JP6090924B2 - User classification device - Google Patents

Description

  The present invention relates to a user classification device, and is particularly suitable for application to a system that promotes health competition among users based on sports performance, and can automatically perform appropriate user grouping from the viewpoint of user motivation and the like. The present invention relates to a user classification device.

  The following conventional techniques exist in a game involving a plurality of users.

  In Patent Document 1, matching is performed using a golfer type degree point calculated from a golfer type evaluation from another person. Further, in Non-Patent Document 1, the user's top ranking is introduced as a system in order for the user to continuously perform sports.

Japanese Patent Laid-Open No. 2008-021104

http://au-sports.auone.jp/

  Non-Patent Document 1 aims to increase the user's health level by directing the user's consciousness toward health and exercise. Specifically, the non-patent document 1 encourages competition among users such as running mileage ranking and abdominal muscle frequency ranking. The content of such health competition is prepared.

  However, the following problems may arise in health competition in which ranking is performed by the total participation method without making a distinction between users. That is, only a part of the entire user is listed in the rank (particularly the higher rank), and a situation may occur in which improvement of motivation is not promoted for most users.

  The following factors have hindered the improvement of such motivation. In other words, when competing for health, the conditions of each user are generally different.

  FIG. 1 schematically shows the motivation inhibiting factors. As examples of users with different conditions, there are users U1 to U3 who have a lot of time to use themselves and can exercise enough, and a user U4 who has little time to use himself and who can not exercise enough, Even if they compete within the same ranking, the user U4 who has less time is only in the lower rank, and improvement in motivation cannot be obtained.

  Here, as one of the major factors causing the difference in whether the usable time is large or small as described above, there is a difference in social conditions as follows.

  For example, the time that you can invest in your own health (free time) differs greatly between regular salaried workers and retired non-employed people, so that these people compete in the same ranking. Matching is not equal in the first place. In such unequal matching, there is a high possibility that motivation does not improve as intended. In addition, even for the same office worker, since the amount of work varies from department to department, it is not preferable to uniformly stand on the same ground (compete within the same ranking).

  Therefore, it is desired that users who are relatively close to each other are grouped together to compete for health / exercise to improve the motivation of many users. However, in the general matching method and grouping method as in Patent Document 1, such an improvement in motivation is not considered.

  An object of the present invention is to provide a user classification device that can classify users according to conditions in view of the above-described problems of the prior art.

  In order to achieve the above object, the present invention provides a user classification device, a schedule acquisition unit that receives schedule information from a plurality of users, an analysis of the schedule information, each activity type for each user, and the activity An activity pattern extraction unit that extracts the consumption time and the degree of restriction for the type as the activity pattern information of the user, and a user classification unit that classifies the plurality of users based on the extracted activity pattern information. This is the first feature.

  Further, according to the present invention, the user classification device further classifies the activity result acquisition unit that receives activity results for a predetermined item from the plurality of users, and classifies the user's result information for the received activity results by the classification unit. A second feature is that a grade creation unit is created for each group of users.

  According to the first feature, each user can be classified according to the characteristics of time use in the schedule.

  According to the second feature, since it is possible to create results of activity results of predetermined items in a group of users having similar time use characteristics, it is possible to increase the user's willingness to the activity. .

It is a figure which shows typically the inhibitory factor of motivation. It is a figure for demonstrating the outline | summary of this invention taking the case where it applies to a health competition system as an example. It is a functional block diagram of a user classification device concerning one embodiment. It is a figure which shows the example which automatically supplements the item of "movement", "sleeping time", and "free time" which are not described from a user's input schedule with the example of a procedure. It is a functional block diagram of the consumption time acquisition part which concerns on one Embodiment. It is a functional block diagram of a place acquisition part concerning one embodiment.

  FIG. 2 is a diagram for explaining the outline of the present invention, taking the case of application to a health competition system as an example. In the present invention, from the life log used by the user, in particular, the data in the schedule book, the time that the user can invest in his / her health is determined, and the group in which the time that can be used by himself / herself is as equal as possible is determined. Identify and do matchmaking.

  For example, depending on whether or not there is a business, and if there is a business, the user can be determined according to the style of business (whether there are many train movements, many walks, meetings, etc.) and time. Classify.

  That is, as schematically shown in FIG. 2, from the user's schedule, a group G1 that is assumed to be a “free person”, and a group G2 that is assumed to be “a sales office worker with n or more turns around” Group G3, which is assumed to be a “salaryman with more than m business trips”, and group G4, which is assumed to be a “office worker-centered salaryman”, can be divided into groups.

  Thus, in each group, meaningful competition is possible between users who have similar restrictions on how to use time. In addition, in the time utilized when performing the said grouping, the property of an idle time etc. are also taken into consideration. For example, the appropriate exercise differs depending on whether the person has a lot of time or a relatively long time. In addition to time, it is preferable to use attribute data that can be estimated from the life log, such as family composition and community characteristics, because further improvement in accuracy can be expected.

  FIG. 3 is a functional block diagram of a user classification device according to an embodiment. The user classification device 1 includes a schedule acquisition unit 2, an activity pattern extraction unit 3, a user classification unit 4, an activity result acquisition unit 5, a grade creation unit 6, and an information provision unit 7. The activity pattern extraction unit 3 further includes a consumption time acquisition unit 31 including a time word dictionary 301, an activity type acquisition unit 32 including a type word dictionary 302, a location acquisition unit 33 including a location word dictionary 303, and a restriction degree database ( A restraint degree calculation unit 34 including a restraint degree DB) 304.

  The schedule acquisition unit 2 acquires schedule information from each of a plurality of users. The schedule information only needs to be composed of text (computer-readable text) input by each user, but at least distinction of each schedule item is provided in advance.

The schedule item typically includes the following three types of information. In the present invention, the following activity pattern extraction unit 3 distinguishes and automatically acquires them.
-Words representing time… Example: AM, PM, 9: 00-10: 30
-Words representing place ... Example: T University, Company I-Words representing content ... Example: Meeting, meeting, going out, business trip

  For example, one item in a user's schedule is described as “9:00 to 10:30 Meeting @ T University”. As described above, the schedule item is described in units of events constituting the schedule.

  When the schedule acquisition unit 2 acquires the schedule information, it may be acquired individually from each user, but may be acquired from the schedule management system 10 in which the schedules of a plurality of users are recorded as illustrated. . The schedule management system 10 may be used by a business operator such as a company and used by a user belonging to a corresponding organization for business purposes. In particular, as a specific example, it is preferable to promote health competition by interlocking with a system used by a company when the company is enrolled in health insurance.

  Especially when linked with such a company system or the like, the schedule acquisition unit 2 may also acquire each user's profile and the like. For example, the department or job type of each user may be acquired. Further, the schedule management system 10 may be operated for each company such as a company, and the schedule and the like may be acquired by the schedule acquisition unit 2 from the plurality of systems. In this case, the company where each user works and the department to which the user belongs can be acquired as a profile.

  That is, the schedule acquisition target user may be a user in one or more organizations (each company or the like), and the user's belonging organization's distinction may be acquired in the profile. A dictionary and a database of each unit 301 to 304 to be described later may be prepared for each group, and may be prepared for each occupation classification of the user in each group. Thus, for example, the present invention can also be applied to each user in a series of companies that are participating in health insurance.

  In addition, when the schedule acquisition unit 2 acquires a schedule from a plurality of users, it is preferable to acquire a schedule within a predetermined period common to each user.

  The activity pattern extraction unit 3 analyzes the schedule information acquired by the schedule acquisition unit 2 using a predetermined dictionary or database (each unit 301 to 304 described later), and each user's schedule (that is, the schedule item for each user) Each activity type that constitutes the activity type) and the consumption time and the degree of restriction for each of the activity types are extracted and used as activity pattern information for each user. At the time of the extraction, the location of each schedule item may be acquired to obtain activity pattern information.

  Each item (consumption time, activity type, place, restraint degree (place may be omitted)) constituting the activity pattern information is determined by each unit 31 to 34 included in the activity pattern extraction unit 3 as follows. Asking.

  The consumption time acquisition unit 31 acquires the consumption time spent for each schedule item. For this reason, as a 1st process, after extracting the word showing time from each schedule item, as the 2nd process, the consumption time spent with respect to the said schedule item is acquired.

  In the first process, a word representing time is automatically extracted with reference to a word registered in advance in the time word dictionary 301 and a description rule thereof (eg, a value of hour and minute is delimited by “colon:”), and a start time And get the end time. In the second process, the consumption time is acquired as the difference between the start time and the end time.

  For example, from the schedule item “9: 00-10: 30 Meeting @ T University” in the above example, the start time “9:00” and the end time “10:30” are acquired, and the consumption time is set as the difference between them. Get as 90 minutes.

  The activity type acquisition unit 32 acquires the activity type of each schedule item. The activity type is obtained by giving a predetermined type classification to the contents in the schedule item. When acquiring the activity type, a word representing the content registered in advance in the type word dictionary 302 is extracted from the description of the schedule item, and the activity type of the word registered in the type word dictionary 302 is referred to. In this way, acquisition is made.

In the type word dictionary 302, for example, “each word representing contents” and “activity type” of each word are registered. Therefore, the activity type acquisition unit 32 referring to the type word dictionary 302 acquires the same activity type even for activities of the same type but described with different names.
Word “meeting”, “meeting”, “meeting”,… → activity type “meeting”
Word “Seminar”, “Training”, “Lecture”,… → Activity type “Seminar”
The words “paid leave”, “annual leave”,… → activity type “paid leave”

  For example, from the schedule item “9:00 to 10:30 Meeting @ T University” in the above example, the word “meeting” representing the content is extracted, and then “meeting” is acquired as the activity type.

  The location acquisition unit 33 refers to the location word dictionary 303 in which a word representing a location and its description rule (for example, a word representing a location appears after the symbol “@”) are registered in advance, and each schedule item Get the location. If a description rule such as “@” is not found, the location is acquired only by word search.

  In addition, the word of the place is basically the concept of the prefecture name or the station name, and then the municipality, the company name such as the company name, establishment name, number of floors in the building, meeting room It continues with names. In the place word dictionary 303, the word is not registered as it is, but is registered after linking with such a hierarchy. By referring to the hierarchical structure, since the location can be extracted from the company name if the company name is known, the process of reading the prefecture, municipality, etc. from the schedule item can be omitted. The location acquisition unit 33 acquires as detailed a location as possible using the hierarchical structure.

  In addition, when the description of the place is not included in the schedule item and the description of the place cannot be extracted, the place acquisition unit 33 is a predetermined place (for example, a work place during the day, Other than daytime, home, etc.) may be acquired as the location.

  Note that one purpose of the location acquisition unit 33 is that there are two schedule items that are adjacent in time for a certain user (however, the times do not have to be continuous), and the acquired locations are different between the schedule items. In this case, it is intended to automatically acquire “movement” as the activity type between the schedule items by informing the activity type acquisition unit 32 of the fact. At this time, in conjunction with the automatic acquisition of “movement”, the consumption time acquisition unit 31 acquires the time spent for the “movement”.

  The processing of the location acquisition unit 33 may be omitted, but if not omitted, the following effects can be obtained. That is, regarding “move”, the user usually omits writing in the schedule, but “move” can be automatically detected by the location acquisition unit 33. Whether there are a lot of “movements” greatly affects the user characteristics, so that more accurate grouping is possible by considering “movements” when grouping. Note that if the “moving” is described in the schedule item by the user, the consumption time acquisition unit 31 and the activity type acquisition unit 32 can automatically detect directly from the description.

Hereinafter, an example of automatic detection of “movement” by the location acquisition unit 33 will be described. It is assumed that a series of schedule items described for a certain user's day is as follows.
(Item 1) 10: 00-12: 00 Group meeting (Item 2) 13: 30-14: 00 Meeting between developers (Item 3) 16: 00-18: 00 Meeting at T University (Miyagi)

  For (Item 1) and (Item 2), the description of the place is not extracted, so “Saitama” is extracted as the place of the affiliated office that is the daytime residence registered as the user's profile. For (Item 3), the description is analyzed to extract “Miyagi”. Therefore, upon receiving the extraction result, the activity type acquisition unit 32 automatically acquires “move” as the activity type between (item 2) and (item 3) where the location has changed. In response to the extraction result, the consumption time acquisition unit 31 acquires the time period “14:00 to 16:00” in which the “movement” is performed, and acquires the consumption time “2 hours”.

  Alternatively, instead of calculating the time difference between items as described above, the consumption time acquisition unit 31 uses the places “Saitama” and “Miyagi” before and after the movement, and a database ( The consumption time required for the movement may be obtained by referring to the place word dictionary 303 as additional information).

  In addition, registration of the location of the affiliated office in the profile can be registered as a predetermined location of the user in each time zone more generally. For example, a predetermined location of the user is registered as the location of the affiliated business office during working hours and the home location outside of working hours, and can be used when there is no description of the location. Further, for example, a holiday may be registered in the same manner for each time zone and day of the week such as home all day. In addition, below, it demonstrates by the workplace and a home as an example of the predetermined location for every time slot | zone.

  As described above, in conjunction with the location acquisition unit 33, the following two automatic acquisition processes can be performed in the same manner as the consumption time acquisition unit 31 and the activity type acquisition unit 32 automatically acquire the contents related to “movement”.

  As a first process, referring to the normal working hours registered as the user's profile and the information on the workplace and home location, the schedule contents of the time periods where there is no description are “moved” (“ Move ").

  After the first process, the sleep time is further acquired as a user profile or general information (general information is artificially set by the administrator of the user classification device 1, etc.), and unprocessed as the second process. “Sleep time” and “Free time” may be acquired together with the consumption time as the activity type for each time zone.

  FIG. 4 shows an example of the first process and the second process. In FIG. 4, when (1) is a user's input schedule, (2) is its profile, and travel time data as shown in (3) is given, the first process (4) As described above, the activity type “movement” is automatically detected and supplemented. Here, as described above, (item B1) is acquired from the difference in location in (item B) and (item C) of the user input items. (Item A0) and (Item C1) are acquired from the working time, the location of the office / home, and the travel time data.

  In other words, the location of (Item A) is the office location “Saitama” and the start time is the same as the start time of working hours, so the previous stage is moving from “Tokyo” to “Saitama” (commuting to work) ), (Item A0) is automatically detected. Similarly, the location of (Item C) is “Miyagi”, and there are only 90 minutes between the end time and working hours, which is longer than 120 minutes required to return from “Miyagi” to “Tokyo” at home. Since it is short, (item C1) is automatically detected.

  Alternatively, (item C1) is defined as “17:30 to 19:30” from the end of working hours, and “16: 0 to 17:30” is blank or a predetermined activity type such as “regular work” is arranged. May be. As for the above-mentioned automatic supplementary processing near the start and end of working hours, a predetermined amount according to the location of the home, the location in the most recent schedule item at that time, and the length of time that is blank A rule may be provided and processing may be automatically performed according to the rule.

  Furthermore, using the sleep time data shown in (5), (item E3) is acquired as shown in (6) by the second process, and a part other than the sleep time is determined outside the end work hours within the day. As the activity type “free time”, (item E1) and (item E2) are acquired. That is, automatic supplementation is possible with a predetermined rule according to the user's profile or the like even in a time zone other than working hours. (7) will be described later.

  In the example of FIG. 4, all the contents of the time zone not specified by the user input shown in (1) are automatically detected. By detecting all, it becomes possible to grasp the user's characteristics (activity pattern information) in more detail, but the supplement by the automatic detection is not applied at all with only the input of (1) being analyzed, or the supplement is partially The present invention can be implemented even when applied to the above.

  In the example of FIG. 4, a part that is not described in the schedule item is automatically detected based on a predetermined activity content such as a predetermined location for each time zone and a bedtime. Similarly, automatic detection processing based on typical information that can be acquired as a profile or the like is possible for each user.

  In addition, when the above-mentioned consumption time acquisition unit 31, activity type acquisition unit 32, and location acquisition unit 33 acquire each item from the text described in the schedule item, the acquisition is performed in a predetermined order, and the description already acquired Acquisition is performed while excluding items from the search target for subsequent acquisition.

  In an example in which the regularity of the description content is taken into consideration, it may be acquired in the order of time, place, and activity type (corresponding content). When the order is applied to the schedule item “9: 00-10: 30 Meeting @ T University” in the above example, the time “9: 00-10: 30” is first extracted, and the description part of the time is excluded. The place “T University” is extracted from “Meeting @T University”, the content “Meeting” is immediately discriminated from the remaining “Meeting”, and the activity type “Meeting” can be obtained.

  Alternatively, if the schedule is input in advance in a predetermined table format and the items of the time, place, and activity type are preliminarily classified in the description of the schedule item, the contents of each item according to the item classification May be obtained.

The restriction degree calculation unit 34 obtains a restriction degree for the activity type acquired by the activity type acquisition unit 32 for each schedule item with reference to the restriction degree DB 304. The degree of restraint is obtained by artificially determining and quantifying in advance the degree to which each activity type restrains the user's time (separately by an administrator of the user classification device 1 or the like). The degree of restriction is recorded in the restriction degree DB 304 as well as the possible range of the value as in the following example.
Activity type: Degree of restraint (minimum value 0, maximum value 100)
Meeting: 80
Seminar: 40
Paid leave: 0

  In addition, when the user's profile is acquired by the schedule acquisition unit 2, the degree of restriction after defining the value for each profile so that the value of the degree of restriction differs depending on the profile, even for the same activity type The DB 304 may be provided, and the restriction degree calculation unit 34 may obtain a restriction degree corresponding to the profile with reference to the user profile.

  For example, if the job type is given in the profile, the activity type “negotiation” has a significant meaning because the job type “sales” is the position of the selling side and has an important meaning. In addition, in the job type “research”, since it is a position on the selling side and is not so serious, the value of the degree of restriction may be set to “20”.

  In one embodiment, the restriction degree calculation unit 34 refers to the restriction degree DB 304 as described above, and obtains the restriction degree as a uniform value corresponding to the activity type (and user profile). In another embodiment, the degree of restriction may be corrected to a value according to the actual situation by further adding details of the activity value in the activity type.

  In order to make the correction, additional information is added to the restriction degree DB 304 as one or more predetermined items to be reflected to reflect the actual situation according to the activity type, and the restriction degree to be corrected according to the content of the item. Define the value. The predetermined item can be acquired by referring to the constraint degree DB 304 as an additional process when the activity type acquisition unit 32 acquires the activity type and / or when the location acquisition unit 33 acquires the location. The restraint degree calculation unit 34 can calculate the modified restraint degree corresponding to the acquired activity type and the predetermined item with reference to the restraint degree DB 304.

  For example, in the case of “meeting” as the activity type, the degree of restriction when it is not corrected can be corrected as follows. When a predetermined “company name” exists as a predetermined item that reflects the actual situation of “meeting” (importance for participating users), whether it is related to other companies rather than a closed meeting within the company Alternatively, the degree of restraint may be “90”. The “company name” may be extracted from the description of the activity content in the activity type acquisition unit 32, or may be extracted from the description of the place in the place acquisition unit 33.

  For example, the activity type “meeting” and the company name “Company B” are extracted from the description of the activity content “Company B and Meeting”, and the description of the activity content and location “Meeting @ Company B” The company name “Company B” is extracted, and in any case, the degree of restraint is given as “90” instead of “80”.

  In addition, for example, for the activity type “meeting”, the activity type acquisition unit 32 further extracts the number of participants, and the constraint level calculation unit 34 acquires a value obtained by correcting the constraint level to a value according to the scale. It may be. In this case, a restriction degree correction method corresponding to the number of people range is defined in the restriction degree DB 304, for example, the degree of restriction is increased as the number of participants increases.

  The number of participants can be extracted by the activity type acquisition unit 32 as described above when described in the schedule item. If it is not described, the number of users who have the same profile and the same schedule as that of the user (for example, the same department and the same), in the same manner as the search process in the other person information acquisition unit 311 described later in FIG. (Corresponding to the number of people describing the meeting schedule).

  The user classifying unit 4 classifies the users into groups based on the activity pattern information (consumption time, activity type, location, restraint degree) of each user extracted by the activity pattern extracting unit 3 as described above. Each of the groups is composed of users having similar activity patterns, and therefore is an appropriate group in the case where a ranking of health competition is provided.

In one example, users are classified based on activity pattern information as follows. For each user i (i = 1, 2, ..., N), the total consumption time spent for the activity type j (j = 1, 2, ..., n (i)) is h. (j), and the degree of restriction for the activity type j is g (j). The time average _mi and the variance V _i of the degree of restriction of the user i are obtained by the following (Expression 1) and (Expression 2), respectively. Note that (7) in FIG. 4 is an example of the activity type, consumption time (total value in each activity type), and degree of restriction obtained from (6).

The obtained average and variance (m _i , V _i ) are used as values that aggregate the activity patterns of the user i, thereby grouping the users. For example, the average and variance (m _i , V _i ) of each user i are given as points on a two-dimensional plane, and clustering is performed for a plurality of points corresponding to the plurality of users by a well-known method. It can be performed. In addition, when giving as points, clustering may be performed on (am _i , bV _i ) by appropriately performing scale conversion using the constants a and b.

  The activity record acquisition unit 5 receives the activity record for a predetermined item from each user who acquired the schedule by the schedule acquisition unit 2. The grade creation unit 6 creates the grade of the user in the received activity result for each group classified by the user classification unit 4, and notifies the corresponding user.

  Here, as specific examples of activity results for predetermined items and creation of results after being classified into groups, each method in the health competition system disclosed in Non-Patent Document 1, etc., for example, sport activity results and their rankings, etc. , You may use. Alternatively, the ranking may be performed not only for health promotion activities such as sports, but also for the learning results (learning time and achievement level of each subject, etc.) in the in-house online training system. The present invention is particularly significant in applying ranking by the well-known method after grouping as outlined in FIG.

  In addition, the activity result acquisition unit 5 further receives user attributes related to the activity results (for example, user vital data in the case of sports) from each user, and in the result creation unit 6, the user classification unit 4 A grade may be created for each user who has the same received user attribute in the classified group. A reference for grouping based on user attributes is prepared in advance. Thus, for example, in the case of sports, rankings can be created between users who share the same time usage and basic physical strength.

  The information providing unit 7 provides each user with predetermined information corresponding to the user characteristics. For this reason, provision information according to user characteristics is prepared in advance. The user characteristics can be obtained as a group classified by the user classifying unit 4 and / or an activity pattern extracted by the activity pattern extracting unit 3. Alternatively, the amount used when classifying by the user classifying unit 4 such as the average and variance according to (Expression 1) and (Expression 2) may be used as the user characteristic. Further, user attributes related to the activity results may be included in the user characteristics.

  The provided information may be, for example, a recommended sport type. For example, it is assumed that a user who is assumed to have a large average degree of restraint in (Equation 1) by a predetermined threshold judgment has a long travel time, and it is not efficient to recommend running. It is possible to recommend simple simple squats. In this case, if a database in which each range of the restriction degree time average value and a sport type recommended to a user belonging to each range are described as a pair is prepared, provision information can be automatically prepared.

  If the present invention is applied to a health competition system, the information providing unit 7 recommends a sport that matches the user characteristics to the user, and the user obtains the activity result of the recommended sport as the activity result acquisition unit 5. It is possible to obtain the ranking of the recommended sports in the users with similar conditions from the score creation unit 6.

  Next, an additional embodiment in the consumption time acquisition unit 31 will be described. In the above description, the assumption that the start time and end time are clearly specified in the schedule item input by the user is used. In this embodiment, when only the start time is written and the end time is not specified, the end time is automatically detected.

  In one embodiment, the start time of the next schedule item (the next schedule item in the time sequence) input by the user may be regarded as an unspecified end time. For example, if (item b) “12:00 to 13:00 lunch break” is described after (item a) “9:00 to a meeting”, only the start time “9:00” is specified. The end time may be considered to be “12:00”.

  At this time, if the location acquired by the location acquisition unit 33 is different between the schedule item (item a) whose end time is not clearly specified and the next schedule item (item b), Thus, “movement” and its consumption time are automatically detected, and the consumption time is obtained by subtracting the time required for the “movement”.

  In another embodiment, as a first policy, it is assumed that a colleague of the user has participated in the same schedule item, and has automatically entered such a schedule item. To detect. As a second policy, assuming that the user has input the same schedule item accurately including the end time in the past or the future, such input is automatically detected. The end time is estimated by comprehensively considering the first and second policies.

  FIG. 5 is a functional block diagram of the consumption time acquisition unit 31 according to this embodiment. In addition to the time word dictionary 301 described above, the consumption time acquisition unit 31 further includes a certainty dictionary 3011 and includes the other person information acquisition unit 311 responsible for the processing of the first policy and the fluctuation dictionary 3012 and the processing of the second policy A self-information acquiring unit 312 that bears the responsibility, and an estimation unit 313 that gives a final result by comprehensively considering the first and second policies.

  The other person information acquisition unit 311 is assumed to be the same schedule item as that of the other person other than the user who describes the schedule item in which the end time is not specified, together with its certainty. Ask. At this time, the other person's schedule is searched as follows. The other person's schedule exists in the schedule acquired by the schedule acquisition unit 2.

  First, the other person who is the search target is in a range where the user and the predetermined profile are common. For example, a company or other company to which the user belongs or another person who has a common business division in the company is selected as a schedule search target.

  Secondly, the search key is a term representing at least the specified start date and time and the contents of the schedule item, and if a location is acquired, the location is also preferably used as the search key. Here, as the term representing the content, “each word representing the content” before being converted into the “activity type” by the activity type acquisition unit 32 is used. Here, since the description of the user giving the search key itself has ambiguity when specifying the event, the certainty degree is defined in advance for each word, and the certainty dictionary 3011 Register in advance.

The following are examples of the contents described in the certainty dictionary 3011.
(About words that describe content)
Meeting ... 0.4 certainty, meeting ... 0.5 certainty ...
(About words that represent places)
T University: Certainty factor 0.8, Miyagi ... Certainty factor 0.4, Tohoku ... Certainty factor 0.2, ...

For example, Mr. A's schedule item is described as “9: 00-Meeting @ T University”, and the other person information acquisition unit 311 performs a search using the date and time, contents “Meeting” and location “T University” as a key. When the schedule item of Mr. B who has the same time starts on the same day is searched as “9:00 to 10:00 meeting @T university”, according to the example of the certainty level, the process ends at “10:00”. The certainty degree of the search result can be calculated as follows. Here, the degree of uncertainty of “meeting” is “1−0.4 = 0.6”, and the degree of uncertainty of “T University” is “1−0.8 = 0.2”.
1- (0.6 * 0.2) = 0.88

In addition, by calculating the certainty degree by the method as described above, a lower certainty degree is calculated for a search result including only one of “meeting” and “T University” than when both are included. You can make it. For example, the degree of certainty described as “9:00 to 10:00 meeting” is “0.4” (consistent with the degree of certainty of “meeting” alone) as follows, and from the above “0.88” Is also a low value. When a search result including both is not obtained, the case where only one of them is included may be further searched, and when determining which one to use, the higher certainty side is adopted. May be.
1-0.6 = 0.4

  In addition, when the result of several different end time hits, those averages are used as end time. The degree of certainty may be increased according to the number of search hits and / or the closeness of the profile of the user who made the search hit (eg, the position is close to the organization). It may be used for weighting.

  The self-information acquisition unit 312 searches the own past and future schedules for content (and location) that is common and clearly specified the start time and end time, and the consumption time (start time and end time). Difference) is added to the start time in the schedule item to be estimated to obtain the one corresponding to the end time not described. When a plurality of searches are hit, an average of the obtained end times is used as a result. Note that the own schedule is acquired by the schedule acquisition unit 2 in the same manner as the schedule of others.

  As in the case of the other person information acquisition unit 311, the term representing the content when searching by the self-information acquisition unit 312 is converted to `` activity type '' by the activity type acquisition unit 32 `` “Each word representing content” is used. At this time, although the content is the same (and location), the expression may be different. For example, there are fluctuations in terms of places such as “T University” and “T University”. In order to absorb such fluctuations and search for those having the same content but different expressions, a fluctuation dictionary 3012 is constructed in advance, and the self-information acquisition unit 312 refers to the fluctuation dictionary 3012 when searching. May be.

The estimation unit 313 uses the end time T1 and the certainty degree p1 acquired by the other person information acquisition unit 311 and the end time T2 acquired by the self information acquisition unit 312 according to the certainty degree p1 as follows. The end time T is estimated proportionally to.
T = p1 * T1 + (1−p1) * T2

  In addition, when only one of the other person information acquisition unit 311 and the self information acquisition unit 312 can acquire the end time, the estimation unit 313 uses the end time acquired by the other one as an estimation result. Similarly, only one of the other person information acquisition unit 311 and the self information acquisition unit 312 may be applied in advance to acquire the end time.

  In the search in both the other person information acquisition unit 311 and the self information acquisition unit 312, if the acquired end time exceeds the start time of the next schedule item in the searched schedule item, the search result Exclude from

  As described above, when the location cannot be acquired in the schedule item by the same method as the automatic estimation of the end time by the consumption time acquisition unit 31 described in FIG. 5, the location acquisition unit 33 can also automatically estimate the location. As described above, in one embodiment, when the location cannot be acquired, the location acquisition unit 33 may automatically determine a predetermined location according to the time zone, such as a workplace described in the user's profile. .

  FIG. 6 is a functional block diagram of the location acquisition unit 33 for automatic estimation, and has functions corresponding to the units of the consumption time acquisition unit 31 of FIG. 5 so that the same name and the corresponding reference number are assigned. .

  The other person information acquisition unit 331 performs a search using a schedule of a user who has the same profile as the user who is the location estimation target as a search target, and the content and date / time as a search key. The certainty level is retained in the certainty dictionary 3031 for the content word, and the search result is obtained together with the certainty level. Even when a search is made from a plurality of users, it may be the same as described with reference to FIG.

  The self-information acquiring unit 332 searches for the past and future schedules of the user himself / herself using the content and the duration of the schedule item as search keys. At the time of the search, the fluctuation dictionary 3032 is referred to so as to absorb the same contents but different expressions. The search key may be only the content. In the estimation unit 333, the estimated location is obtained by proportionally proportioning the location acquired by the other person information acquisition unit 331 and the location acquired by the self-information acquisition unit 332. Alternatively, only one of the two results may be used.

  According to the user classification device 1 as described above, typically, as described in the activity results acquisition unit 5, the grade creation unit 6 and the information provision unit 7, using the classification result in the user classification unit 4, the health Competition-related services can be provided. More specifically, the health competition may be provided in the following manner.

(1) Health competition game in the same company For example, the amount of walking is completely different between people in office work and those outside. It is clear that when such people compete with each other for health, people who are outside the company are more advantageous from their lifestyles. For this reason, people who can use a similar time are searched and labeled from in-house data such as the above schedule. Competition is promoted by competing within the labeled groups. This label may be, for example, “business with travel distance of km or more”, “office worker with many meetings”, and so on.

(2) Health competition game between similar departments and companies. For example, in the sales of company A and company B, those departments are similar and work styles are very similar. To match. You may show an indicator such as “If you reduce 1 ton of fat overall, you will win”.

  1 ... User classification device, 2 ... Schedule acquisition unit, 3 ... Activity pattern extraction unit, 4 ... User classification unit, 5 ... Activity result acquisition unit, 6 ... Grade creation unit, 7 ... Information provision unit

Claims (9)

A schedule acquisition unit that receives schedule information from a plurality of users;
Activity pattern extraction that analyzes the schedule information and extracts each of the activity types for each user, the consumption time for the activity type, and the degree that the activity type binds the user's time as the activity pattern information of the user And
A user classifying unit comprising: a user classifying unit that classifies the plurality of users based on the extracted activity pattern information. The schedule acquisition unit receives schedule information for each schedule item in each user,
The activity pattern extraction unit further includes:
A consumption time acquisition unit for acquiring the consumption time spent for each schedule item;
An activity type acquisition unit that acquires the content word of each schedule item and acquires an activity type corresponding to the content word by referring to a predetermined dictionary;
The user classification device according to claim 1, further comprising: a constraint degree calculation unit that calculates the degree corresponding to the acquired activity type by referring to a predetermined database. The activity pattern extraction unit further includes a location acquisition unit that acquires a location in each schedule item by referring to a predetermined dictionary,
When the location acquired by the location acquisition unit is different in adjacent schedule items in the same user, the consumption time acquisition unit and the activity type acquisition unit, respectively, consumption time corresponding to between the schedule items, The user classification device according to claim 2, wherein an activity type of location movement is acquired. The schedule acquisition unit receives the schedule information together with the profile of each user, and the profile includes a predetermined location in each time zone of each user,
4. The user classification device according to claim 3, wherein the location acquisition unit acquires the predetermined location as a location when the location cannot be acquired from a schedule item by referring to the predetermined dictionary. The schedule acquisition unit receives the schedule information together with the profile of each user,
The consumption time acquisition unit, when only the start time is described in the schedule item, and the end time is not described,
Search for items with the same start time, content word, and location from the schedule items of users with a common profile, and refer to the specified certainty dictionary for the content word and location. The other person information acquisition unit for obtaining the certainty degree regarding the end time,
A self-information acquisition unit that searches the user's own schedule items for which the content word and location are the same, and the start time and end time are described,
An estimation unit for estimating an end time not described based on the end time and the certainty degree acquired by the other person information acquisition unit and the end time acquired by the self-information acquisition unit. The user classification device according to claim 3 or 4, characterized in that An activity results acquisition unit that receives activity results for predetermined items from the plurality of users;
The grade creation part which creates the said grade information of the said user with respect to the said received activity result for every group of the user classified by the said classification | category part is further provided, The provision in any one of Claim 1 thru | or 5 characterized by the above-mentioned. User classification device.   The user classification apparatus according to claim 6, wherein the predetermined item relates to a user health promotion activity.   The user classification device according to claim 1, further comprising an information providing unit that provides a user with predetermined information corresponding to the extracted activity pattern information.   9. The system according to claim 1, wherein the plurality of users are configured as users in one or more organizations, and the schedule acquisition unit receives schedule information together with information on the distinction of an affiliated organization from each user. The user classification device according to the above.