the extracted words can be considered an example of an index (feature parameter) that represents the characteristics of the information in that group. The feature parameter is not limited to the words themselves extracted from the text data, but may also be, for example, conceptual words such as synonyms, or words representing semantic concepts extracted from sentences indicated by the text data.

[0043] The natural language processing 111 includes analyzing the co-occurrence of each word (step S3). Co-occurrence here refers to the phenomenon of two different words being used simultaneously in one sentence. In the example of FIG. 3, the co-occurrence relationship is analyzed for each word in the word group 23 extracted from the text 20 (step S3). Specifically, for "apple," "red," "sweet," and "fruit" are analyzed as words having a co-occurrence relationship. For "red," "apple," "sweet," and "fruit" are analyzed as words having a co-occurrence relationship.

[0044] Co-occurrence relationships are analyzed sentence by sentence. When a speech is input to the terminal device 5 as speech, one sentence is obtained by the text conversion process 511 of the processor 51. One sentence is, for example, a portion of speech data sandwiched between silent periods of a length equal to or greater than a threshold. In this case, the text data is input from the terminal device 5 to the processor 11 of the server 1 sentence by sentence, for example, by dividing the text data into minutes. This allows the processor 11 to analyze the co-occurrence of each word sentence by sentence.

[0045] Processor 11 stores each word in word group 23 extracted from text 20 in memory 12 together with co-occurrence information indicating the co-occurrence relationships obtained for the words (step S4). In the example of Figure 3, information 24A about "apple" and information 24B about "red" are stored in word DB 122.

[0046] Word DB 122 stores words as first feature parameters for each group. As an example, word DB 122 has word DB 122A for group A, word DB 122B for group B, etc. as shown in Fig. 3. In this case, processor 11 stores information 24A, 24B, etc. in word DB 122A for group A indicated in identification information 21 added to text 20.

[0047] The frequent word list generation process 113 includes generating a frequent word list 25 using words extracted from text data input during a predetermined period and stored by group in the word DB 122 (step S5). The frequent word list 25 is an example of information that lists a predetermined number of words that have been extracted most frequently among the words stored in the word DB 122, that is, words that appear most frequently in text data associated with a certain group. In the example of Fig. 3, the frequent word list 25 is a list that lists the 10 most frequently appearing words among the words stored in the word DB 122A for group A, arranged in descending order of appearance frequency, together with the number of times each word was extracted.

[0048] For a certain word, the higher the frequency of appearance, the higher the interest in the group's discussion, and the lower the frequency, the lower the interest in the group's discussion. In other words, for a certain word, the higher the frequency of appearance, the higher the importance of the information in the group, and the lower the frequency, the lower the importance of the information in the group's discussion.

[0049] The networking process 112 includes a co-occurrence network generation process 114. Fig. 4 is a diagram for explaining the co-occurrence network generation process 114. The co-occurrence network generation process 114 includes generating a network data model 26 (step S6) using co-occurrence information stored in association with words stored in the word DB 122. Referring to Fig. 4, the network data model 26 refers to information that represents the relationships between words stored in the word DB 122 based on the co-occurrence information.

[0050] The networking process 112 includes generating a co-occurrence network 27 (step S7) using the network data model 26 and the frequent word list 25. Referring to Fig. 4, the co-occurrence network 27 is a graph in which frequently occurring words for a target group are defined as nodes 71, and co-occurrence relationships between corresponding words are connected between the nodes 71 as edges 72.

[0051] Nodes 71 have the appearance frequency of the corresponding word, and as an example, have a display style according to the appearance frequency. For example, in the example of Fig. 4, the higher the appearance frequency, the larger the node 71, and the lower the appearance frequency, the smaller the node 71. Furthermore, nodes 71 whose appearance frequency is higher than a predetermined frequency may be displayed in red, and nodes 71 whose appearance frequency is lower than the predetermined frequency may be displayed in black.

[0052] The edges 72 indicate the degree of co-occurrence between words, and as an example, the thickness of the edges varies depending on the degree of co-occurrence. For example, in the example of Figure 4, the edges 72 are thicker the higher the co-occurrence, and thinner the lower the co-occurrence. By generating such a co-occurrence network 27, multiple words used in a discussion by a certain group and their co-occurrence relationships can be represented.

[0053] Preferably, the natural language processing 111 includes determining whether each extracted word is positive or negative. The determination may be made based on a dictionary for determination, which is stored in advance or loaded from another device. In this case, the node 71 may have a display mode corresponding to the determination result. For example, in the example of FIG. 4, the co-occurrence network 27 includes a node 71A that is displayed in a color (e.g., blue) that represents a negative word. This allows multiple characteristics used in a discussion within a group to be represented.

[0054] The arithmetic processing executed by the processor 11 includes a presentation processing 115. The presentation processing 115 includes a presentation source group determination processing 116. The presentation processing 115 also includes a presentation word extraction processing 117.

[0055] The presentation source group determination process 116 is a process for determining another group (second group) from which presentation feature parameters to be output to the terminal device 5 of the target group (first group) are obtained. The presentation feature parameters refer to feature parameters (second feature parameters) of the second group that do not correspond to the feature parameters (first feature parameters) of the first group. Specifically, the presentation source group determination process 116 refers to words (presented words) that do not correspond to words stored in the word DB of the first group among the words stored in the word DB of the second group. The presentation word extraction process 117 refers to a process for extracting presentation words.

[0056] The source group determination process 116 includes determining a second group from among a plurality of groups other than the first group based on a first index of each of the characteristic parameters obtained for groups A to N. The first index refers to an index representing the difference in the characteristic parameters for each group, that is, an index representing the difference in the words used in the discussions for each group. Since a group whose characteristic parameters differ more greatly from the first characteristic parameter of the first group is more likely to be engaged in a different discussion from the first group, it is considered effective for information sharing to present information about such a group as the second group. Therefore, the source group determination process 116 preferably includes determining, as the second group, a group whose characteristic parameters differ more greatly from the first index.

[0057] The first index may be, for example, the ratio R of the frequency of occurrence of words with a high frequency of occurrence (hypernyms) among the words stored in the word DB of the first group to the frequency of occurrence of words stored in the word DB of another group. In this case, it can be said that the lower the ratio R, the greater the difference from the feature parameters of the first group.

[0058] 5 is a diagram for explaining the presentation source group determination process 116 and the presentation word extraction process 117. Referring to Fig. 5, the presentation source group determination process 116 includes determining k words, which are predefined in descending order of appearance frequency in the frequent word list 25 of the first group, as hypernyms, and extracting hypernyms from each of the frequent word lists 25B to 25N of the other groups (step S11).

[0059] 5, k is three, and three hypernyms, "artificial intelligence," "economy," and "development," obtained from frequent word list 25, are extracted from frequent word lists 25B to 25N, respectively. Two words, "economy" and "development," are extracted from frequent word list 25B, and one word, "artificial intelligence," is extracted from frequent word list 25N.

[0060] The presentation source group determination process 116 includes calculating a ratio R for groups other than the first group (step S12). The ratio R may be, for example, a ratio of hypernyms in the frequent word lists 25B to 25N of the other groups.

[0061] As an example, the ratio R is the ratio of the number of times that a hypernym is extracted to the total number of times that m words predefined in order of highest frequency of appearance in each of the frequent word lists 25B to 25N are extracted. In the example of Fig. 5, m is 4, and the ratio RB of hypernyms in group B is 0.59, and the ratio RN of hypernyms in group N is 0.31.

[0062] The presentation source group determination process 116 includes determining a second group using the proportion R of hypernyms calculated for each group other than the first group (step S13). In the example of FIG. 5, the group N with the smallest proportion R is determined to be the second group. This is because the smaller the proportion R, the greater the difference between the feature parameters and the first feature parameters. Therefore, by selecting the group with the smallest proportion R as the second group and extracting a presentation term from the second feature parameters, information can be shared efficiently among multiple users.

[0063] The presentation source group determination process 116 includes extracting presentation words from the words stored in the word DB of the determined second group (step S14). Extracting presentation feature parameters includes extracting presentation feature parameters based on a second index of feature parameters among the second feature parameters that do not correspond to the first feature parameters. The second index may be any criterion used to extract words to be presented in the first group, and may be, for example, the frequency of appearance. As another example, when a negative or positive judgment is made for each word, the second index may be the judgment result or a combination with the frequency of appearance.

[0064] In the example of Fig. 5, the word "environment" is extracted as the most frequently extracted word among the words that are included in the frequent word list 25N of group N determined to be the second group and that are not included in the first characteristic parameter. By presenting the suggested words extracted in this way to the first group, the words that are of most interest in the group (second group) that is most different from the words used in the discussion become suggested words.

[0065] The presentation process 115 includes a presentation information generation process 118. The presentation information generation process 118 includes a process for outputting the presentation feature parameters in association with the first group, and as an example, includes generating information to be displayed on the terminal device 5 that has input text data to which the identification information of the first group has been added. Specific screen displays on the terminal device 5 will be described later with specific examples.

[0066] Fig. 6 is a diagram showing an example of the flow of a support method for supporting information sharing among multiple users in the support system 100. Figs. 7 and 8 are schematic diagrams of display screens 40 and 47 displayed on the display 54 of the terminal device 5.

[0067] Display screen 40 in Fig. 7 is a screen that is displayed on display 54 when processing is started on terminal device 5. Display screen 40 includes button 41 for instructing the start of processing, i.e., the start of voice input of utterance content, button 42 for instructing the end of processing, i.e., the end of voice input of utterance content, and button 46 for selecting a group. The processing in Fig. 6 is started when a group is selected with button 46 and button 41 is pressed.

[0068] As shown in Fig. 1, a terminal device 5 is provided for each of the multiple users. The multiple users are organized into multiple groups, each consisting of one or more users. The microphone 55 of the terminal device 5 acquires the content of each user's speech.

[0069] 6, a voice signal corresponding to the content of a user's speech acquired by microphone 55 is input to processor 51 of terminal device 5 (step S31). Processor 51 converts the voice signal into text by performing a speech recognition process on the voice signal, thereby generating text data (step S33). Processor 51 transmits the generated text data to server 1 together with identification information of the group set by button 46 on display screen 40 (step S35). As a result, text data representing the content of each of the one or more users constituting each group is input to processor 11 of server 1.

[0070] Each time a user's voice is obtained from microphone 55, an audio signal is input to processor 51. Processor 51 repeats steps S33 and S35 each time an audio signal is input. As an example, processor 51 analyzes the audio signal to consider audio during a silent period equal to or greater than a threshold as one sentence, and converts the audio signal for that period into text data for one sentence. As a result, text data is input to server 1 sentence by sentence.

[0071] Text data is input to the server 1 from each of the multiple terminal devices 5 along with group identification information. The processor 11 of the server 1 extracts words from the text data by performing natural language processing on the text data (step S37) and stores the extracted words in the word DB 122 for each group. Extracting words from the text data refers to calculating feature parameters for information in each of the multiple groups. The processing of step S37 includes the processing of steps S1 to S4 in FIG. 3. As a result, words used in discussions about each of the multiple groups are stored in the word DB 122 for the corresponding group.

[0072] Processor 11 generates a frequent word list from word DB 122 for each group (step S39). The generation of the frequent word list includes the process described in step S5 of Fig. 3. As a result, frequent word list 25 for each group is generated.

[0073] Processor 11 generates a co-occurrence network for each group (step S41). Generating a co-occurrence network for each group involves performing the processes described in steps S6 and S7 of Figure 4 for each group. This generates co-occurrence network 27 for each group.

[0074] Processor 11 determines a second group that will be a presentation source group for the first group from among the multiple groups other than the first group (step S43). The determination of the presentation source group includes the processing described in steps S11 to S13 of FIG. 5. In this way, the second group that will be a presentation source group for the first group is determined. In step S43, processor 11 determines a presentation source group for each of the multiple groups as a first group, by performing steps S11 to S13 of FIG. 5 for each of the multiple groups.

[0075] The processor 11 extracts a suggested word (step S45). The suggested word is a presented feature parameter, and extracting the suggested word refers to extracting, as a presented feature parameter, a feature parameter from among the second feature parameters that does not correspond to the first feature parameter by comparing the first feature parameter for the first group with the second feature parameter for the second group. Extracting the suggested word includes the process described in step S14 of FIG. 5. As an example, the processor 11 extracts, as a suggested word, the word that is obtained as the second feature parameter and that is extracted most frequently among words that are not included in the group of frequently occurring words that is the first feature parameter.

[0076] The processor 11 generates presentation information (step S47). Generating the presentation information is a process for presenting the presentation feature parameters to one or more users constituting the first group, and includes, for example, generating information to be displayed on the terminal devices 5 of the users constituting the first group. Therefore, the processor 11 transmits the presentation information generated in step S47 to the terminal devices 5 of the users constituting the first group (step S49).

[0077] The display screen 40 includes a display unit 45 that displays words used in a discussion between multiple users. The processor 51 of the terminal device 5 displays on the display unit 45 based on the presentation information received from the server 1 (step S51).

[0078] The presentation information is information for presenting words that are first feature parameters based on the co-occurrence information, and for displaying presentation words that are presentation feature parameters together with the words that are first feature parameters on the display unit 45. As a result, the presentation words that are presentation feature parameters are displayed on the display unit 45 together with the words that are first feature parameters presented based on the co-occurrence information.

[0079] As an example, processor 11 uses the co-occurrence networks generated for each of the first and second groups to generate, as presentation information, information for adding and displaying suggested words to the co-occurrence network of the first group in a co-occurrence relationship that corresponds to the co-occurrence relationship in the second group. Specifically, processor 11 generates co-occurrence network 27A displayed on display unit 45 of FIG. 7, in which suggested words extracted from the second group are arranged in positions based on the co-occurrence network generated for the second group in co-occurrence network 27 of FIG. 4 generated for the first group. Display unit 45 of FIG. 7 displays co-occurrence network 27A in which nodes 73 representing suggested words such as "tendency," "number of people," "increase," "teacher," "home," and "environment" are added to co-occurrence network 27.

[0080] The shape of the co-occurrence network 27A displayed on the display unit 45 is not limited to the network shape shown in Fig. 7. As another example, it may be a tree shape as shown on the display unit 45 on the display screen 47 in Fig. 8. In the tree display, the processor 11, for example, uses the node 71 that is connected to the largest number of other nodes 71, 73 by edges 72 as a reference, and arranges the other nodes in order from the reference node 71. This makes it easier for the users who make up the first group to understand the transition and structure of the discussion in their own group.

[0081] Furthermore, the co-occurrence network 27A displayed on the display unit 45 may be switchable between a network shape and a tree shape. In this case, the presentation information from the server 1 includes information for presenting the network shape, the tree shape, and both shapes. The display screens 40 and 47 each include a button 44 for instructing switching between the network shape and the tree display. When the button 44 on the display screen 40 is pressed, the processor 51 of the terminal device 5 switches the display screen 40 to the display screen 47. That is, the display on the display unit 45 is switched from the network shape to the tree shape. When the button 44 on the display screen 47 is pressed, the processor 51 switches the display screen 47 to the display screen 40. That is, the display on the display unit 45 is switched from the tree shape to the network shape. This allows the users constituting the first group to switch the display depending on the situation, making it easier to understand the progress and structure of the discussion in their own group.

[0082] When generating co-occurrence network 27A, processor 11 extracts from co-occurrence network 27 the words to which the suggested words were connected by edges in the co-occurrence network of the second group. Processor 11 connects the suggested words to the words extracted from co-occurrence network 27 by edges 72.

[0083] As a result, the users of the first group can understand the words that frequently appear in the discussions of the second group by using the co-occurrence network 27A displayed on the display unit 45. Furthermore, the users can understand the words that frequently appear in the discussions of the second group, along with their co-occurrence with words used in the discussions of their own group (first group). This allows information about the second group to be shared with the users of the first group. Furthermore, the users of the first group can obtain new ideas in the discussions of their own group that are in line with the discussions of the second group, thereby supporting the discussions of the first group.

[0084] Preferably, in the co-occurrence network 27A, node 73 is displayed in a form different from node 71. This allows the presented word to be displayed on the display unit 45 in a manner that distinguishes it from the words in the first group. One example of a different form is a different display color. FIG. 7 shows an example in which the color of a node differs depending on whether the node is determined to be positive, negative, or neither (normal word), whether it is a newly added word, or whether it is a presented word (word of another group). This allows users who make up the first group to easily identify the presented word, even if it is presented together with words from the first group, by using the co-occurrence network 27A displayed on the display unit 45.

[0085] Preferably, edges 72 connecting nodes 73 hold the degree of co-occurrence in the co-occurrence network of the second group, so that users of the first group can understand the degree of co-occurrence of presented words in the second group from co-occurrence network 27A displayed on display unit 45.

[0086] Preferably, processor 11 of server 1 executes steps S37 to S39 every time text data is input from terminal device 5. That is, processor 11 executes steps S37 to S39 every time a sentence is uttered in each group. As a result, every time a user utters a sentence, a co-occurrence network is generated according to the content of the utterance.

[0087] Preferably, the processor 11 of the server 1 executes steps S43 to S49 at a predetermined timing. The predetermined timing may be, for example, when a predetermined number of new words (e.g., 10 words) are added to the co-occurrence network 27 of the first group. As a result, the second group is determined in accordance with the progress of the discussion, and words with a high frequency of appearance in the second group are presented as suggested words. As a result, information is repeatedly shared with groups less relevant to the discussion in real time as the discussion progresses. This allows multiple users to share information as a whole.

[0088] Preferably, the processor 51 of the terminal device 5 receives presentation information from the server 1 and presents the updated information when updating the display unit 45. In other words, the processor 11 causes updated information of the presentation feature parameters to be presented every time a presentation feature parameter is extracted. The updated information is, for example, a display notifying that an update has been made.

[0089] 7, display screen 40 includes display 43 notifying an update. Display 43 is a display that notifies an update by, for example, blinking when display unit 45 is updated. Display 43 may be text or an image such as a character. By displaying display 43, users constituting the first group can understand that co-occurrence network 27A has been updated, that is, that new presentation words extracted from a new second group have been displayed, even if many words are displayed.

[0090] Display 43 may be a button that changes the display to notify that an update has been made to display unit 45. In this case, when the button on display 43 is pressed, processor 51 may update display unit 45. This allows the users constituting the first group to understand that co-occurrence network 27A will be updated, that is, that new presentation words extracted from a new second group will be displayed, and allows them to display the new presentation words by operating the button.

[0091] As another example, the update information may be audio or a combination of audio and visual information, so that the users of the first group can know that the co-occurrence network 27A has been updated, that is, that a new presentation word extracted from the new second group has been displayed, even if they look away from the display 54 of the terminal device 5.

[0092] The inventors conducted an experiment to evaluate the influence of the support system 100 on discussions within one's own group when suggested words from other groups are presented. The experiment was conducted by dividing 20 university student participants into five groups of four people each, and having them simultaneously discuss the same topic. Suggested words were presented to three of the five groups, group A, while only the words of one's own group were presented to two groups, group B, without any suggestion.

[0093] In the experiment, 30-minute discussions based on the following themes 1 and 2 were conducted three times in total, with members being switched to avoid bias in the participation history of groups A and B. Prior to the experiment, a discussion using a practice discussion theme was conducted. Discussion practice topic: "How to use different means of communication within the laboratory" Theme 1: "Attempts to introduce AI into anti-bullying measures in elementary and junior high schools" Theme 2: "Artificial Intelligence and Security"

[0094] The evaluation index used was word complexity. Word complexity is an index that indicates the degree of word repetition and is expressed as the ratio of the number of word types to the total number of words in a discussion. Indicators based on the number of words and word types are known to be effective as automatic evaluation indexes for speech content, as they are relatively easy to analyze and do not involve human subjectivity.

[0095] When calculating word complexity, each unit token obtained by dividing a sentence using the morphological analyzer MeCab was considered a single word. The number of different types of unit tokens (number of different words) was used as the number of word types. Word complexity was calculated for words that appeared in discussions in each of the multiple groups.

[0096] Figures 9 to 11 show the experimental results. Figure 9 shows the counting results of the number of words that appeared in each group, the number of different words, and the number of words from other groups that were presented. Figure 10 shows the time series changes in word complexity in the discussion of Theme 1, and Figure 11 shows the time series changes in word complexity in the discussion of Theme 2.

[0097] From Figure 9, it can be seen that the number of words and different words appearing in group a is generally higher than that in group b for both themes 1 and 2. When comparing the number of words appearing in group a for each of themes 1 and 2, there is a tendency for theme 1, which presented many suggested words, to have a higher number of words appearing and different words than theme 2, which presented fewer suggested words. From this, it can be said that the number of words appearing and different words appearing tends to be higher when more suggested words are presented.

[0098] In Figures 10 and 11, the word complexity of Group B decreases over the course of the discussion. In contrast, the word complexity of Group A decreases less than that of Group B. For example, when comparing the word complexity of each group after the discussion has ended (after 30 minutes of discussion), in both Figures 10 and 11, Group A exceeds the word complexity of Group B. This shows that the group that was presented with suggested words (Group A) experienced less decline in complexity than the group that was not (Group B).

[0099] This is presumably because, in group b, which did not provide suggested words, the discussion tended to delve deeper into specific points (topical subjects such as words or topics) as the discussion progressed, whereas in group a, which provided suggested words, the participants incorporated perspectives on the discussion topics that had come up in other groups into their own group's discussion, and the discussion was based on a wider range of points. Therefore, this experiment verified that by using the support system 100 to provide suggested words from other groups, information is shared between the participants' own group and other groups, effectively promoting information sharing among multiple users. Furthermore, it was verified that when this support system 100 is used in discussions involving multiple groups, discussions based on multiple points of discussion are promoted in each group, and lively discussions can be supported.

[0100] <3. Notes> The present invention is not limited to the above-described embodiment, and various modifications are possible. [Explanation of symbols]

[0101] 1: Server 2: Theme 3: Communication line 5: Terminal device 11: Processor 12: Memory 13: Communications Department 20: Text 21: Identification information 22:Morpheme group 23: Word group 24A: Information 24B: Information 25: Frequent Word List 25B: Frequent Word List 25N: Frequent Word List 26: Network Data Model 27: Co-occurrence network 27A: Co-occurrence network 40:Display screen 41: Button 42: Button 43 :Display 44: Button 45: Display section 46: Button 47:Display screen 51: Processor 52: Memory 53: Communications Department 54: Display 55:Mike 71: Node 71A: Node 72: Edge 73: Node 100: Support System 111: Natural Language Processing 112: Networked Processing 113: Frequent word list generation process 114: Co-occurrence network generation processing 115: Presentation processing 116: Presentation group determination process 117: Presentation word extraction processing 118: Presentation information generation processing 121: Computer Programs 511: Text processing 512: Display processing 521: Computer Programs 522: Group information storage unit 122: Word DB 122A: Group A word database 122B: Word DB for Group B R: Ratio RB: Ratio RN: Ratio

Claims

1. an input unit that distinguishes between groups and accepts input of text data; a processor for executing a calculation process using the text data, The processor, Calculating a first feature parameter by performing natural language processing on the text data classified into a first group; extracting, as presentation feature parameters, feature parameters that do not correspond to the first feature parameters among the second feature parameters by comparing the first feature parameters with second feature parameters obtained from text data classified into a second group; and outputting the presentation feature parameters from an output device in association with the first group; The processor is configured to calculate a feature parameter for each of a plurality of groups including the first group from the text data classified into the plurality of groups, and to determine the second group from among a plurality of other groups other than the first group based on a first index of each of the plurality of feature parameters. Computing device.

2. The processor is configured to execute the arithmetic processing each time the text data is input to the input unit. The computing device of claim 1 .

3. The first index represents a difference between the first characteristic parameter of the first group and the characteristic parameters of a plurality of other groups other than the first group.

3. A computing device according to claim 1 or 2.

4. The processor is configured to extract the feature parameter for presentation based on a second indicator of a feature parameter among the second feature parameters that does not correspond to the first feature parameter. The computing device according to any one of claims 1 to 3.

5. The feature parameters are words obtained from the text data, The outputting of the presentation feature parameters includes: presenting words that are the first feature parameters based on co-occurrence information obtained by the natural language processing of the text data; presenting the words obtained from the text data classified into the second group, which are the presentation feature parameters, together with the words which are the first feature parameters. The computing device according to any one of claims 1 to 4.

6. Displaying together with the word that is the first characteristic parameter includes presenting the word that is the presentation characteristic parameter separately from the word that is the first characteristic parameter. The computing device according to claim 5 .

7. the processor executes the arithmetic processing each time the text data is input to the input unit; The processor further comprises: and presenting updated information of the feature parameters for presentation each time the feature parameters for presentation are extracted. The computing device according to any one of claims 1 to 6.

8. A method for supporting information sharing among multiple users, comprising: inputting, for each of a plurality of groups each composed of one or more users among the plurality of users, the content of each of the one or more users constituting each group; Calculating feature parameters for information in each of the plurality of groups by performing natural language processing on the text representing the utterance content; extracting, as presentation feature parameters, feature parameters for one of the plurality of groups that do not correspond to the feature parameters for the one of the plurality of groups by comparing the feature parameters for the other of the plurality of groups; presenting the presentation characteristic parameters to the one or more users constituting the one group; the other group of the plurality of groups is determined from a plurality of other groups other than the one group of the plurality of groups, based on a first index of each of the feature parameters for each of the plurality of groups calculated from the text classified into each of the plurality of groups; How to help.