JP5600370B1 - User behavior analysis apparatus and behavior analysis program in community - Google Patents

Abstract

  Participation situation analysis unit 13 that analyzes the user's involvement situation in the community site for each of a plurality of divided periods obtained by dividing the analysis target period using index values relating to a plurality of types of behaviors of the user in the community site. And a change pattern specifying unit 14 that specifies which change pattern corresponds to each of a plurality of users by analyzing how the participation status has changed over a plurality of divided periods, and during the analysis target period By cross-tabulating the number of participants in the campaign in which the user is participating, for each of multiple change patterns, using the participation status information that indicates which campaign the multiple users have participated in. And a relationship analysis unit 15 for analyzing the relationship between the change pattern of the campaign and the plurality of campaigns. A campaign that was carried out in order, to be able to understand the causal relationship between changes in the behavior of the user in a community site.

Description

  The present invention relates to a user behavior analysis apparatus and a behavior analysis program in a community.

  Conventionally, various campaigns are generally performed as a measure for promoting the sale of products. The content of the campaign is diverse and is based on the advertising strategy of the company that sells the product. For example, present sweepstakes for purchasers of goods, provision of free samples using stores and streets, recruitment of monitors for new products, etc. are widely implemented.

  When conducting a campaign, of course, it costs a certain amount. Companies run campaigns with the expectation of an effect (such as sales increase) that exceeds the cost they spend, but it is not certain what campaigns are producing results. Therefore, it is desired to be able to confirm the effectiveness of the campaign as objectively as possible. This is because the subsequent advertising strategy can be optimized according to the confirmation result.

  Conventionally, various techniques for measuring the effect of a campaign executed for a specific product have been proposed (see, for example, Patent Documents 1 to 3). In the sales promotion campaign effect measurement system described in Patent Document 1, a plurality of stores and a base server are connected via a communication line, and sales results of products transmitted from each store are stored for each date, store, and product. . Then, the stored sales data is compared with the sales promotion campaign information, and the change in the number of sales is monitored in real time as the effect of the sales promotion campaign.

  In addition, the sales information analysis device described in Patent Document 2 holds a correspondence relationship between each area and an area cluster that is a group of areas having similar attributes regarding residents. The sales information analyzer acquires the date and time when the sales promotion measure was implemented, and also acquires the product sales volume in each area before and after the implementation of the sales promotion measure. The sales information analysis apparatus totals the product sales volume in each area for each area cluster, and calculates the increase rate of the product sales volume by the sales promotion measure for each area cluster.

  Moreover, in the campaign effect calculation system described in Patent Document 3, mROI analysis processing is performed on campaign measure information. The mROI analysis process identifies the customer group that purchased the target product during the campaign period and the other customer groups, and calculates the overall profit boosting effect from the difference in sales increase rate between these customer groups.

  By the way, in recent years, many community sites (SNS sites) have been provided on the Internet, and users with common fields of interest, values, and purposes gathered to actively communicate information such as information sharing and free remarks. It has been broken. In this community site, remarks about products that the user is interested in are made, and it is known that word-of-mouth by the remarks greatly affects the sales of the products.

  Recently, advertising that uses word-of-mouth is also being carried out in anticipation of being more cost-effective than conventional advertising that promotes mass media to unspecified number of people. For example, product promotion messages are published on the net as content with impact and topicality, attracting users' interests and interests, and communicating the goodness of products to many people through word-of-mouth effects.

  In addition, a technique for analyzing a causal relationship between information posted to the SNS and purchase of a product has been proposed (see, for example, Patent Document 4). In the campaign management apparatus described in Patent Document 4, a description suggesting the purchase of a product is searched from information posted by the customer to the SNS, and it is estimated whether the customer has purchased the product based on the search result. To do. In addition, this Patent Document 4 also describes that a customer type relating to a royalty (loyalty, patronage) for a campaign store is determined.

JP 2005-128553 A JP 2009-169698 A JP 2013-005799 A JP 2013-246747 A

  According to the techniques described in Patent Documents 1 to 3, it is possible to analyze how much the sales amount of the target product has increased due to the implementation of the campaign based on the actual value of the product purchase. Further, according to the technique described in Patent Document 4, it is possible to estimate whether or not a target product for the campaign has been purchased from the content of the user's comments in the community.

  However, the techniques described in Patent Documents 1 to 4 have a problem in that it is impossible to specifically analyze how the campaign implementation affects the user's behavior in the community. . As described above, a user's behavior within a community regarding a certain product has a considerable influence on the sales of the product. Therefore, it is desirable to analyze and understand the causal relationship between the contents of the campaign and the user's behavior in the community.

  The present invention has been made to solve such problems, and how the campaign conducted for the promotion of the product affects the user's behavior in the community. The purpose is to be able to grasp.

  In order to solve the above-described problems, in the present invention, index values related to a plurality of types of behavior of users on a community site, which are index values related to a plurality of users in a desired analysis target period, and users in the analysis target period The following analysis is performed using the participation status information indicating the participation status of the campaign. That is, for each of a plurality of divided periods obtained by dividing the analysis target period, the user's involvement status in the community site is analyzed for each of the plurality of users, and how the involvement status changes over the plurality of divided periods is analyzed. Thus, the change pattern is specified for each of a plurality of users. For each of the plurality of change patterns, the number of participants in the campaign in which the user is participating is cross-counted to analyze the relationship between the plurality of change patterns and the plurality of campaigns.

  According to the present invention configured as described above, a change pattern of a user's behavior on a community site in a desired analysis target period is analyzed, while a user showing the same change pattern is executed during the analysis target period. It is analyzed how many people have participated in which of the campaigns. Thereby, based on the analysis result, it is possible to infer a causal relationship between the campaign executed for the sales promotion of the product and the change in the behavior of the user in the community site. As a result, it is possible to grasp how the actual campaign affects the user's behavior in the community site.

It is a block diagram which shows the function structural example of the action analysis apparatus by 1st Embodiment. It is a figure which shows an example of the index value which the index value acquisition part of this embodiment acquires. It is a figure which shows an example of the participation status information which the participation status information acquisition part of this embodiment acquires. It is a block diagram which shows the example of a specific function structure for analyzing a participation level by the participation condition analysis part which concerns on 1st Embodiment. It is a figure for demonstrating the processing content of the total value calculation part which concerns on 1st Embodiment. It is a figure which shows the monthly involvement level of each user analyzed by the involvement situation analysis part which concerns on 1st Embodiment. It is a figure for demonstrating the processing content of the change pattern specific | specification part by 1st Embodiment. It is a figure for demonstrating the processing content of the change pattern specific | specification part by 1st Embodiment. It is a figure for demonstrating the processing content of the relationship analysis part by 1st Embodiment. It is a block diagram which shows the function structural example of the action analysis apparatus by 2nd Embodiment. It is a block diagram which shows the example of a specific function structure for analyzing a participation type by the participation condition analysis part which concerns on 2nd Embodiment. It is a figure for demonstrating the processing content of the average value calculation performed by the teacher data generation part by 2nd Embodiment. It is a figure for demonstrating the processing content of the non-hierarchical cluster analysis performed by the teacher data generation part which concerns on 2nd Embodiment. It is a figure which shows the engagement type according to each user's month analyzed by the participation condition analysis part which concerns on 2nd Embodiment. It is a figure for demonstrating the processing content of the index value totalization part by 2nd Embodiment.

(First embodiment)
DESCRIPTION OF EXEMPLARY EMBODIMENTS Hereinafter, a first embodiment of the invention will be described with reference to the drawings. FIG. 1 is a block diagram illustrating a functional configuration example of a “user behavior analysis apparatus in a community” according to the first embodiment. As shown in FIG. 1, the behavior analysis apparatus 100 according to the first embodiment has, as its functional configuration, an index value acquisition unit 11, a participation situation information acquisition unit 12, an engagement situation analysis unit 13, a change pattern specification unit 14, and a relationship. An analysis unit 15 is provided.

  Each of the functional blocks 11 to 15 can be configured by any of hardware, DSP (Digital Signal Processor), and software. For example, when configured by software, each of the functional blocks 11 to 15 is actually configured by including a CPU, RAM, ROM, and the like of a computer, and stored in a recording medium such as a RAM, ROM, hard disk, or semiconductor memory. Is realized by operating.

  The index value acquisition unit 11 acquires index values related to a plurality of types of user actions on a community site on the Internet. This index value is a value indicating an index related to behaviors related to a plurality of users in a desired analysis target period. It is assumed that the community site here is a community site related to a specific product opened by a company that implements various campaigns for sales promotion of a product.

  FIG. 2 is a diagram illustrating an example of the index value acquired by the index value acquiring unit 11. In the first embodiment, as an example of the index, seven of the number of PV (page views), the number of sessions, the staying time, the number of comment posts, the length of comments, the number of applause, and the number of applause are used. Note that the seven indicators listed here are merely examples, and it is not essential to use all of them. Moreover, you may use indices other than these seven if it can represent a user's action.

  The analysis target period can be arbitrarily set by the operator who performs the analysis. In the example of FIG. 2, the analysis target period is set to 9 months. The index value acquisition unit 11 uses a plurality of seven index values from the log data related to the community site as parameters indicating what actions each of the plurality of users took during the analysis target period from January to September. Get for each user.

  Note that it is not necessary for individual users to specifically specify “who”, and it is sufficient if they can be distinguished from each other. For example, any one of a user ID, a user name, and a nickname used in the community site can be used as information for identifying each user. Although FIG. 2 shows five users, there are actually more than this.

  The participation status information acquisition unit 12 acquires participation status information indicating which campaign a plurality of users have participated among a plurality of campaigns performed for sales promotion of a product during the analysis target period. FIG. 3 is a diagram illustrating an example of participation status information acquired by the participation status information acquisition unit 12. In the example of FIG. 3, among the plurality of campaigns 1 to n performed during the analysis target period, the presence / absence of participation is distinguished by a flag of 1 for a user participating and 0 for a non-participating user. Note that FIG. 3 is shown for explanation, and it is not essential that the participation status information is in this data format.

  Here, whether or not the user has participated in each campaign can be grasped by, for example, an interview survey using a questionnaire. By conducting an electronic questionnaire provided on the community site, it is possible to easily collect the participation status of each user for each campaign. Further, based on action data taken by the user on the community site (for example, whether or not a reply is made to the presented theme), the presence or absence of participation in the campaign may be determined.

  The participation situation analysis unit 13 performs a predetermined calculation on the seven index values related to the user's behavior for each of a plurality of divided periods obtained by dividing the analysis target period, and based on the result of the calculation, the user's community site Analyzing the participation status in each user. In the first embodiment, one month is set as an example of the division period. That is, the user's involvement status is analyzed for each month from January to September.

  In the first embodiment, as an example of the involvement status, a “participation level” that represents the degree of involvement of the user with the community site is analyzed. Specifically, the participation status analysis unit 13 calculates a participation point by performing a predetermined weighted addition operation on seven index values for each of a plurality of divided periods obtained by dividing the analysis target period. Then, based on the calculated size of the participation point, which level corresponds to the participation level of the plurality of stages is analyzed for each of the plurality of users.

  FIG. 4 is a block diagram illustrating a specific functional configuration example for analyzing the participation level by the participation status analysis unit 13. As shown in FIG. 4, the participation situation analysis unit 13 includes a total value calculation unit 131, a weight calculation unit 132, an engagement point calculation unit 133, and a level determination unit 134 as specific functional configurations.

  As shown in FIG. 5, the total value calculation unit 131 calculates the total value SUMij (i = 1 to 9) of the index values of all users for each of the seven index values for each of a plurality of divided periods from January to September. , J = 1 to 7) are calculated respectively. The weight calculation unit 132 multiplies the reciprocal of the total value SUMij calculated for each index value by the total value calculation unit 131 by 0.1, and calculates the index of each month by the calculation “Wij = 0.1 / SUMij”. A weight Wij for each value is calculated.

  The participation point calculation unit 133 multiplies each of the index values for each user (see FIG. 2) for each month by the weight Wij calculated by the weight calculation unit 132, and totals the values obtained for each index. Thus, the monthly participation points of each user are calculated. For example, the seven index values for January of user 1 shown in FIG. 2 are respectively multiplied by the weights W1j (j = 1 to 7) of the index values for January calculated by the weight calculation unit 132, By summing the obtained seven values, the January participation points of the user 1 are calculated. Similarly, the participation points for January are calculated for other users. By performing this for each of January to September, it is possible to calculate the monthly participation points of each user.

  The level determination unit 134 determines the participation level of each user according to the size of the participation point calculated by the participation point calculation unit 133 for each of a plurality of divided periods from January to September. For example, three threshold values are set with respect to the size of the participation point, and a level comparison between the three threshold values and the participation point of each user determines which level the user belongs to among four levels of participation levels. . The level determination method shown here is merely an example, and the present invention is not limited to this method.

  For example, the values of the participation points calculated by the participation point calculation unit 133 are arranged in descending order, and after 0-point users are excluded, the percentage of the ranking is specified. Then, three thresholds are set with respect to the magnitude of this percentage, and each user corresponds to which of the four levels of participation level by comparing the three thresholds with the percentage specified from the participation point of each user. You may make it determine.

  FIG. 6 is a diagram showing the monthly participation level of each user analyzed by the participation situation analysis unit 13 by the above processing. In the example of FIG. 6, level 1 has the smallest participation point, and level 4 has the largest participation point. That is, the participation points are large in the order of level 1, level 2, level 3, and level 4.

  Returning to FIG. The change pattern specifying unit 14 analyzes how the participation level analyzed by the participation status analysis unit 13 for each of the plurality of divided periods (monthly from January to September) has changed over the plurality of divided periods. Which of a plurality of predetermined change patterns corresponds to each of a plurality of users. In the first embodiment, the change pattern specifying unit 14 includes, as a plurality of change patterns, a first pattern in which the participation level is increased, a second pattern in which the participation level is in a maintained state, and a first pattern in which the participation level is decreased. Which of the three patterns corresponds is specified for each of a plurality of users.

  7 and 8 are diagrams for explaining the processing contents of the change pattern specifying unit 14 according to the first embodiment. In the first embodiment, the analysis target period from January to September is divided into three periods. That is, January to March is defined as period 1, April to June is defined as period 2, and July to September is defined as period 3. Then, for each period, the one with the highest participation level (the one with the largest participation point) is set as the user's participation level in that period. For example, as shown in FIG. 7, when the participation level of the user 1 from January to March is level 1, level 2, and level 4, respectively, the level 4 having the largest participation point is selected as the period of the user 1. The level of involvement in 1.

  In this way, three participation levels are set for each of the periods 1 to 3 for one user. The change pattern specifying unit 14 determines whether the change pattern of the participation level corresponds to increase, maintenance, or decrease from period 1 to period 2. Similarly, it is determined from the period 2 to the period 3 whether the change pattern of the participation level corresponds to an increase, a maintenance, or a decrease. Then, according to the determined combination of the two change patterns, it is specified for each of a plurality of users whether the involvement level corresponds to an increase, maintenance, or decrease pattern throughout the entire analysis target period.

  FIG. 8 shows whether the change pattern of the participation level in the entire analysis target period falls, rises, maintains, or falls from the combination of the change pattern in period 1 to period 2 and the change pattern in period 2 to period 3. FIG. According to the example shown in FIG. 8, in the case of the user 1 shown in FIG. 7, the change pattern in the period 1 to the period 2 is lowered (↓), and the change pattern in the period 2 to the period 3 is also lowered (↓). The change pattern of the participation level in the entire analysis target period is the third pattern of decrease.

  In this example, the analysis target period of 9 months is divided every three months and three periods 1 to 3 are set. However, the present invention is not limited to this method. For example, if the participation status analysis unit 13 divides the analysis target period into three divided periods instead of nine and analyzes the participation level, the change pattern specifying unit 14 uses the obtained three participation levels as they are and the change pattern Can be specified.

  In the above embodiment, the analysis target period is divided into three periods, two change patterns are determined, and the change pattern of the entire analysis target period is specified from the combination of the two change patterns. It is not limited. For example, the analysis target period may be divided into two periods, and the change pattern of the entire analysis target period may be directly specified from the change in the participation level in the two periods.

  The relationship analysis unit 15 cross-tabulates the number of participants in the campaign in which the user identified as corresponding to any change pattern by the change pattern specifying unit 14 for each change pattern of ascending, maintaining, and descending. By analyzing the relationship between multiple change patterns and multiple campaigns.

  FIG. 9 is a diagram for explaining the processing contents of the relationship analysis unit 15 according to the first embodiment. FIG. 9A shows a table in which the participation status information indicating the participation status in each campaign shown in FIG. 3 and the change pattern specified for each user by the change pattern specifying unit 14 are combined. Yes.

  Based on the information shown in this table, the relationship analysis unit 15 totals how many users are participating in which campaign whose change pattern indicates “up”. In addition, the number of users who participate in which campaign whose change pattern indicates “maintain” is totaled. Further, the number of users who participate in which campaign whose change pattern indicates “down” is totaled. FIG. 9B shows the result.

  Moreover, as shown in FIG.9 (c), the relationship analysis part 15 may total | count how many users correspond to this according to a change pattern. Further, as shown in FIG. 9 (d), the relationship analysis unit 15 further uses the total results shown in FIGS. 9 (b) and (c) to determine how much from each change pattern group for each campaign. You may make it total whether the number of people participates.

  As described above in detail, according to the first embodiment, while a change pattern of a user's behavior on a community site in a desired analysis target period is analyzed, a user who shows the same change pattern is analyzed. It is analyzed how many campaigns participated in the campaigns executed during the period. Thereby, based on the analysis result, it is possible to infer a causal relationship between the campaign executed for the sales promotion of the product and the change in the behavior of the user in the community site. As a result, it is possible to grasp how the actual campaign affects the user's behavior in the community site.

  For example, the campaigns that change the level of involvement in the community site for the products targeted for the campaign were classified as “increased”, which campaigns did or did not participate, and those classified as “maintained” Users who have participated or did not participate and who are classified as “descent” can know which campaign has or has not participated.

  The behavior of users who have a change pattern classified as “rising” in the community site (especially comment posting and applause) has a significant impact on other users and contributes to an increase in product sales. It can be hypothesized that there is a possibility. Based on this hypothesis, by confirming which campaign the user whose change pattern is classified as “rising” is participating, or what type of campaign is participating in the promotion effect, It is possible to estimate a campaign that seems to be high.

  In addition, it is possible to make a hypothesis that a user whose change pattern is classified as “rising” has a high possibility of purchasing a product. Based on this hypothesis, it is possible to estimate a campaign that seems to have a high sales promotion effect by confirming a campaign in which the change pattern is classified as “rising” and a large number of users participate.

  Or, if you look at a campaign in which a group of users whose change pattern is classified as “down” is participating, the trend is that most of them participate in Campaign 1 but not in Campaign 2, Because the experience expected in Campaign 1 was not obtained, motivation did not take hold until the time of Campaign 2, or the desire evoked in Campaign 1 could not be continuously responded in Campaign 2. Can also be inferred.

  In the first embodiment, the example in which three change patterns of ascending, maintaining, and descending are specified based on the change in the participation level has been described, but the present invention is not limited to this. For example, the change pattern specifying unit 14 determines the first pattern if the participation point has risen beyond a predetermined value over two periods, and the second pattern if the participation point is in a maintenance state within a predetermined range. If the point has fallen beyond a predetermined value, it may be determined that the point corresponds to the third pattern. In this case, the participation status analysis unit 13 does not need to analyze the participation level by calculating up to the participation points.

(Second Embodiment)
Next, a second embodiment of the present invention will be described with reference to the drawings. FIG. 10 is a block diagram illustrating a functional configuration example of the “user behavior analysis apparatus in a community” according to the second embodiment. Note that in FIG. 10, those given the same reference numerals as those shown in FIG. 1 have the same functions, and thus redundant description is omitted here.

  As shown in FIG. 10, the behavior analysis apparatus 100 ′ according to the second embodiment further includes an index value totaling unit 16 and an ordering unit 17. Further, the behavior analysis apparatus 100 ′ according to the second embodiment includes an engagement situation analysis unit 13 ′ and a change pattern identification unit 14 ′ instead of the involvement situation analysis unit 13 and the change pattern identification unit 14 illustrated in FIG. ing. In the second embodiment, the involvement status analysis unit 13 ′ analyzes “involvement type” representing how a user is involved in a community site as an example of the involvement status.

  Specifically, the participation situation analysis unit 13 ′ performs a predetermined clustering operation on seven index values related to the user's behavior for each of a plurality of divided periods obtained by dividing the analysis target period. Which type of participation type is applicable is analyzed for each of a plurality of users. In the present embodiment, as an example, each user is described as being classified into four types of participation types. The four types of involvement are, for example, type A that is not so involved in the community site, type B that is relatively involved, type C that is mainly involved in comment posting, and type D that is uniformly involved.

  FIG. 11 is a block diagram illustrating a specific functional configuration example for analyzing the participation type by the participation state analysis unit 13 ′. As shown in FIG. 11, the participation situation analysis unit 13 ′ includes a teacher data generation unit 135 and an engagement type determination unit 136 as specific functional configurations.

  The teacher data generation unit 135 aggregates the seven index values in the entire analysis target period for each user, and uses the aggregated values on the coordinate plane specified by the combination of two indices out of the seven indices. Four cluster center points corresponding to the types of participation types are identified. Then, a teacher who assigns a plurality of users to any of the four clusters based on the relationship between the four cluster center points and the position on the coordinate plane specified by the combination of the aggregate values obtained for the two indices. Generate data. Hereinafter, the processing of the teacher data generation unit 135 will be described in detail.

  First, the teacher data generation unit 135 aggregates seven index values for the entire analysis target period for each user. For example, as shown in FIG. 12, the teacher data generation unit 135 calculates an average value for each index value in a monthly unit for each analysis period for each of a plurality of users. That is, the teacher data generation unit 135 calculates the average value of each index value for each of a plurality of users by totaling each index value for 9 months and dividing the total number by 9. . Although the average value is calculated here, it may be a total value.

  Next, the teacher data generation unit 135 performs non-hierarchical cluster analysis by applying a known k-means method using the average value of each index value calculated as described above. The k-means method is an analysis method for performing clustering by giving a prototype (cluster center point) representative to each of a predetermined number of clusters and assigning each individual to the closest prototype.

  FIG. 13 is a diagram for explaining the processing contents of this non-hierarchical cluster analysis. In the non-hierarchical cluster analysis, first, the center points of the four clusters corresponding to the four types of involvement are specified on the coordinate plane specified by the combination of two indexes among the seven indexes related to the user's behavior.

  For example, as shown in FIG. 13, a coordinate plane is set in which two indices, the number of PVs and the number of sessions, are taken on the vertical axis and the horizontal axis. Then, reference points (many small points shown in FIG. 13) corresponding to each user are plotted at coordinate positions determined by the average value of the PV numbers calculated for each user and the average value of the number of sessions. In this state, first, initial values of the center points P1 to P4 are given to the four clusters. For example, a plurality of reference points are randomly selected to form a cluster, and the center points of the plurality of reference points included in the cluster are set as initial values.

  Next, the distances between all the reference points and the four cluster center points P1 to P4 are calculated, and all the reference points are reclassified into clusters having the smallest distances from the cluster center points P1 to P4. Then, the center positions of the reclassified reference points are set as new center points P1 'to P4' of the cluster. The processing from the distance calculation to the center point setting is repeated until the cluster center points P1 to P4 no longer change (until P1 to P4 = P1 'to P4').

  When the four cluster center points P1 to P4 are determined in this way, the four cluster center points P1 to P4 are plotted at positions on the coordinate plane specified by the combination of the two average values of the PV number and the session number. Based on the positional relationship with the reference points, teacher data in which a plurality of users corresponding to each reference point are assigned to any of a plurality of clusters is completed. The teacher data generation unit 135 calculates such teacher data for each of the combinations of the other two indices out of the seven indices.

  The involvement type determination unit 136 includes the teacher data generated by the teacher data generation unit 135, and seven index values (see FIG. 2) for each of a plurality of divided periods (every month) acquired by the index value acquisition unit 11. From these relationships, a known discriminant analysis method is applied to determine which type a user belongs to among the four types of involvement. The discriminant analysis method is a method of constructing a discriminant model using teacher data in which it is clear to which group an individual belongs, and discriminating to which group an individual with unknown affiliation belongs using the model.

  In other words, the participation type discriminating unit 136 first uses the teacher data generated for each combination of two indicators by the teacher data generation unit 135 to create a reference for type discrimination for each combination of two indicators. A discrimination coefficient for each participation type is obtained. And a user's involvement type is discriminate | determined by applying a discriminant coefficient with respect to seven index values according to the month acquired about each user by the index value acquisition part 11. FIG.

  For example, the participation type discriminating unit 136 uses a discriminant coefficient for discriminating the type of the combination of the PV number and the session number from the teacher data generated by the teacher data generating unit 135 for the combination of the two indicators of the PV number and the session number. Ask. Then, by applying this determination coefficient to the monthly PV number and session number of the user 1 shown in FIG. 2, the January involvement type regarding the combination of the PV number and the session number is determined for the user 1.

  The involvement type determination unit 136 also determines the participation type in the same manner for the combination of the other two indicators among the seven indicators of the user 1 in January. Thereby, based on the index value of January of the user 1 acquired by the index value acquisition unit 11, a plurality of involvement types are determined from a plurality of combinations of indices. Then, the involvement type determination unit 136 determines the participation type to which the most belongs as the January involvement type of the user 1. Similarly for other users, the January involvement type is determined. By performing this for each of January to September, it is possible to determine the monthly participation type of each user. FIG. 14 shows an example of the monthly participation type of each user determined by the participation type determination unit 136 by the above processing.

  The index value totaling unit 16 totals seven index values for each user analyzed by the teacher data generation unit 135 as belonging to each participation type for each of the four types of participation types. As described above, the teacher data indicates which user belongs to which participation type. The index value totaling unit 16 uses the teacher data to calculate the average value of the index values of the users belonging to the four types of participation types, as shown in FIG.

  The order assignment unit 17 performs the same weighted addition operation as the weight calculation unit 132 and the participation point calculation unit 133 illustrated in FIG. 4 for the total value (average value of each index value) calculated by the index value total unit 16. To calculate the participation points. And based on the magnitude | size of the participation point, an order is given to four types of participation types. As an example, an order is given to four types of involvement so that the higher the participation point, the higher the ranking. For example, in the order of participation type A → B → C → D, an order is given such that the participation type D is the highest and the participation type A is the lowest.

  The change pattern specifying unit 14 ′ determines how the participation type (see FIG. 14) analyzed for each of a plurality of divided periods (every month from January to September) by the participation status analyzing unit 13 ′ over a plurality of divided periods. And whether the change pattern corresponds to the change pattern of ascending, maintaining, or descending is specified for each of a plurality of users. In 2nd Embodiment, change pattern specific | specification part 14 'is 1st pattern, when the order of the rank provided to each participation type by the order assignment part 17 is rising, and the order of rank is in a maintenance state. Is determined to fall under the second pattern and the third pattern when the rank of the rank is lowered.

  Also in the second embodiment, the change pattern specifying unit 14 ′ divides the analysis target period from January to September into three periods 1 to 3, similarly to the method shown in FIGS. -The change pattern of the involvement type in the whole analysis target period is specified for each of a plurality of users according to the combination of the change pattern of the relationship type in the period 2 and the change pattern of the participation type in the period 2 to the period 3.

  In the second embodiment configured as described above, the change pattern of the user's behavior on the community site in the analysis target period is analyzed, while the same change pattern is displayed, as in the first embodiment. It is analyzed how many users participated in which campaign among a plurality of campaigns executed during the analysis target period. Thereby, based on the analysis result, it is possible to infer a causal relationship between the campaign executed for the sales promotion of the product and the change in the behavior of the user in the community site. As a result, it is possible to grasp how the actual campaign affects the user's behavior in the community site.

  In the second embodiment, an example has been described in which the involvement type is determined for all combinations of two indicators out of the seven indicators, and the most frequently determined involvement type is specified as the user involvement type. However, the present invention is not limited to this. For example, the participation type may be determined only for a combination of indices that are particularly related to the participation type to be determined.

  In the first and second embodiments, the change pattern of the participation situation (participation level or participation type) analyzed by the participation situation analysis units 13 and 13 ′ is specified, and the change pattern of the participation situation and the campaign are specified. Although the example in which the relationship analysis unit 15 analyzes the relationship with the participation status is described, the present invention is not limited to this. For example, the relationship between the participation status (participation level or participation type) itself analyzed by the participation status analysis units 13 and 13 ′ and the participation status in the campaign may be analyzed.

  In this case, the participation situation analysis units 13 and 13 'do not necessarily have to analyze the participation situation of the user's community site for each of a plurality of divided periods. For example, when it is desired to analyze the relationship between a specific campaign and the participation status of the user in the community site, the participation status analysis units 13 and 13 ′ are set to seven index values only for the period in which the campaign is executed. On the other hand, it is possible to perform the above-described calculation and analyze the user's involvement status based on the result of the calculation.

  In this case, the relationship analysis unit 15 performs the participation status analysis unit 13, 13 ′ for each of a plurality of participation statuses (four levels of participation levels or four types of participation types) analyzed by the participation status analysis unit 13, 13 ′. By cross-counting the number of users who participated in the campaign of the user identified as corresponding to any of the participation situations, the relationship between the plurality of participation situations and the specific campaign can be analyzed.

  It should be noted that the present invention is not limited to the analysis of the relationship between one specific campaign and the participation status of the user in the community site, but a plurality of campaigns executed during the same analysis target period as in the first and second embodiments, and the user community It is also possible to analyze the relationship with the engagement status at the site. From this analysis result, for example, which campaign the user classified into the highest participation level 4 participated or did not participate, and which campaign the user classified into the engagement type D participated or did not. I can understand that.

  In addition, each of the first and second embodiments described above is merely an example of a specific example for carrying out the present invention, and the technical scope of the present invention should not be interpreted in a limited manner. It will not be. That is, the present invention can be implemented in various forms without departing from the gist or the main features thereof.

DESCRIPTION OF SYMBOLS 11 Index value acquisition part 12 Participation condition information acquisition part 13, 13 'Participation condition analysis part 14, 14' Change pattern specific | specification part 15 Relation analysis part 16 Index value total part 17 Order assignment part 100 Behavior analysis apparatus 131 Total value calculation part 132 Weight calculation unit 133 Participation point calculation unit 134 Level determination unit 135 Teacher data generation unit 136 Participation type determination unit

Claims (9)

An index value acquisition unit that acquires an index value related to a plurality of types of behavior of a user in a community site on the Internet, and acquires the index value related to the behavior related to the plurality of users in a desired analysis target period;
A participation status information acquisition unit for acquiring participation status information indicating which campaign the plurality of users have participated in among the plurality of campaigns performed during the analysis target period;
For each of a plurality of divided periods obtained by dividing the analysis target period, a predetermined calculation is performed on the index values related to the plurality of types of behaviors, and based on the result of the calculation, the state of involvement of the user in the community site is determined. An engagement analysis unit for analyzing each of the plurality of users;
Analyzing how the participation status analyzed for each of the divided periods by the participation status analyzing unit has changed over a plurality of divided periods, and determining which of the plurality of predetermined change patterns corresponds to the plurality of the change patterns A change pattern specifying unit to be specified for each user;
For each of the plurality of change patterns, by cross-counting the number of participants in the campaign in which the user identified as corresponding to any one of the change patterns by the change pattern specifying unit is crossed, A user behavior analysis device in a community comprising a relationship analysis unit that analyzes relationships with a plurality of campaigns.   The participation status analysis unit calculates a participation point by performing a predetermined weighted addition operation on the index values related to the plurality of types of behavior for each of a plurality of divided periods obtained by dividing the analysis target period. 2. The method according to claim 1, wherein the level of participation in the community site of the user is analyzed for each of the plurality of users based on the participation point, and which level corresponds to a plurality of levels of participation. User behavior analysis device in the community of.   The participation status analysis unit performs a predetermined clustering operation on the index values related to the plurality of types of behavior for each of a plurality of divided periods obtained by dividing the analysis target period, thereby allowing the user to participate in the community site. 2. The user behavior analysis device in a community according to claim 1, wherein, as a situation, the type of participation type among a plurality of types is analyzed for each of the plurality of users. The participation status analysis unit includes a teacher data generation unit that generates teacher data as a configuration for performing the clustering operation, and a participation type determination unit that determines the participation type using the teacher data,
The teacher data generation unit aggregates the index values related to the plurality of types of actions in the entire analysis target period for each user, and uses the total value to combine two indexes among the plurality of types of indexes. The center point of a plurality of clusters corresponding to the plurality of types of participation is specified on the coordinate plane specified in step (b), and the center point of the plurality of clusters and the total value obtained for the two indicators are specified. Generating the teacher data by assigning the plurality of users to any one of the plurality of clusters from the relationship with the position on the coordinate plane,
The participation type discriminating unit is configured so that the plurality of users are respectively connected to the plurality of users based on the relationship between the teacher data generated by the teacher data generation unit and the index values related to the plurality of types of actions for each of the plurality of divided periods. 4. The user behavior analysis device in a community according to claim 3, wherein which type of participation type corresponds to the type is determined.   The change pattern specifying unit includes, as the plurality of change patterns, a first pattern in which the participation level is increased, a second pattern in which the participation level is in a maintained state, and a third pattern in which the participation level is decreased. 3. The user behavior analysis device in a community according to claim 2, wherein one of the plurality of users is identified.   The change pattern specifying unit includes, as the plurality of change patterns, a first pattern in which the participation point is rising above a predetermined value, a second pattern in which the participation point is in a maintenance state within a predetermined range, and the participation The user behavior analysis device in a community according to claim 2, wherein for each of the plurality of users, it is specified which of the third patterns the points fall below a predetermined value. For each of the plurality of types of participation types, an index value totaling unit that totals the index values related to the plurality of types of actions related to each user analyzed as belonging to each type of participation by the teacher data generation unit;
An order assignment that calculates a participation point by performing a predetermined weighted addition operation on the total value calculated by the index value totaling unit and gives an order to the plurality of types of participation based on the size of the participation point And further comprising
The change pattern specifying unit includes, as the plurality of change patterns, a first pattern in which the rank order is increasing, a second pattern in which the rank order is in a maintained state, and a rank pattern in which the rank rank is decreasing. The user behavior analysis device in the community according to claim 4, wherein which of the three patterns corresponds to each of the plurality of users. An index value acquisition unit that acquires an index value related to a plurality of types of behavior of a user in a community site on the Internet, and acquires the index value related to the behavior related to the plurality of users in a desired analysis target period;
A participation status information acquisition unit for acquiring participation status information indicating the participation status of the plurality of users in the campaign during the analysis target period;
Participation situation analysis unit that performs a predetermined calculation on the index values related to the plurality of types of behavior and analyzes the participation situation of the user in the community site for each of the plurality of users based on the result of the calculation When,
For each of a plurality of participation situations analyzed by the above-mentioned participation situation analysis unit, by cross-counting the number of participants in the campaign in which the user identified as corresponding to any of the participation situations by the participation situation analysis unit An apparatus for confirming the effectiveness of a campaign, comprising: a relationship analysis unit for analyzing a relationship between the plurality of participation situations and the campaign. Index value acquisition means for acquiring an index value related to a plurality of types of behavior of a user in a community site on the Internet, the index value relating to the behavior related to the plurality of users in a desired analysis target period;
Participation status information acquisition means for acquiring participation status information indicating which campaign of the plurality of users participated in the plurality of campaigns during the analysis target period;
For each of a plurality of divided periods obtained by dividing the analysis target period, a predetermined calculation is performed on the index values related to the plurality of types of behaviors, and based on the result of the calculation, the state of involvement of the user in the community site is determined. A participation status analysis means for analyzing each of the plurality of users;
Analyzing how the participation status analyzed for each of the divided periods by the participation status analyzing means has changed over a plurality of divided periods, and determining which of a plurality of predetermined change patterns corresponds to Cross-tabulation of the number of participants in the campaign in which the user who has been identified as one of the change patterns by the change pattern specifying unit is participating for each of the plurality of change patterns. A user behavior analysis program in a community for causing a computer to function as a relationship analysis means for analyzing a relationship between the plurality of change patterns and the plurality of campaigns.

Images