JP2003085347A - Device and method for supporting analysis of organization communication - Google Patents

Abstract

PROBLEM TO BE SOLVED: To facilitate the analysis of effects of an event on an organization communication in an organization communication means such as a mailing list. SOLUTION: The interval and contents of an event performed in the operation of a mailing list and others are inputted by an analyzer. An analysis support device analyzes the log of the mailing list using a publicly known analysis method. An analysis result display 400 showing the graph 412 of the time-series variation of analysis result index values before and after the event is provided to users. The quantitative analysis results of the index values are displayed together with an event information area 420 showing the start point 414a and end point 414b of the event and the name and description of the event. Thus, the users can understand how the state of the communication on the mailing list is changed by the effect of the event.

Description

Detailed Description of the Invention

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to an apparatus and method for supporting communication analysis using an information system used for organizational communication such as an electronic bulletin board and a mailing list.

[0002]

2. Description of the Related Art In recent years, as communication means in communities and corporate activities on the Internet, organizational communication means using information technology such as electronic mail systems and WWW-applied mailing lists and bulletin boards have been used. One of the features of the organizational communication means using information technology is that the message log can be stored, and there is a technology for analyzing and applying this message log in the past.

For example, a system has been proposed in which a title, a speaker, a date, a body, and the like among the header information included in a message are stored in a database and a corresponding message log is searched in a natural language (example: JP-A-11-242545). In addition, a technique of applying such header information to network management (path delay analysis, log accumulation, etc.) (for example, Japanese Patent Laid-Open No. 06-059993),
Techniques (eg, Japanese Patent Laid-Open No. 6-259345, Japanese Patent Laid-Open No. 11-15757, Japanese Patent Laid-Open No. 06-062046) that are applied to agents that perform some processing on the receiving side (for example, classification of electronic mail) are also known. There is. However, these technologies are not considered from the viewpoint of organizational communication analysis.

On the other hand, regarding the method itself for analyzing a message log from the viewpoint of organizational communication, an example in which its usefulness was proposed and discussed academically (Advances in
social network analysis: Research in the social an
d behavioral sciences, pp.167-203, Newbury Park, C
A: Sage, 1996 ACM 0-89791-782-0 / 96/04, JCMC 3 (4)
June 1998) and an example published in literature (1998 ACM 0-
89791-282-9 / 88/0324) are known. But,
These analysis methods need to be carried out manually, and the implementation is time-consuming. In addition, these are specialized only for one organization communication, and there is a problem that the method itself needs to be modified depending on the analysis target.

On the other hand, as a device that supports message log analysis from the viewpoint of organizational communication, a technique for visualizing relational information between speakers (eg, Takahashi, Kitayama,
Kaneko: Quantification and Visualization of Organization Awareness in Network Communities, Transactions of Information Processing Society of Japan, Vol.40,
No.11, pp3988-3999, Nov 1999.), a technique for sharing related information (eg, Japanese Patent Laid-Open No. 10-301905), a technique for calculating various indicators including visualization using related information (Pajek:
http://vlado.fmf.uni-lj.si/pub/networks/pajek/def
ault.htm) is known.

[0006]

Although these devices and systems can obtain and display quantitative information about communication within an organization, the quantitative analysis result is displayed in the series of communication. It was not enough for the purpose of analyzing it by associating it with qualitative information such as events and topics that appeared in.

The present invention has been made in view of the above problems, and an analysis in which a quantitative analysis of a communication log of an organizational communication means is associated with qualitative information such as an event or topic related to the organizational communication. It is an object of the present invention to provide an apparatus and method capable of supporting the above.

[0008]

In order to solve the above-mentioned conventional problems, the present invention is an apparatus for supporting analysis of an organizational communication log, which is an analysis target event related to the analysis target organizational communication. Means for accepting the designation of, the means for analyzing the log for the analysis section related to the designated event, and the means for displaying the log analysis result for the analysis section in association with the information of the event. An organizational communication analysis support device is provided.

In a preferred aspect, the apparatus according to the present invention further comprises means for accepting designation of the analysis section associated with the event, and performs log analysis for the analysis section designated by this means.

In another preferred aspect, in the apparatus according to the present invention, a section before and after the event is set as the analysis section related to the event, and the section before and after the event are set. The information of the change of the log analysis result of is presented.

In another preferred aspect, the apparatus according to the present invention receives designation of a plurality of events as events to be analyzed, and displays the log analysis results for the analysis sections of the designated events in an overlapped manner. Means are further provided.

In another preferred aspect, the apparatus according to the present invention is a means for accepting designation of a reference event among a plurality of events to be analyzed, and a log analysis result of each event in each analysis section. The change pattern is further compared with the change pattern of the log analysis result in the reference event analysis section, and means for controlling the display form of the log analysis result of each event based on the comparison result.

In another preferred aspect, the apparatus according to the present invention receives a designation of a plurality of events as an analysis target event, and obtains a log analysis result for an analysis section of each designated event. A means for accepting the designation of the event of interest from among these events, and searching the log analysis result for each event,
Means for obtaining an event having a log analysis result similar to the log analysis result for the event of interest.

In still another preferred aspect, the apparatus according to the present invention receives a designation of a plurality of events as an analysis target event, and obtains a log analysis result for an analysis section of each designated event. It further comprises means for accepting an input of a search condition for the log analysis result, and means for searching the log analysis result for each event to obtain an event having a log analysis result satisfying the search condition.

Further, the device according to the present invention is a device for supporting the analysis of the organizational communication log, and means for accepting the designation of the utterance noticed by the user from the organizational communication log to be analyzed, And a means for displaying the log analysis result of the analysis section in association with the information of the statement.

[0016] In a preferred aspect of the present invention, the means for accepting the designation of the utterance focused by the user displays each utterance included in the log in a tree structure based on the reference relationship. It is characterized in that the designation of the notice of interest is accepted in association with this display.

In still another preferred aspect, the present invention is characterized by comprising means for comparing and displaying log analysis results for respective analysis sections of a plurality of utterances designated as a target of attention.

[0018] In another preferred aspect, the present invention is based on a partial tree starting from a message designated by a user as a target of interest in a tree structure of messages in the log, and an analysis section related to the target message of interest. And means for determining

According to another preferred aspect, the present invention provides a means for obtaining, for each message designated by the user as a target of interest, a log analysis result for an analysis section corresponding to each message and a log analysis result. It further comprises means for accepting an input of a search condition, and means for searching the log analysis result for each of the analysis sections to obtain an analysis section having a log analysis result satisfying the search condition.

In another preferred aspect, the present invention accepts comparative analysis information of log analysis results of a plurality of analysis sections as the search condition, and a plurality of combinations of the log analysis information satisfy the comparative analysis information. It is characterized by comprising means for retrieving a combination of analysis sections.

[0021]

BEST MODE FOR CARRYING OUT THE INVENTION Preferred embodiments of the present invention will be described below with reference to the drawings.

In this embodiment, the organizational communication means electronic mail, mailing list, bulletin board,
This refers to communication composed of multiple parties using organizational communication means to which information technology is applied, such as chat systems and video and audio teleconferencing systems.

In these organizational communications, for example, e-mail, mailing list, bulletin board,
In the case of a chat system or the like, as a record of the communication, information about who communicated with whom, when, and what kind of content is held as text data. Also, in the case of a teleconferencing system such as video and audio, as a record of that communication, in addition to information about who was communicated to whom and when, audio and video streaming data,
Alternatively, the analysis result of the communication content obtained by applying the existing voice recognition technology to the streaming data is held. In this embodiment, it is premised that the organizational communication is recorded in some form as described above.

The following is a description of the organizational communication analysis support device of the present embodiment, taking a mailing list (hereinafter appropriately abbreviated as ML) as an example of the organizational communication means.

FIG. 1 shows a schematic configuration of the apparatus of this embodiment. This device roughly includes a read control unit 10, a storage unit 15, an input control unit 20, an analysis unit 30, a search unit 40, and a display control unit 50. This device is typically realized by executing a program that describes the function of each functional module described below in a computer system.

The read control unit 10 is a functional module for reading an organizational communication group to be analyzed, that is, a communication log from a server that manages a mailing list and the like, and registering it in the storage unit 15, and is provided with an organizational communication group reading unit 12. ing. The input control unit 20 is a functional module for receiving inputs of various instructions and data from a user or the like and performing processing corresponding thereto, and the organizational communication selection unit 22,
An event editing unit 24, an event selecting unit 25, an event analysis section designating unit 26, and a message log analysis section designating unit 28 are provided. The analysis unit 30 is a functional module that actually performs analysis processing on the log to be analyzed, and includes an event analysis unit 32 and a message log analysis unit 3.
4 and. The search unit 40 is a functional module that performs a search process such as a similarity search on the log analysis result by the analysis unit 30, and includes an event analysis result similarity search unit 42 and a message log analysis result similarity search unit 44. The display control unit 50 is a functional module that performs processing for displaying analysis results, search results, and the like on a monitor or the like,
Organizational communication group display means 52, message log display means 54, event and log aggregation result display means 5
6. An event analysis result display means 58 and a message log analysis result display means 59 are provided. The details of each of these parts will be described below.

FIG. 2 is a diagram schematically showing an example of the ML selection screen 100 generated by the organization communication group display means 52 and displayed on a monitor or the like. Starting with the selection screen 100, some schematic diagrams of user interface screens of the apparatus will be illustrated below, but these are all configured using a well-known GUI (graphical user interface) technique, such as a mouse. By operating the pointing device, the desired display item can be selected and the execution instruction of the processing associated therewith can be issued.

On the ML selection screen 100, a list list 110 of one or more MLs which are candidates for analysis is displayed.
Substantial data such as messages of each ML displayed in the list 110 is stored in advance by the organization communication group reading means 12 from the mailing list server or the like in the storage unit 15.
Alternatively, the header information to be displayed in the list 110 may be read first, and the individual M
The substantive data of L may be read in the folder or file in which the message logs of the ML group are stored only when it is needed by the user's selection or the like.

Each horizontal row of the ML list 110 indicates information of one mailing list. The information of each ML includes the identification information (“I
D)), the name ("ML name"), and path information ("Comment log", "Access log", and "User information") indicating the file storage location of the comment log, access log, and user information are displayed. Further, an explanation sentence (“explanation”) regarding the ML is shown. The comment log is a log showing the history of comments made by each user to the ML, and the access log is a log showing the history of access (including only browsing access) made by each user to the ML. The user information file is a file that stores registration information (e-mail address, etc.) of each user registered in the ML. These log and user information files are created and managed by a general ML server.

The ML list 11 shown in FIG.
0 is just an example, and instead, only the ML name may be displayed as a list. Further, when a hierarchical relationship or the like exists between the MLs, a display form that reflects the hierarchical relationship or the like can be used instead of the list list.

When the user operates the mouse or the like to move the cursor to the desired ML row on the list list 110 and gives a selection instruction by clicking or the like, the organizational communication selection means 22 accepts the selection of the ML and displays it. In the area 120, the name of the ML, the storage location of the utterance log and the user information, the explanation, etc. are displayed in more detail than the list list 110. The user can read this display area 120 and confirm whether the ML is the desired ML.

The ML message log is generally composed of a plurality of utterance data. FIG. 3 shows an example of the data structure of the message log data, and one row of the table shown in this figure is the statement data of one statement. The speech data is
The raw data generally includes information such as utterance ID, utterance source ID, speaker name, utterance date and time, utterance title, and utterance content. Here, the "speech ID" is identification information of the speech itself. The “speaker ID” is the ID of the message that is the target of the message (that is, the message referred to by the message). That is, for a comment B such as a comment sent to a certain comment A, the ID of the comment A is set as the comment source ID. Since the utterance that is the origin of a series of topics does not have such a reference utterance, special identification information (“-1” in the example of FIG. 3) not used for the utterance ID is set as the utterance source ID. The "speaker name" is information such as the name of the user who made the remark, and the "remark date and time" is the date and time (year, month, day, etc.) at which the remark was made. And the "speaking title" is the title given by the speaker to the statement,
“Comment content” is the text of the message.

The raw data of the message log held by the ML server itself is shown in FIG. 3, and it may be complicated to handle the data during the communication analysis process. Therefore, it is also preferable to convert this raw data into an easy-to-handle data structure that conforms to the purpose of analysis. For example, as an example, the raw data of the log is divided into a table showing the relationship between the utterances shown in FIG. 4 and a table relating to the content of each utterance shown in FIG. In the table of FIG. 4, a unique speaker ID is used instead of the speaker name. For each utterance, the ID of the speaker and the ID of the speaker who is the source of the utterance are registered, and the relationship between the speakers can be easily detected.

ML as one of the information of the ML
Additional information such as operator name, start time, participation condition, purpose, and attribute value (profile) of the participating user is prepared on the ML server side, which is read and stored in this device, and is stored in the ML server. It can also be used as a target or a flag. In addition, as information regarding the ML, as shown in FIG. 6, attribute information of each participating user of the ML is prepared,
It is also possible to incorporate this into the apparatus of this embodiment and use it for analysis or the like.

The above is an example of a mailing list,
In the bulletin board and chat system on the Web page, who, where, and when
Information about whether or not page b has been browsed may be retained as an access log, and this access log can be treated as an analysis target together with the message log. The access log of FIG. 7 shows the UR of the accessed page.
L, access time, and accessor name are registered. It is also preferable to convert the raw data of the access log into a data structure that is easy to handle in analysis processing as shown in FIG. In the data structure example of FIG. 8, in addition to the data items shown in FIG. 7, an “access ID” is assigned to each access target page, and the accessor is indicated by the user ID instead of the name. Has registered "access location" information that indicates the sites (including mirror sites) actually accessed by. By representing the page to be accessed and the accessing person by ID, it is possible to expect the efficiency of the analysis process.

Next, the event editing means 24, the event selecting means 25, and the event and log totalization result displaying means 26 will be described with reference to FIG. In addition, in FIG.
3 schematically shows an example of an event editing user interface screen generated by the event and log aggregation result display means 26.

The event editing means 24 is a means for providing a user interface for setting various kinds of information about the event related to the ML selected as the analysis target. As an event, for example, a web page related to a request for participation in ML, an announcement of a request for participation in ML by direct e-mail delivery, and a We linking with ML.
There are actions to be taken from the ML management side to the participating user side in the operation of ML, such as a questionnaire survey on the topic of ML on page b. In addition, a characteristic utterance made by the ML operator or the participant on the ML can be regarded as a kind of event.

An analyst who intends to analyze the log has an event editing screen 20 of FIG. 9 displayed in response to the selection of ML.
At 0, an event can be added, changed, or deleted from the ML via the event editing unit 24. Multiple events can be set for one ML. Each event is, for example, "event name",
It can be represented by “event section” and “detailed description of event”. Here, the "event name" is a name for identifying the event, and the "detailed description of the event" is a descriptive text of the content of the event. "Event section"
Is information indicating a period in which the event is performed, and is typically represented by a start date and time and an end date and time. Of course, there is an event at one time point without a time width, but in that case, only the start date and time of the "event section" need be set. The analyst inputs such information into the event edit screen 20.
The contents of the event can be set by inputting each input field of the event information input area 220 of 0. Then, the event registration button 23 is clicked.
When 0 is pressed, an “event” having the input information in the area 220 as attribute information is registered in this device.

A list of events registered in the ML in this way is displayed in the event list list 240. When a desired event is selected on the list 240 with a mouse or the like and the edit button 250 is clicked, the registration information of the event is displayed in the input area 220, and the registration information can be re-edited. . Also, by selecting a desired event on this list 240 and clicking the delete button 260, the event can be deleted.

Further, the log total display area 210 of the edit screen 200 displays the log total result of the ML.
In the example of FIG. 9, a time-series graph (broken line) of the message log aggregation result is shown as the log aggregation result. This time-series graph shows the time-series changes in the aggregated values of the aggregated items corresponding to the conditions input by the analyst. here,
The condition is represented by a combination of values selected by the analyst in each of the “index” column 212, the “time unit” column 214, and the “flag” column 216. The “index” is a type prepared in advance for the items to be aggregated in the message log. For example, for ML, the number of remarks, the number of remarks (replies) to the remarks, the number of speakers, the number of new speakers, the number of subscribers, And so on. Further, the "flag" is a classification attribute prepared in advance for the participating members (users) of the ML, and may be, for example, the type of job, the department to which the user belongs, or the age group. The target of aggregation is defined by the combination of these “index” and “flag”. The “time unit” is a designation of a unit of a time axis delimiter when a time series graph is displayed. Also, the display time range displayed in the total display area 210 can be changed according to the instruction of the analyst. Further, not only the graph of such a time series change but also the numerical value of the aggregation result can be displayed together.

Further, in the total display area 210, the event editing area 2 is displayed in accordance with the time series graph of total items.
Information on the section of the event set to 20 is displayed.
The illustrated example is for an event with a certain time width, and lines indicating the start time and end time of the event are shown. By looking at the relationship between the event section and the time-series graph, the analyst can see the effect of the event on the target aggregate item.

The above is the case of the mailing list,
As for the bulletin board and the chat system, the above-mentioned access log information is also included in the bulletin board and chat system, so that it is also possible to display the same tabulation result for this access log.

Next, the event selecting means 25 and the event analysis section specifying means 26 will be described with reference to FIG. In addition, FIG. 10 is for inputting conditions of each of these means.
It is a figure which shows typically an example of the event selection screen 300 which the event and log totaling result display means 56 provides. The analyst can call the selection screen 300 by selecting a desired ML from the list of MLs and inputting a display instruction of the event selection screen.

In the registered event list list 310 of the event selection screen 300, a list of events registered by the editing means 24 is displayed for the selected ML. The analyst can select the event selection means 25 from this list 3
The event selected by mouse operation or the like on 10 is recognized as the event to be analyzed. Further, the event analysis section designation means 26 recognizes the analysis section input in the analysis section designation field 315 as the analysis section of the analysis target event. In this example, the section length on one side of the event section is input as the event analysis section. Therefore, the event analysis section of the input length is set on each of the front side and the rear side of the event section. That is, in the case of an event having no time width, as shown in FIG. 11A, an analysis section corresponding to the section length is set before and after the time point of the event, and in the case of an event having a time width, FIG. As shown in (b), an analysis section having the section length is set before and after the event section. Note that, hereinafter, the analysis section before the event section (or the time point of the event) is referred to as a “previous section”, and the analysis section after the event section is referred to as a “post section” as necessary.

When an event is selected and an analysis section is designated and the analysis section registration button 320 is pressed by a mouse operation or the like, the event is registered as an analysis target, and the section is registered as an analysis section corresponding to the event. .
Events registered as analysis targets in this way are
It is displayed in the analysis target event list list 330. In order to remove the event once selected as the analysis target from the analysis target, the event may be selected on the list 330 and the analysis target deletion button 340 may be pressed.

Next, the analysis processing by the event analysis means 32 for the analysis target event thus selected will be described. As an analysis method, the comparison of the aggregated values of various prepared aggregated items before and after the event is the basis.
For example, the most basic method for analyzing ML message logs is to compare the values of simple indicators such as the number of utterances, the number of replies, the number of speakers, the number of new participants, and the number of subscribers in the pre-event section and the post-event section. Analysis is possible. That is, it is calculated whether the respective index values (or the one selected by the analyst) have increased or decreased in the front section and the rear section of the event, or the degree of the increase or decrease (change rate, etc.). And present it to the analyst. Further, it is also possible to calculate a change in those index values from the start point of the previous section through the event section to the end point of the subsequent section. By this analysis, it is possible to grasp how the event selected by the event selecting means 25 affects activation or inactivation of the ML as a transition of each index.

As another example, a network analysis, which is one of mathematical social science analysis methods, is performed on the message log, and the indexes calculated from the network analysis are pre-event intervals and post-event intervals. It is also possible to display, as an analysis result, how the transition is made. The information for this network analysis is
Alternatively, it can be obtained from the reference relationship between speakers. Since the information about the reference relation of the utterance is included in the raw data of the message log (FIG. 3), the network analysis can be performed on the raw data, but the utterance as shown in FIG. 4 can be obtained from the raw data. If the relational information of is previously created, it is possible to quickly analyze using this relational information.

FIG. 12 shows an example of a data structure of relational information between speakers (relational information about how much one speaker speaks to another speaker) created in this network analysis. In this example, the sender's speaker ID is in the left column.
Therefore, the speaker ID of the destination is registered in the center column, and the frequency of the speech directed from the source to the destination is registered in the right column. This relation information can be automatically calculated from the message log in the event analysis section.

The network analysis index can be calculated as follows based on this relation information. For example, “Density”, which is a typical index for network analysis, should be calculated as the ratio of the number of one-way utterance relationships that actually occurred in the analysis section to the theoretical total number of possible utterance relationships in that section. You can Here, the one-way utterance relationship is a relationship when two or more speakers have at least one utterance directed from one to the other. If there is a speaker pair having such a relationship, the one-way statement relationship is counted as 1 for the pair. Even if the speaker pair has two-way utterances or a plurality of utterances, only one is counted as the one-way utterance relationship.

The "direct coupling degree", which is an index similar to this, focuses on the two-way speaker relationship instead of the one-way speaker relationship. That is, in this case, the relationship between the speakers is counted only when there is bidirectional communication between the speakers, and the ratio of the count result to the theoretical total number of relationships is directly calculated. The degree of coupling.

For example, by calculating the density and the degree of direct coupling for the front section, the event section, and the rear section of the event and displaying the transition in the form of a graph or the like, the event for the communication occurring in the ML can be displayed. It can be useful for analyzing impacts. Here, by displaying the transitions of both the density and the direct coupling degree, it becomes possible to perform a comparative analysis in which the directionality of the utterance relationship is taken into consideration.

The result of the analysis by the event analysis means 32 is displayed on the monitor or the like by the event analysis result display means 58. FIG. 13 schematically shows an example of the display form of the event analysis result. In the analysis result screen 400 of FIG. 13, the transition before and after the event of the index value of the “new participant” of the “sales” type user for ML is displayed in the analysis result area 410 as a graph 412.
The broken lines 416a and 416b are the start points of the previous section,
Indicates the end point of the latter section. Since this graph 412 is displayed together with the event start point 414a and the event end point 414b, it is easy to understand the transition of the index value before and after the event. In addition, in the event information area 420,
The registered information about the event related to the analysis result is displayed, and the analyst can easily understand the content of the analysis target from this display. In this example, the transition of the index value of the so-called simple index, which is the number of new participants, is displayed, but it is of course possible to display the transition of the index value obtained by the network analysis. The analyst can appropriately specify the analysis target desired by himself or herself and obtain the analysis result by switching the index type of the analysis target specified in the index column 416 or the attribute of the user of the analysis target specified in the flag column 418. You can

Further, instead of the transition graph of the index value as in the example of FIG. 13, a network diagram of the relationship between speakers before and after the event can be displayed as the analysis result. 14
Shows an example of this display. In this example, the analysis result display column 43
At 0, a network diagram 430a of the speaker relationship before the start of the event and a network diagram 430b of the speaker relationship after the end of the event are displayed side by side for easy comparison. Speakers (participating users) of the mailing list (ML) are shown in the network diagrams 430a and 430b.
A node 432 representing is shown. Then, in the period covered by each network diagram, when there is a speech directed from another speaker on the ML to another speaker, an edge 434 with an arrow pointing from the former to the latter is displayed. . When there is bidirectional speech exchange between speakers, an edge of a bidirectional arrow is displayed. By creating and displaying such a network diagram before and after the event, the analyst can grasp the influence of the event on the communication in the ML. For example, in the example of FIG. 14, the network diagram 430b after the end of the event has more nodes 432 and edges 434 than the network diagram 430a before the start of the event, so that communication is activated by the event to be analyzed. You can easily understand what you have done.

Although the above is an example of the analysis result for one event, a plurality of events may be selected as the analysis target by the event selecting means 25. In this case, it is of course possible to individually display the analysis results of each analysis target in separate display windows or display areas, but instead of this, as shown in FIG. It is also possible to display the results in an overlapping manner (hereinafter, such a display is referred to as an overlap display). In the example of FIG. 15, transition graphs 442a to 442c of index values (the number of new ML participants in the sales job category in this example) before and after a plurality of events are superimposed and displayed in the analysis result display area 440.

Further, on the display screen 400, the analysis target event list area 450 displays a list of events registered as analysis targets by the event selecting means 25. The analyst selects an event for which the analysis result is to be displayed in the overlap display from the list by operating the mouse or the like, and displays the overlap display button 455.
By pressing with the mouse or the like, a transition graph of the analysis results of the selected events is displayed in the analysis result display area 44.
It can be displayed on top of zero.

Here, when comparing and displaying the analysis results of a plurality of events, it is easier to see the overlapped analysis results when the widths of these event analysis sections are equal.
It is often desirable for practical use. However, it is general that each event is individually registered and the widths of the event section and the analysis section are not equal. Therefore, it is also preferable to provide a means for aligning the analysis section width (here, the time width from the start point of the pre-event section to the end point of the post-event section) for a plurality of overlapping display target events. The event analysis section resetting instruction unit 460 on the display screen 400 is an example of a user interface of this means. In this example, in order to make the analysis interval width uniform, 2
Offers two ways. One is to set one of a plurality of events selected as overlap display targets as a reference, and reset the analysis section widths of other events to match the analysis section width of the reference event. The reference event is
Ask the analyst to make a selection. The other method is to obtain the average value of the analysis interval widths of a plurality of events selected as overlap display targets and align the analysis interval widths of the respective events with the average value. The event analysis section reset instruction unit 460 is a user interface for selecting a desired one from these two methods. Of course, the illustrated two methods are merely examples, and other methods can be used to make the event analysis section widths uniform. It is also preferable to provide a mode in which the analysis results of a plurality of events are displayed in an overlapped manner without performing such resetting processing for making the analysis section widths uniform.

Further, in the overlap display of the analysis results of a plurality of events, the analyst designates a certain event as a reference, and the increase / decrease of the attention index value in the analysis section of the reference event and the attention of other events. It is also preferable to compare the increase and decrease of the index value in the analysis section and display the analysis result according to the comparison result.

For example, the increase / decrease of the reference index value of the reference event and the increase / decrease of the reference index value of another event are binary-compared, and the index value transition graph of the event showing the increase / decrease different from the reference event is highlighted. Processing such as is conceivable. The binary comparison of index value increase / decrease is a comparison focusing only on whether the index value has increased or decreased before and after the event. For example, in the example of FIG. 15, the index value of the reference event (graph 442a) increases in the analysis section, while “event 5”
Then, since the index value is decreasing in the analysis section, the transition graph 442c of "event 5" is highlighted 445.
Has been applied.

Further, a display method in which attention is paid to the range of increase / decrease of the index value in the analysis section can be considered. That is, for example, the index value increase / decrease range of each event is compared with the index value increase / decrease range of the reference event, and the transition graph of the event showing the increase / decrease range larger than the index value increase / decrease range of the reference event is highlighted. Instead of designating the reference event, the increment / decrement width itself serving as a threshold may be designated by a numerical value.

The comparison method designation section 470 of the display screen 400 illustrated in FIG. 15 is a user interface for selecting these two methods for highlighting. In addition,
The comparison method for determining the highlighted display is not limited to the above two methods.

As described above, the event analysis result display means 58 can display the event of the mailing list and the quantitative analysis result of the mailing list before and after this event in association with each other. This display allows the analyst to easily analyze what event is activating the mailing list and how.

Next, the event analysis result similarity search means 42
Will be described. The event analysis result similarity search unit 42 is a unit for searching for an analysis result similar to the event analysis result by the event analysis unit 32. FIG. 16 shows an example of a user interface screen for inputting instructions for processing by the search means 42 and displaying search results. In the search target display area 510 of the similarity search screen 500 shown in FIG. 16, each analyzed event is displayed as a list of similar search targets. When a similar search target event is selected from this list display by a mouse operation or the like and the search start button 520 is clicked, a list of events having analysis results similar to the selected event is displayed in the search result display area 530. .

Whether or not the analysis results of the events are similar to each other can be determined based on, for example, a binary pattern of whether the target index value is increasing or decreasing in each event analysis section. In this case, when the index value increases in the analysis section of the event designated as the similar search target, the similar search searches for the event whose index value increases in the analysis section. Further, not only binary increase / decrease but also attention is paid to a change rate of increase or decrease, and a process of searching an event whose change rate is within a predetermined difference from the similar search target event as a similar event may be considered.

Further, when a plurality of indexes are obtained in the log analysis, a combination pattern of whether each index value increases or decreases in the analysis section is obtained for each event. For example, the pattern is such that the index A increases, the index B increases, the index C decreases, and so on. In this case, this combination pattern is searched for as an “similar” event if it is close to an event close to the combination pattern of the similar search target events. In this case, the combination pattern is regarded as a coordinate point in the n-dimensional space (n is the number of indices), for example, with “increase” being a numerical value “1”, “decrease” being a numerical value “−1”, and the like search target event It suffices to perform processing such as obtaining an event at a distance equal to or smaller than a predetermined value from the corresponding coordinate point as a similar event.

Further, as a user interface screen for similarity search, a screen exemplified in FIG. 17 can be considered. The example of FIG. 16 searches for an event having an analysis result similar to the analysis result of the selected event.
In the example of, the condition about the analysis result is input and the event satisfying the condition is searched. In the illustrated example, a condition setting area 610 for specifying conditions is provided. In this example, a plurality of condition items 612 can be designated, and the condition content 614 can be designated for each of the items 612. The condition items 612 include a section of interest in the search (event section, entire analysis section, previous section, subsequent section, etc.), an index of interest (the number of utterances, the number of speakers, etc.), a time range to be searched, and the like. For indicators to focus on,
You can specify multiple condition items. As the condition content, the length of the section, the start point and the end point of the time range, etc. can be specified. As for the condition content regarding the index, the same condition as that of the comparison method illustrated in FIG. 15 can be designated, such as comparison of increase / decrease binary values regarding the index value or comparison of increase / decrease range. Also, a logical relationship 616 when applying each of the condition items 612 can be designated. The logical relationship 616 is logical AND, O
Various logical relationships such as R and exclusive OR can be designated.

When a desired condition is input (or selected from preset options) in this condition designation area 610 and a search start button 620 is clicked, the area 61 is clicked.
A list of events that satisfy the condition shown in 0 is displayed in the similar search result display area 630.

In the above example, "event" such as an event having an analysis result similar to the selected event or an event satisfying the specified condition is obtained as the search result. On the other hand, it is conceivable to perform a search with the entire time series of index values of the mailing list log including the part where the event analysis section is not set as the search range. In this case, for example, the time series of the index value is sequentially examined from the beginning, and processing such as extracting a part similar to the increase or decrease of the index value in the analysis section of the selected event or a part satisfying the specified condition is performed. I'll do it. In this case, as the search result, a display form such as a list display of the start point and the end point of the portion extracted by satisfying the condition or the like can be considered.
In addition, a display form in which a portion satisfying the condition or the like is highlighted in a time series graph such as the log aggregation result display area 210 in FIG. With such a display, it is possible to perform an analysis in the opposite direction to that of investigating the influence of an event such as what caused the change in the index value of the extracted portion and what caused the change. The search option designation unit 52 shown in the examples of FIGS.
Reference numerals 5 and 625 are user interfaces for inputting a selection instruction as to whether only the event is to be searched or time-series analysis data of the entire log is to be searched as described here.

Further, in the above example, the similarity search is performed by paying attention to the time series change of the index value obtained by the analysis, but a network diagram of the speaker relationship which is a kind of analysis result (for example, FIG. 14). Network diagram 430a, 43
(See 0b), a method of performing similarity search processing is also possible. In this case, by indexing the structural characteristics of the network diagram using a well-known method, it is possible to use the method similar to the similar search method for simple indexes such as the number of speakers, which has been described so far, between speakers before and after the event. It is possible to perform a similarity search for changes in the structural properties of relationships. In addition,
A general algorithm such as graph theory may be used to index the structural characteristics of the network diagram.

As described above, the configuration of the apparatus according to the embodiment is taken as an example in the case where the “event” during the mailing list implementation period and the quantitative analysis result obtained by the analysis of the message log of the mailing list are associated and presented to the analyst. Described the function. Next, as another example, an example will be described in which the topic of a statement appearing in the message log is focused on and the quantitative analysis result of the log is displayed in relation to the topic. In this process,
Organizational communication group reading means 12, organizational communication selecting means 22, message log display means 5
4, message log analysis section designation means 28, message log analysis means 34, message log analysis result display means 5
9, and each module of the message log analysis result similarity search means 44 is used.

In this example, the organization communication group reading means 12 reads the organization communication log group such as a mailing list, and the analyst uses the organization communication selecting means 22 to select a log to be analyzed. Is similar to the case of the above event analysis.

This example differs from the event analysis in that the event analysis is qualitative in analyzing a conceptual “event” that is not necessarily included in the log as data. In contrast to the viewpoint, in this example, individual statements included in the log are used as a qualitative viewpoint for analysis. In the above example, the analysis section is set based on the event, but in this example, the analysis section is set and analyzed based on the statement selected by the analyst. This makes it possible to analyze changes in organizational communication (activation, inactivation, etc.) due to the statement and a series of statements derived from it.

FIG. 18 schematically shows an example of the analysis section designation screen 700 for setting the analysis section in this example. This screen 700 shows the message log display means 5.
4 is generated. On this screen, the message log display area 710 displays a group of messages included in the message log of the mailing list (ML) selected as an analysis target by the organizational communication selection means 22. In the example of FIG. 18, area 710 displays bibliographical information such as the title of a message, the name of the speaker, the mail address of the speaker, and the date and time of the message for each message. If the number of utterances is too large to be displayed all at once in the area 710, all the utterances included in the log can be displayed by scrolling the display range. This area 7
When the analyst selects one utterance by operating the mouse on 10, the utterance content (that is, the utterance text) of the selected utterance is displayed in the utterance content area 720. By checking the content of this message, it is possible to determine whether the selected message is the desired one. The analyst repeats selection in the area 710 and confirmation of the content of the statement in the area 720 until the statement desired to be analyzed is found.

Note that, in the message log display area 710, a tree display (so-called thread display) based on the reference relationship of utterances is displayed as shown in FIG. 19 instead of the simple list format display shown in FIG. It is also possible to do so. In this display, a reply to this comment, a further reply to the reply, and the like are shown in a tree structure, starting from the comment that started the topic. In this tree structure, it is considered that one tree rooted in a utterance without a reference (start of a topic) is associated with one topic. In the example of FIG. 19, trees for three topics are shown. In addition, a partial tree (also called “branch”) consisting of a group of lower words starting from a node of a certain statement in one tree is regarded as showing a partial topic related to a large topic of the entire tree. be able to. By displaying the message group in the message log in a tree structure in this way, it becomes easier for the analyst to find the message of interest.

When the desired utterance can be selected by referring to the areas 710 and 720, the analyst designates the analysis section based on the utterance.

The analysis section may be, for example, a section having a predetermined width period before and after the utterance. In this case, when the analyst selects the statement, the analysis section is automatically determined. Of course, the analyst can also specify the period width before and after the statement here.

It is also conceivable that the analysis section is a distribution period of the utterance dates and times of the utterance groups included in the tree or the partial tree starting from the selected utterance in the tree structure of the utterances in the log. Also in this case, when the analyst selects the statement, the analysis section is automatically determined.

Since the same or similar topic may be spread over a plurality of utterance trees in the log, the analyst is asked to select a plurality of utterances, and each of the utterances is used as a starting point. It is also possible to collectively set the distribution range of the utterances of the partial tree group as one analysis section. When an analysis section is designated for a topic composed of a plurality of partial trees in this way, in the message list and the tree structure displayed in the message log display area 710, the messages of those common topics are not always displayed continuously. is not. Therefore, in this case, in the area 710, the starting point of each partial tree is sequentially selected, and the analysis section designation end button (not shown) or the like is designated by a mouse or the like. It is desirable to be able to explicitly determine the section.

Further, the analyst, on the analysis section information input area 730, for the analysis section thus determined,
You will be asked to enter a name for identifying the section (“Message log analysis section name”), a detailed description of the section that cannot be represented by just the name (“Message log analysis section description”), and so on. These pieces of input information are treated as attribute information of the analysis section. Further, in each analysis section, identification information (“message log analysis section ID”) is set for mutual identification on the system of the present apparatus, and this apparatus automatically sets a unique value. Give.

In this way, when the analysis section is specified by selecting the utterance and the input of the attribute information by the analysis section is finished, the analyst analyzes the analysis section by pressing the registration button 740 with a mouse or the like. It can be registered as a processing target.

A plurality of analysis sections that are the target of the analysis processing can be registered. A plurality of registered analysis sections are displayed as a list in the registration section display area 750. To delete the registered analysis section, select the section on the list display and press the delete button 760.

When one analysis section is selected on the registered section display area 750 by a mouse operation or the like and the single analysis button 770 is pressed, the message log analysis unit 34 performs the single analysis process on the selected section. Be seen. In this single analysis, the time-series changes such as the simple aggregation index and the network analysis index similar to those in the event analysis described above are required for the selected message analysis section.

When two analysis sections are selected on the registered section display area 750 and the comparison and analysis button 780 is pressed, the message log analysis unit 34 performs comparison and analysis processing on the selected sections. In a comparative analysis,
For example, a time-series change of a simple tabulation index or a network analysis index is obtained for each selected analysis section. Also,
It is also conceivable to obtain a feature amount (increase / decrease or change rate at that time) of the time-series change of each of these analysis sections and compare and display them.

The result of such analysis processing by the message log analysis means 34 is displayed on the monitor or the like by the message log analysis result display means 59. As a display method, in the case of single analysis of one analysis section, it is conceivable to display the index in the analysis section as a time series graph or display the speaker relationship of the analysis section as a single network diagram. . In the case of a comparative analysis of two analysis sections, time series graphs of changes in indicators by simple aggregation or network analysis in the two analysis sections are displayed side by side so that they can be compared with each other, and It is conceivable to create a network diagram of the relationships between speakers for the messages that occurred in a section and compare and display them. The display screen of the results of the single analysis and the comparative analysis described here and the operation method thereof may be the same as the case of the display by the event analysis result display means 58 in the case of the above-described event analysis (see, for example, FIGS. reference). in this case,
In addition to displaying the analysis results in a graph, the information (title, speaker name, utterance content, etc.) of the utterance that is the basis of the analysis section is displayed.

As described above, according to the present embodiment, the content (or topic) of the message in the message log and the result of the quantitative analysis of the message log analysis section are presented to the analyst in association with each other. can do. This makes it possible to easily analyze what utterance content activates (or deactivates) the ML.

Message log analysis result similarity search means 44
Is a means for performing similarity search similar to the event analysis result similarity search means 42 on the analysis result of each analysis section by the message log analysis means 34.

For example, in the similarity search for the result of a single analysis, the analyst designates one analysis section as a reference for the similar search, and the index value increase / decrease (or change) in the section and the similar index value increase / decrease (or change). Alternatively, it is possible to perform processing such as searching for an analysis section indicating a change) or designating a search condition for one or more indexes and searching for an analysis section having an index value that satisfies this search condition. As the search condition, it is also possible to use a condition such that the index value is increased (or decreased) by a designated width. In addition,
The analysis section that serves as a reference for similarity search is a continuous time range based on the statement selected by the analyst, or a partial tree (or group of statements) of statements starting from the statement (or group of statements) selected by the analyst. Can be specified.

A similarity search can also be performed on the comparative analysis results of the above-mentioned two analysis sections. In this case, it is conceivable that the search condition is the increase or decrease of a single index or a combination of change patterns in the two analysis sections. In this case, for example, if you specify two analysis intervals that are the criteria for the search,
A pair of other analysis sections having similar comparative analysis results to those of the two sections is searched. It is also possible to specify a combination of single or multiple index increase / decrease, change, search range, time unit, etc. as a search condition and search for a combination of analysis sections having an analysis result that matches this condition.

Further, in the similarity search of the network diagram, it is also possible to perform a similarity search of the message log analysis section having the corresponding transition of the index, on condition that the transition of the value in which the structural characteristic of the network diagram is indexed. It should be noted that the search screen and the operation method for the single analysis and the comparative analysis may be the same as those in the example of the event analysis result display means 58 described above.

As described above, according to the present embodiment, an event related to a mailing list or a statement appearing in the mailing list is designated, and the analysis result of the mailing list log is associated with qualitative information such as the event or the content of the statement. Can be presented to the user. This allows the user to easily understand how the communication on the mailing list has changed due to such events and statements.

As a result, it is possible to assist the analyst in various analyzes of organizational communication using the organizational communication technique that is routinely performed. This enables the operation of organizational communication based on the analysis result. Organizational communication operation based on the analysis results means, for example, referring to the factors of activated organizational communication as a reference for the operation of another organizational communication, or identifying and improving factors of inactive organizational communication. It is to be.

Although the mailing list has been taken as an example in the above, the device of this embodiment is of course applicable to analysis support of organizational communication using other means such as a bulletin board.

[Brief description of drawings]

FIG. 1 is a diagram showing an overall configuration of an apparatus according to an embodiment.

FIG. 2 is a diagram schematically showing an example of a user interface screen for selecting a mailing list to be analyzed.

FIG. 3 is a diagram showing an example of comment log raw data of a mailing list.

FIG. 4 is an example of a result obtained by converting raw data of a comment log into a form suitable for analysis processing.

FIG. 5 is an example of a result obtained by converting raw data of a comment log into a form suitable for analysis processing.

FIG. 6 is a diagram showing an example of user information.

FIG. 7 is a diagram showing an example of raw access log data on a bulletin board.

FIG. 8 is a diagram showing an example of a result of converting raw access log data on a bulletin board into a form suitable for analysis processing.

FIG. 9 is a diagram schematically showing an example of a user interface screen for event registration.

FIG. 10 is a diagram schematically showing an example of a user interface screen for selecting an event to be analyzed.

FIG. 11 is a diagram for explaining a configuration example of an event analysis section.

FIG. 12 is a diagram showing an example of inter-speaker relationship data.

FIG. 13 is a diagram schematically illustrating an example of a display screen of a log analysis result regarding an event.

FIG. 14 is a diagram schematically showing another example of a display screen of a log analysis result regarding an event.

FIG. 15 is a diagram schematically showing a display example in which analysis results of a plurality of events are overlapped and displayed.

FIG. 16 is a diagram showing an example of a user interface screen for designating a similar search for an event analysis result and displaying a result of the similar search.

FIG. 17 is a diagram showing an example of a user interface screen for designating a similar search for an event analysis result and displaying a result of the similar search.

FIG. 18 is a diagram showing an example of a user interface screen for setting processing of an analysis section based on comments on a mailing list.

FIG. 19 is a diagram for explaining an example of a display form of a group of messages on a mailing list.

[Explanation of symbols]

10 read control unit, 12 organizational communication group reading unit, 15 storage unit, 20 input control unit, 22 organizational communication selection unit, 24 event editing unit, 25 event selection unit, 26 event analysis section designation unit, 28 message log analysis section designation Means, 30
Analysis unit, 32 event analysis unit, 34 message log analysis unit, 40 search unit, 42 event analysis result similarity search unit, 44 message log analysis result similarity search unit, 50 display control unit, 52 organization communication group display unit, 54 message log Display means, 56 event and log aggregation result display means, 58 event analysis result display means, 59 message log analysis result display means, 400 analysis result screen, 410 analysis result area,
412 (temporal transition of index value) graph, 414a event start point, 414b event end point, 420 event information area.

   ─────────────────────────────────────────────────── ─── Continued front page    (72) Inventor Ken Yoshioka             430 Green, Sakai, Nakai-cho, Ashigaragami-gun, Kanagawa Prefecture             Inside of Fuji Xerox Co., Ltd. (72) Inventor Chito Hayashi             430 Green, Sakai, Nakai-cho, Ashigaragami-gun, Kanagawa Prefecture             Inside of Fuji Xerox Co., Ltd. (72) Inventor Nobuhiro Yamazaki             430 Green, Sakai, Nakai-cho, Ashigaragami-gun, Kanagawa Prefecture             Inside of Fuji Xerox Co., Ltd.

Claims (17)

[Claims] 1. A device for supporting analysis of a log of organizational communication, said means for accepting designation of an analysis target event relating to the analysis target organizational communication, and said log for the analysis section relating to the designated event. An organizational communication analysis support device comprising: means for performing analysis of the above; and means for displaying the log analysis result for the analysis section in association with the event information. 2. The organizational communication analysis support device according to claim 1, further comprising means for accepting designation of the analysis section related to the event, and performing log analysis for the analysis section designated by this means. 3. A section before and after the event is set as the analysis section related to the event, and information on a change in the log analysis result in the preceding section and the subsequent section is presented. The organizational communication analysis support device according to claim 1, characterized in that: 4. The organizational communication analysis support according to claim 1, further comprising means for accepting designation of a plurality of events as an analysis target event and overlappingly displaying a log analysis result regarding an analysis section of each of the designated events. apparatus. 5. A means for accepting designation of a reference event among a plurality of events to be analyzed, a change pattern of a log analysis result in each analysis section of each of these events, and a log in the analysis section of the reference event. The organizational communication analysis support apparatus according to claim 4, further comprising: a unit that compares a change pattern of the analysis result and controls a display form of the log analysis result of each event based on the comparison result. 6. A means for accepting designation of a plurality of events as events to be analyzed, obtaining a log analysis result for an analysis section of each of the designated events, and means for accepting designation of an attention event from the respective events. 2. The organizational communication analysis support according to claim 1, further comprising: a means for searching the log analysis result for each event to obtain an event having a log analysis result similar to the log analysis result for the event of interest. apparatus. 7. A means for accepting designation of a plurality of events as an analysis target event, a means for obtaining a log analysis result for an analysis section of each designated event, and a means for accepting an input of a search condition for the log analysis result. 2. The organizational communication analysis support device according to claim 1, further comprising means for searching the log analysis result for each of the events to obtain an event having a log analysis result satisfying the search condition. 8. An apparatus for supporting analysis of a log of organizational communication, comprising means for accepting a designation of a message focused on by a user from the log of the organizational communication to be analyzed, and analysis relating to the designated message. An organizational communication analysis support device comprising: means for analyzing the log for a section; and means for displaying a log analysis result for the analysis section in association with information about the statement. 9. The means for accepting the designation of the utterance focused by the user, displays the utterances included in the log in a tree structure based on the reference relationship, and displays the utterance in association with the display. 9. The organizational communication analysis support device according to claim 8, wherein designation of a remark of interest is accepted. 10. The organizational communication analysis support device according to claim 8, further comprising means for comparing and displaying log analysis results for respective analysis sections of a plurality of utterances designated as attention targets. 11. The method according to claim 8, further comprising means for determining an analysis section related to the remarked target utterance based on a partial tree whose origin is a remark specified by the user as a remarked target in the tree structure of remarks in the log. Organization communication analysis support device. 12. A means for obtaining a log analysis result for an analysis section corresponding to each utterance specified by the user for each utterance, and a means for accepting an input of a search condition for the log analysis result, 9. The organizational communication analysis support device according to claim 8, further comprising means for searching the log analysis result for each analysis section to obtain an analysis section having a log analysis result satisfying the search condition. 13. A means for receiving comparative analysis information of log analysis results of a plurality of analysis sections as the search condition and searching for a combination of a plurality of analysis sections whose combination of log analysis information satisfies the comparative analysis information. 13. The organizational communication analysis support device according to claim 12, wherein: 14. A method of supporting analysis of a log of organizational communication using a computer system, comprising: receiving a designation of an analysis target event related to the organizational communication of the analysis target; and relating to the designated event. And a step of displaying the log analysis result of the analysis section in association with the event information, and displaying the log analysis result of the analysis section. 15. A device for supporting analysis of an organizational communication log by using a computer system, comprising the step of accepting a designation of a utterance of interest from a user from the organizational communication log to be analyzed. And a step of displaying the log analysis result of the analysis section in association with the information of the statement. 16. A step of accepting a designation of an analysis target event related to an analysis target organizational communication in a computer system, a step of analyzing the organizational communication log for an analysis section related to the specified event, A program for executing the step of displaying the log analysis result for the analysis section in association with the event information. 17. A step of receiving, in a computer system, a designation of a utterance of interest from a user in an organizational communication log to be analyzed, and analyzing the organizational communication log for an analysis section related to the designated utterance. And a step of displaying the log analysis result of the analysis section in association with the information of the utterance, and a program for executing.