JP2015127929A - Method for determining optimal team for discussing subject composed of members whose conversation was recorded - Google Patents

Abstract

PROBLEM TO BE SOLVED: To solve the problem that there is no technique of determining an optimal team of members (personnel assignment) for discussing a subject, i.e., and that because a leader is decided depending on a subject and a relative relationship with other speakers except for a case a leader is decided from the start, a conventional technique focuses on a technique of determining a leader of a conversation though there is important information on members other than the leader for determining an optimal team.SOLUTION: From a recorded conversation, unit time zones are set in such a way that a time point at which a composition ratio of the numbers of utterances in the conversation has dramatically changed is a separation point. In each of the unit time zones, a plurality of frequently-appearing nouns are extracted in order of descending frequency, and a subject of a conversation in each unit time zone is derived. By matching a plurality of members whose numbers of utterances constitute larger portions of all numbers of utterances in each unit time zone with a subject of a conversation in the unit time zone, optimal personnel assignment for discussing the subject is determined.

Description

  The present invention relates to an optimal staffing method for limiting the number of people and providing a place for discussion in solving a certain subject, and in particular, recording a conversation between three or more persons existing in a certain place. From the above, it is related with the method of discriminating the optimal group composition in the said 3 or more persons.

  There was no prior art on this type of team composition discrimination method. An example of the prior art to be approximated is disclosed in Patent Document 1. According to this patent document 1, the amount of speech is calculated from the act of speaking by a speaker, and the lead is specified.

  An example of the prior art that approximates is disclosed in Non-Patent Document 1. According to this Non-Patent Document 1, the overall composition ratio of the number of utterances corresponding to the “turn” is calculated from the act of speaking by the speaker, and the leader is specified.

JP-A-2005-275536

  A study on the functional evaluation of leaders in consensus building in small community development-Focusing on the "Kannon Community Planning Society" in Kiryu City 2013 Conference p-311

  This prior art is aimed at discriminating the individual who is the leader of the conversation, and is not aimed at discriminating the most suitable group composition (multiple staffing) for discussing a certain subject. It was.

  According to the determination method of Patent Document 1, when the conversation leader unilaterally speaks frequently, it is possible to determine the individual who is the leader of the subject indicated by the content of the utterance. However, since the discussion would not be successful unless another utterance following the content of the utterance was frequently performed, it was not possible to determine the optimal team composition for discussing the subject.

  Also, for example, when a leader is determined from the beginning in a conference, the frequency of utterances by persons other than the leader is relatively remarkably higher because the frequency of the utterance of the leader is significantly higher than that of other utterers. In the case of low, it was extremely difficult to determine who other than the leader was actively involved in the conversation.

  According to the determination method of Non-Patent Document 1, the number of utterances corresponding to “turn” is calculated, and the number of utterances unilaterally uttered can be excluded to accurately determine the leader involved in the discussion. However, it was not possible to determine who could become a leader instead when the leader was absent.

  Leaders are determined according to the subject matter and in a relative relationship with other speakers, so there was important information for members other than the leaders in determining the optimal team composition. So, the focus was only on the technology that discriminates conversation leaders.

  Therefore, the main object of the present invention is to provide the best mix of optimal personnel assignment for discussing a subject from a record of the conversation, even when a person who frequently speaks in a conversation is mixed. It is to provide a method for discriminating.

  In the invention of claim 1, similarly to the leader discriminating method of Non-Patent Document 1, using the data calculated from the recorded conversation, the overall composition ratio of the number of utterances of the speaker corresponding to the “turn”, The degree of involvement of all speakers in the conversation is determined.

  Furthermore, in the invention of claim 1, a plurality of nouns that frequently appear in a certain time zone are extracted from the recorded conversation from the top, and the subject of the conversation in that time zone is derived.

  Furthermore, in the invention of claim 1, the degree of involvement of all the speakers in a certain time zone determined from the recorded conversation and the subject of the conversation in the time zone are adapted to discuss the subject. The correct staffing.

  The invention of claim 3 is dependent on claim 1 and the time zone for extracting a plurality of nouns frequently appearing from the recorded conversation from the top is determined between the time points when the change in the overall composition ratio of the number of utterances in the conversation is large. Is a classification means for classifying as a unit time zone.

  According to this invention, even if a person who frequently speaks in a certain conversation, the overall composition ratio of the number of utterances exchanged in a unit time zone and the subject of the conversation are matched from the record of the conversation. By doing so, it is possible to accurately determine the best mix of the optimal staffing when discussing a specific subject.

  According to the present invention, regardless of whether there is a subject that a person who frequently speaks in a conversation intentionally leads or controls, the subject that is actually discussed is derived according to the involvement of the speaker. Therefore, it is possible to accurately determine the subject corresponding to the optimal staffing for the discussion.

It is a matrix figure used for correspondence with a plurality of members with a large whole composition ratio of the number of utterances in a certain time zone, and the subject of conversation in the said time zone.

  In the left column shown in the matrix diagram of FIG. 1, the transition of the time zone is entered. In the same column, the overall composition ratio of the number of utterances of each member in the time zone is entered. Enter a number of frequently used nouns in the time zone to help you understand the subject of the conversation.

  As shown in FIG. 1, the left column of the matrix is not limited to time and date, but may be any unit as long as it represents a time transition.

  As shown in Fig. 1, in the middle column of the matrix diagram, if you fill in the horizontal composition of the number of utterances of each member, the composition ratio of each member becomes more visible. Only numerical values may be entered.

  As shown in FIG. 1, frequent nouns in a certain time zone may be entered in the right column of the matrix diagram, but the frequent nouns are extracted from the utterances of members having a large overall composition ratio of the number of utterances in the time zone. It is also possible to extract the utterances that are included and enter the summary of the utterance contents.

  INDUSTRIAL APPLICABILITY The present invention can provide a technique for discriminating an optimal group composition for discussing a certain subject from among persons who have recorded conversations, and determining an optimal personnel arrangement of members in a company or an organization. It contributes to.

  Currently, many companies and organizations have a database that records the conversations of members and customers, and the database is expected to grow further in the future, and the need for solutions using such databases is growing. Yes.

  The present invention is a technique for easily discriminating an optimum group configuration corresponding to a certain subject for smooth and efficient communication of members of a company or organization, and the overall composition ratio of the number of utterances in the database. By developing software that has both judgment means for judging the time between changes as a unit time zone and derivation means for deriving a subject in a certain time zone from the database, the industrial applicability has jumped. Increase.

Claims (3)

A method of determining a combination of three or more persons suitable for discussing a specific subject from a conversation between three or more persons existing in a certain place,
Deriving means for deriving the subject of discussion in a certain time zone from data in which conversations between the three or more people are stored as text files, and the number of utterances of each of the three or more people in a certain time zone A method for discriminating the optimum combination of persons, comprising a procedure for recognizing the overall composition ratio of the above and a change in the subject of discussion derived from the deriving means.   In deriving the subject of discussion in the certain time zone, a plurality of nouns that frequently appear in the certain time zone are extracted from the stored data from the top, and the subject of conversation in the time zone is derived from the plurality of nouns. The determination method according to claim 1, further comprising first derivation means.   First division means for dividing the time zone with a point when a large change occurs in the overall composition ratio of the number of utterances is further provided, and the overall composition ratio of the number of utterances in the time zone divided by the first division means The determination method according to claim 1, wherein a plurality of large members and the subject of conversation in the time period derived in claim 2 are matched to determine an optimal personnel assignment for discussing the subject.