WO2009110253A1 - Information recommendation system, information recommendation server device, and information recommendation method - Google Patents

Abstract

This system aims to narrow information utilizing the viewing tendency of an individual user relative to all the users. An information index calculation means (12) calculates an information index being an index based on a viewed tendency being how the information is viewed with respect to each information, from the viewing histories of the information by a plurality of users. A user index calculation means (13) calculates a user index based on the viewing tendency being how the user views information, from the information index calculated about the information viewed by the user with respect to each user. An index matching means (16) extracts information having an information index similar to the user index calculated about the user from a group of recommendation object information.

Description

Information recommendation system, information recommendation server device, and information recommendation method

The present invention relates to an information recommendation system for information published on a network, and in particular, information using a viewed tendency as a method of browsing individual information and a browsing tendency as a method of browsing by individual users. Regarding the recommendation system.

When the amount of information disclosed on the network becomes enormous, it is impossible to browse all the information disclosed in time. Also, random viewing results in a lot of wasted time. For this reason, a technique for narrowing down information to be browsed preferentially from a large number of information disclosed on the network is important.

A typical example of this type of technology is keyword extraction. Information extraction using keywords is particularly effective when the field or range of information to be extracted is determined to some extent. Moreover, if the field of information browsed in the past and the keywords used are extracted from the browsing history of each user, and the information is periodically searched using the extracted keywords, etc., the browsing tendency of the users is reflected. Information can be narrowed down.

An example of another technique related to information narrowing is described in Patent Document 1. In Patent Document 1, a third party can browse the browsing history of others such as experts, so that a useful web page can be browsed without special skills. For example, referring to a doctor's browsing history may help extract medical information.
JP 2005-148843 A

The method described in Patent Document 1 is effective depending on how it is used because it enables browsing of information similar to that of an expert simply by selecting the browsing history of the expert. However, since the browsing tendency of each user is not considered at all, it is not possible to narrow down information reflecting the browsing tendency of the user.

On the other hand, according to the method of extracting keywords or the like browsed in the past from the browsing history of the user and searching for information using the extracted keywords or the like, it is possible to narrow down the information reflecting the browsing tendency of the user Become. However, the browsing tendency of the user in this case is a browsing tendency regarding the content of information.

The browsing tendency of the user has a browsing tendency determined by the relationship with other users in addition to the browsing tendency regarding the contents of information as described above. For example, if you prefer to browse information that is not viewed by others, and vice versa, you may prefer to browse information that many people are browsing. It is an example of a simple browsing tendency. If it becomes possible to narrow down information using such a relative browsing tendency, it becomes possible to narrow down information effectively for a user who has a particularly remarkable browsing tendency.

The present invention has been proposed in view of such circumstances, and an object of the present invention is to make it possible to narrow down information by using the relative browsing tendency of individual users viewed from all users. It is in.

The information recommendation server device of the present invention calculates an information index, which is an index based on a browsing tendency, which is how the information is browsed, for each piece of information from the browsing history of information by a plurality of users. A user index for calculating a user index based on a browsing tendency, which is a method of browsing the information of the user, from the information index calculated for the information browsed by the user for each user. Computation means, and index matching means for extracting information having an information index similar to the calculated user index for the user from a recommendation target information group.

The information recommendation system of the present invention transmits the information recommendation request including the user identifier to the information recommendation server device of the present invention and the information recommendation server device, and the information recommendation result including the information identifier returned as a response thereto. A terminal device that receives the information, and an information providing server device that provides the terminal device with information specified by the information identifier.

In the information recommendation method of the present invention, a) the information index calculation means is an index based on the browsing tendency, which is how the information is browsed for each piece of information from the browsing history of information by a plurality of users. A step of calculating an information index, and b) a user index calculation means for browsing the user information from the information index calculated for the information browsed by the user for each user. A step of calculating a user index based on a browsing tendency; and c) a step of index matching means extracting, from the recommendation target information group, information having an information index similar to the calculated user index for the user. Including.

According to the present invention, it is possible to narrow down information using the relative browsing tendency of individual users viewed from all users. The reason is that, from the browsing history, an information index that is an index based on the browsing tendency, which is how individual information is browsed, is calculated, and for each user, the information index calculated for the information browsed by the user is calculated. In order to calculate a user index based on a browsing tendency that is a way of browsing the information of the user, and to extract information having an information index similar to the user index calculated for the user from the recommended target information group is there.

It is a block diagram of a 1st embodiment of the present invention. It is a flowchart which shows an example of operation | movement of the form of the 1st form of this invention. It is a flowchart which shows the procedure which calculates the recognition parameter | index which is one of the information parameters | indexes in the 1st Embodiment of this invention. It is a flowchart which shows the procedure which calculates the fixing parameter | index which is another one of the information parameter | index in the 1st Embodiment of this invention. It is a flowchart which shows the procedure which calculates the user parameter | index in the 1st Embodiment of this invention. It is a flowchart which shows the process example of the parameter | index matching means in the 1st Embodiment of this invention. It is a figure which shows the specific example of the information browsing history in the 1st Embodiment of this invention. It is a figure which shows the type | formula which calculates | requires the distance between the indices in the 1st Embodiment of this invention. On the plane centered on the recognition index and the establishment index, the information index of the provided information browsed by the entire user, the information index of the provided information viewed by a certain user, and the user of the user It is the figure which plotted the parameter | index (double-circle). It is a block diagram of the 2nd Embodiment of this invention. It is a figure which shows the specific example of the information browsing history in the 2nd Embodiment of this invention. It is a flowchart which shows an example of the operation | movement of the 2nd Embodiment of this invention. It is a flowchart which shows the procedure which calculates the residence parameter | index which is one of the information parameters | indexes in the 2nd Embodiment of this invention. It is a flowchart which shows the procedure which calculates the user parameter | index in the 2nd Embodiment of this invention. It is a figure which shows the type | formula which calculates | requires the distance between the indices in the 2nd Embodiment of this invention. It is a block diagram of the 3rd Embodiment of this invention. It is a figure which shows the specific example of the information browsing history in the 3rd Embodiment of this invention. It is a flowchart which shows an example of the operation | movement of the 3rd Embodiment of this invention. It is a flowchart which shows the procedure which calculates the time difference parameter | index which is the information parameter | index in the 3rd Embodiment of this invention. It is a flowchart which shows the procedure which calculates the user parameter | index in the 3rd Embodiment of this invention. It is a figure which shows the type | formula which calculates | requires the distance between the indices in the 3rd Embodiment of this invention. It is a block diagram of the 4th Embodiment of this invention. It is a flowchart which shows the example of a process of the information query means in the 4th Embodiment of this invention, and a parameter | index matching means. It is a specific example of the information parameter | index in Example 1 of this invention. It is a specific example of the user parameter | index in Example 1 of this invention.

Explanation of symbols

1, 1A to 1C: Information recommendation servers 11, 11A, 11B ... Browsing history storage units 12, 12A, 12B ... Information index calculation means 13, 13A, 13B ... User index means 14, 14A, 14B ... Index storage unit 15, 15A, 15C ... information query means 16, 16A-16C ... index matching means 2 ... network 3 ... information providing server 4 ... terminals 41, 41C ... information reference means 101-104 ... information recommendation system

Next, the best mode for carrying out the present invention will be described in detail with reference to the drawings.

“First Embodiment”
Referring to FIG. 1, an information recommendation system 101 according to the first embodiment of the present invention includes an information recommendation server 1, an information providing server 3, a terminal 4, and a network 2 connecting them. Yes.

The information providing server 3 is a computer that provides information to the terminal 4 through the network 2. The type of information to be provided is not limited to documents, but may be multimedia information such as images and sounds. Hereinafter, each piece of information provided by the information providing server 3 is referred to as provision information. Although one information providing server 3 is shown in FIG. 1, a plurality of information providing servers 3 may exist.

The network 2 connects the information recommendation server 1, the information providing server 3, and the terminal 4 so that they can communicate with each other. The network 2 may be a public network or a closed network such as a LAN.

The terminal 4 is a computer such as a personal computer or a mobile phone, and includes information reference means 41. The information reference unit 41 has a function of browsing information provided by the information providing server 3, a function of transmitting an information recommendation request to the information recommendation server 1 according to a request from the user, and a response transmitted from the information recommendation server 1. Receiving information recommendation results and presenting them to the user through a display or the like. The information recommendation request transmitted to the information recommendation server 1 includes a user identifier for identifying the user. The information recommendation result received from the information recommendation server 1 includes an information identifier for identifying information provided by the information providing server 3. The information reference means 41 may be realized by dedicated hardware or software, or a general-purpose web browser may be used.

The information recommendation server 1 receives the information recommendation request transmitted from the terminal 4, and sends the information recommendation result including the identifier of the provided information suitable for the user specified by the user identifier included in the recommendation request to the terminal 4. It is a computer with a function to return. The information recommendation server 1 according to the present embodiment includes a browsing history storage unit 11, an information index calculation unit 12, a user index calculation unit 13, an index storage unit 14, an information query unit 15, an index matching unit 16, and the like. It has. Each of these means has the following functions.

The browsing history storage unit 11 stores an information browsing history that is a history of browsing information provided by the user of the information recommendation system 101 on the information providing server 3. Here, in the case of the present embodiment, the information browsing history is a set of a user identifier, an information identifier of the browsed provided information, and a browse date and time when the provided information is browsed. As such information browsing history, a browsing log in the information providing server 3 or an access log such as a proxy server connected to the network 2 can be used. As the information identifier, a URI may be used, or a document ID in a document management system may be used.

The information index calculation means 12 uses the information browsing history stored in the browsing history storage unit 11 to calculate an information index representing a viewed tendency, which is how each information is browsed, and this calculated information index. Is stored in the index storage unit 14.

The user index calculation means 13 uses the information browsing history stored in the browsing history storage unit 11 and the information index stored in the index storage unit 14 to browse each user's information. A user index representing a tendency is calculated, and the calculated user index is stored in the index storage unit 14.

The index storage unit 14 stores the information index for each provided information calculated by the information index calculation unit 12 and the user index for each user calculated by the user index calculation unit 13.

The information query unit 15 receives an information recommendation request including a user identifier from the terminal 4 and transmits the user identifier to the index matching unit 16. Also, the information recommendation result transmitted from the index matching means 16 is received, and this information recommendation result is transmitted to the terminal 4 as the recommendation request source.

The index matching unit 16 receives the user identifier transmitted from the information query unit 15, searches the index storage unit 14 for the user index of the user specified by the user identifier, and searches for the user index searched for. Are matched with the information index of each provided information in the index storage unit 14, and a list of information identifiers of provided information having an information index similar to the user index is extracted. The list of information identifiers is transmitted to the information query unit 15 as an information recommendation result.

Next, the overall operation of this embodiment will be described in detail. It is assumed that the information browsing history as shown in FIG. 7 is stored in advance in the browsing history storage unit 11 of the information recommendation server 1.

Referring to the flowchart of FIG. 2, the information index calculation unit 12 acquires a browsing history from the browsing history storage unit 11 (S21), and calculates an information index for each provided information using the acquired browsing history. (S22, S23).

In this embodiment, the information index is composed of a set of two indices, a recognition index and a retention index.

Here, the recognition index is an index indicating how many users of the user group of the information recommendation system 101 have viewed the provided information during a predetermined period d, and which provided information is It only represents what is widely recognized. The recognition index is calculated according to the flowchart of FIG.

Further, the fixing index is an index indicating how many times each user of the information recommendation system 101 browses the provided information during the period d, and how long the provided information is browsed. To express. Here, in order to prevent an arbitrary recommendation result from being manipulated, even if a plurality of times of browsing within a unit time are viewed, it is handled that only one browsing has been performed within the unit time. The fixing index is calculated according to the flowchart of FIG.

The information index calculation means 12 stores, for each provided information, the information identifier of the provided information, the calculated recognition index, and the established index in the index storage unit 14 (S24).

Here, with reference to the flowchart of FIG. 3 and the flowchart of FIG. 4, a calculation method of the recognition index and the establishment index will be described. In addition, in calculating each index, the total number of users of the information recommendation system 101 is acquired in advance.

Referring to the flowchart of FIG. 3, the information index calculation unit 12 pays attention to one provided information recorded in the browsing history storage unit 11 and extracts the total number of users who browsed the provided information at least once in the period d. (S32). Next, a ratio of the total number of extracted users to the total number of users of the information recommendation system 101 is calculated, and this ratio is set as a recognition index of the provided information (S33). The information index calculation means 12 repeats such calculation for all the remaining individual provision information recorded in the browsing history storage unit 11 (S31s, S31e).

Referring to the flowchart of FIG. 4, the user index calculation means 13 pays attention to one provided information and one user recorded in the browsing history storage unit 11 and determines whether or not the user has browsed the provided information. The unit time is calculated for each unit time in the period d (S43), the total number of unit times that have been browsed is calculated (S44), and the ratio of the calculated total number to the total number of unit times in the period d As a fixing index related to the user (S45). For the same provision information, the fixing index for the remaining users is calculated in the same manner (S42s, S42e), and the average value of the fixing index of all the users for the provision information is calculated, and the calculated average value is used as the provision index. It is set as an information fixing index (S47). Note that a user who does not browse the provided information at all during the period d (a user whose fixing index calculated in step S45 is 0) is excluded from the elements for calculating the average value (S46). The reason is that the fixed index of certain provided information is an index indicating how many times the provided information has been browsed on the assumption that the provided information has been browsed. The user index calculation means 13 repeats the same calculation for all remaining provision information (S41s, S41e).

Referring to the flowchart of FIG. 5, the user index calculation unit 13 acquires the browsing history from the browsing history storage unit 11 (S51), and further provides the information browsed from the index storage unit 14 for each individual user (S52s). Are obtained (S53), and a user index is calculated (S54).

In the present embodiment, the user index is composed of a set of two indexes, a recognition index and a retention index, in the same manner as the information index. Here, the recognition index of the user index is calculated as the average value of the recognition index of the provided information that has been viewed by each individual user, and the retention index of the user index is the information provided by the individual user. Calculated as the average value of the fixing index. The user index represents a browsing tendency indicating how each individual user is viewing information that is positioned for the user group.

The user index calculation means 13 stores the user identifier of the user and the user index (recognition index, fixing index) of the user in the index storage unit 14 (S55).

Referring to the flowchart of FIG. 6, when the index matching unit 16 receives the user identifier transmitted from the information query unit 15 (S61), the index matching unit 16 receives the user identifier specified by the received user identifier from the index storage unit 14. A user index is acquired, and information indexes of all provided information are extracted from the index storage unit 14 (S62).

Next, the index matching means 16 pays attention to the extracted one piece of provided information, and calculates the distance between the information index of the provided information and the user index of the user (S64). In this embodiment, since the user index and the information index are composed of a combination of a recognition index and a retention index, the distance between the user index and the information index is on a plane around the recognition index and the retention index. It is a distance and is calculated by the formula shown in FIG. The index matching unit 16 calculates the distance between the information index and the user index for all the remaining extracted extracted information (S63s, S63e). Finally, the index matching unit 16 generates a list in which the information identifiers of the provided information are arranged in ascending order of the distance, and returns the list to the information query unit 15 (S65).

Here, the index matching means 16 may set a certain threshold value and exclude the information identifier of the provided information whose distance is larger than the threshold value from the list. In addition, the index matching unit 16 may remove the identifier of the provided information that the user has already browsed from the generated list. Which provided information in the list has been browsed is determined by searching the browsing history storage unit 11 with the set of information identifier and user identifier of the provided information in the list, and the corresponding set is stored in the browsing history storage unit 11. If so, it can be determined that it has been browsed, and if not, it has not been browsed.

Next, the effect of this embodiment will be described.

According to the present embodiment, it is possible to recommend the information in consideration of the positioning of the individual provided information for the user group and the browsing tendency of the individual users. The reason for this is that an information index, which is how each of the provided information is browsed, is calculated from the browsing history of the provided information of the user group, and the usage is calculated from the average of the information indices of one or more provided information viewed by each user. This is because the user information of the user is calculated, and the provided information having the information index similar to the user index is used as the recommended information.

In this embodiment, a recognition index is used as an information index and a user index. The recognition index of information is generally an index indicating whether or not the information is viewed by many people, and the recognition index of the user index is generally a tendency to browse only information viewed by many people. It is an index indicating whether there is a tendency to browse exclusively information that has not been browsed much. Therefore, for users who tend to browse the information browsed by many people, the information provided by the information providing server 3 can recommend the information viewed by many people. It is possible to recommend information that has not been browsed so much to users who tend to browse information that has not been browsed so much.

Also, in this embodiment, a fixing index is used as an information index and a user index. The fixing index is generally an index indicating whether or not the information is repeatedly viewed when viewed, and the fixing index of the user index generally browses information that is repeatedly viewed many times. It is an index indicating whether there is a tendency, or whether the information tends to be browsed exclusively for information that is not browsed many times. Therefore, for users who tend to exclusively browse information that is repeatedly viewed, information that is repeatedly viewed among the information provided by the information providing server 3 can be recommended. It is possible to recommend information that has not been browsed many times to users who tend to browse information that has not been browsed many times.

In this embodiment, since a set of a recognition index and a fixing index is used, for example, information that is viewed by many people and that is repeatedly viewed many times when viewed is exclusively viewed. Can recommend such information to users who are browsing the information that has been viewed by many people and has not been viewed repeatedly. Can recommend such information to users who are exclusively browsing information that has not been viewed by many people and that is repeatedly viewed when viewed. Information that can be recommended and recommended to users who are browsing information that has not been browsed by many people and that has not been browsed too often when viewed. it can.

FIG. 9 shows an information index of provided information browsed by the entire user, an information index of provided information browsed by a certain user, and a user on a plane around the recognition index and the established index. The user index ◎ is plotted. In the case of this example, provision information having an information index ★ in the vicinity of the user index ◎ is recommended for the user.

In this embodiment, since the information index is calculated based on the number of browsing without the duplication, an arbitrary recommendation result operation by a specific user intentionally browsing specific information, etc. Can be prevented.

“Second Embodiment”
Referring to FIG. 10A, the information recommendation system 102 according to the second exemplary embodiment of the present invention is compared with the information recommendation system 101 according to the first exemplary embodiment illustrated in FIG. Instead, the information recommendation server 1A is provided.

Compared with the information recommendation server 1 of the information recommendation system 101 according to the first embodiment, the information recommendation server 1A is a browsing history storage unit 11, an information index calculation unit 12, a user index calculation unit 13, and an index storage unit. 14. Instead of the index matching unit 16, a browsing history storage unit 11A, an information index calculation unit 12A, a user index calculation unit 13A, an index storage unit 14A, and an index matching unit 16A are provided. Each of these means has the following functions.

The browsing history storage unit 11 </ b> A stores an information browsing history that is a history of browsing information provided by the user of the information recommendation system 102 on the information providing server 3. Here, in the case of the present embodiment, the information browsing history is a set of a user identifier, an information identifier of the browsed information, a browsing date and time when the information is browsed, and a residence time, as illustrated in FIG. 10B. That is, the residence time is added to the information browsing history stored in the browsing history storage unit 11 in the first embodiment. The residence time is a time during which the user has browsed the provided information. As the information browsing history including such residence time, an access log such as a browsing log in the information providing server 3 or a proxy server connected to the network 2 can be used.

The information index calculation unit 12A uses the information browsing history stored in the browsing history storage unit 11A as an information index that represents a browsing tendency, which is how each information is browsed, as a recognition index and a fixing index. In addition, a residence index is calculated, and the calculated information index is stored in the index storage unit 14A. Here, the retention index of a certain provided information is an index representing how much time the user spends browsing the provided information. The staying index may be used as a standard for advertisement placement.

The user index calculation means 13A uses the information browsing history stored in the browsing history storage unit 11A and the information index stored in the index storage unit 14A, and is a browsing method for browsing individual user information. A user index representing a trend is calculated, and the calculated user index is stored in the index storage unit 14A. Here, in the case of the present embodiment, the user index is a set of a recognition index, a fixing index, and a staying index.

The index storage unit 14A stores the information index for each provided information calculated by the information index calculation unit 12A and the user index for each user calculated by the user index calculation unit 13A.

The information query unit 15 is the same as that in the first embodiment, receives an information recommendation request including a user identifier from the terminal 4, and transmits the user identifier to the index matching unit 16A. Further, the information recommendation result transmitted from the index matching means 16A is received, and this information recommendation result is transmitted to the terminal 4 as the recommendation request source.

The index matching unit 16A receives the user identifier transmitted from the information query unit 15, searches the index storage unit 14A for the user index of the user specified by the user identifier, and searches the user index. Are matched with the information indexes of the individual provided information in the index storage unit 14A, and a list of information identifiers of the provided information having information indexes similar to the user indexes is extracted. The list of information identifiers is transmitted to the information query unit 15 as an information recommendation result.

Next, the overall operation of this embodiment will be described in detail. It is assumed that the browsing history storage unit 11A of the information recommendation server 1A stores information browsing history as shown in FIG. 10B in advance.

Referring to the flowchart of FIG. 11, the information index calculation unit 12A acquires the browsing history from the browsing history storage unit 11A (S111), and calculates the information index for each piece of information using the acquired browsing history ( S22, S23, S114).

In this embodiment, the information indicator is composed of a set of three indicators: a recognition indicator, a retention indicator, and a stay indicator. The meaning and calculation method of the two indicators, the recognition indicator and the establishment indicator, are the same as those in the first embodiment. The fixing index is calculated according to the flowchart of FIG.

Referring to the flowchart of FIG. 12, the information index calculation unit 12A pays attention to one provided information recorded in the browsing history storage unit 11A, and calculates the sum of the residence time for each browsing of the provided information (S122). ). Next, a quotient obtained by dividing the sum of the residence times by the product of the unit time and the number of browsing times is obtained, and the obtained quotient is used as a residence index of the provided information (S123). The information index calculation unit 12A repeats such calculation for all the remaining individual provision information recorded in the browsing history storage unit 11 (S121s, S121e).

Finally, the information index calculating unit 12A stores the information identifier of the information, the calculated recognition index, the fixing index, and the staying index in the index storage unit 14A (S115).

Next, referring to the flowchart of FIG. 13, the user index calculation means 13A acquires the browsing history from the browsing history storage unit 11A (S131), and further browses individual users (S132s) from the index storage unit 14A. The information index of the obtained information is acquired (S133), and the user index is calculated (S134).

In the present embodiment, the user index is composed of a set of three indicators, a recognition indicator, a retention indicator, and a stay indicator, in the same manner as the information indicator. The meaning and calculation method of the user index recognition index and the retention index are the same as those in the first embodiment. The retention index of the user index is the average value of the retention index of the provided information that has been viewed by individual users, and the browsing of how much information each user has appealed to the user group Represents a trend.

The user index calculation unit 13A stores the user identifier of the user and the user index of the user as a set in the index storage unit 14A (S135).

The index matching unit 16A performs the same operation as the index matching unit 16 in the first embodiment except for the following points. The index matching means 16A uses the formula shown in FIG. 14 when calculating the distance between the user index and the information index in step S64 of FIG. That is, in the present embodiment, the distance between the user index and the information index is a distance in a space with the recognition index, the fixing index, and the staying index as axes.

According to the present embodiment, the same effects as the first embodiment can be obtained, and at the same time, the following effects can be obtained.

In this embodiment, retention is used as an information index and a user index. The retention index of the information index is generally an index indicating whether or not the information is viewed for a long time by one browsing, and the recognition index of the user index generally browses information that is viewed for a long time by one browsing. It is an index that indicates whether there is a tendency, or whether there is a tendency to browse exclusively information that has not been browsed for a long time in a single browsing. Therefore, for a user who tends to browse information that is browsed for a long time only once, information that has been browsed for a long time can be recommended out of the information provided by the information providing server 3. For users who tend to browse information that is not browsed for a long time by one browsing, it is possible to recommend information that is not browsed for a long time by browsing once, among the information provided by the information providing server 3.

Further, in this embodiment, since a set of a recognition index, a fixing index, and a staying index is used, for example, information that has been browsed by many people and is browsed over and over again when viewed. For users who are exclusively browsing information that is browsed for a long time, it is possible to recommend a method of recommending such information.

“Third embodiment”
Referring to FIG. 15, the information recommendation system 103 according to the third exemplary embodiment of the present invention is compared with the information recommendation system 101 according to the first exemplary embodiment illustrated in FIG. Instead, it is different in that an information recommendation server 1B is provided.

Compared with the information recommendation server 1 of the information recommendation system 101 according to the first embodiment, the information recommendation server 1B is a browsing history storage unit 11, an information index calculation unit 12, a user index calculation unit 13, and an index storage unit. 14. Instead of the index matching unit 16, a browsing history storage unit 11B, an information index calculation unit 12B, a user index calculation unit 13B, an index storage unit 14B, and an index matching unit 16B are provided. Each of these means has the following functions.

The browsing history storage unit 11 </ b> B stores an information browsing history that is a history of browsing provided information published on the information providing server 3 by a user of the information recommendation system 103. Here, in the case of the present embodiment, the information browsing history is a set of a user identifier, an information identifier of the browsed information, a browsing date and time of browsing the information, and a time difference, as illustrated in FIG. That is, a time difference is added to the information browsing history stored in the browsing history storage unit 11 in the first embodiment. The time difference is a time indicating how much the user browses the provided information with a delay from the reference time. The reference time can be the time when the provided information is browsed for the first time or when the provided information can be disclosed on the information providing server 3. As an information browsing history including such a time difference, an access log such as a browsing log in the information providing server 3 or a proxy server connected to the network 2 can be used. Note that the browsing date and time in FIG. 16 is the browsing date and time used when calculating the time difference, and may be deleted after the time difference is calculated.

The information index calculation means 12B uses the information browsing history stored in the browsing history storage unit 11B to calculate a time difference index as an information index that represents a browsing tendency, which is how each information is browsed. The calculated information index is stored in the index storage unit 14B. Here, the time difference index of certain provision information is an index representing how early the provision information is referred to.

The user index calculation unit 13B uses the information browsing history stored in the browsing history storage unit 11B and the information index stored in the index storage unit 14B, and is a browsing that is a way of browsing information of individual users. A user index representing a trend is calculated, and the calculated user index is stored in the index storage unit 14B. Here, the user index is a time difference index in the present embodiment.

The index storage unit 14B stores the information index for each provided information calculated by the information index calculation unit 12B and the user index for each user calculated by the user index calculation unit 13B.

The information query unit 15 is the same as that in the first embodiment, receives an information recommendation request including a user identifier from the terminal 4, and transmits the user identifier to the index matching unit 16B. Further, the information recommendation result transmitted from the index matching means 16B is received, and this information recommendation result is transmitted to the terminal 4 as the recommendation request source.

The index matching unit 16B receives the user identifier transmitted from the information query unit 15, searches the index storage unit 14B for the user index of the user specified by the user identifier, and searches the user index. Is matched with the information index of each provision information in the index storage unit 14B, and a list of information identifiers of the provision information having the information index similar to the user index is extracted. The list of information identifiers is transmitted to the information query unit 15 as an information recommendation result.

Next, the overall operation of this embodiment will be described in detail. It is assumed that the information browsing history as shown in FIG. 16 is stored in advance in the browsing history storage unit 11B of the information recommendation server 1B.

Referring to the flowchart of FIG. 17, the information index calculation unit 12B acquires the browsing history from the browsing history storage unit 11B (S171), and calculates the information index for each piece of information using the acquired browsing history ( S172).

In this embodiment, the information index is composed of a time difference index. The time difference index is calculated according to the flowchart of FIG.

Referring to the flowchart of FIG. 18, the information index calculation unit 12B pays attention to one provided information recorded in the browsing history storage unit 11, and calculates the sum of time differences for each user who browsed the provided information (S182). ). Next, a quotient obtained by dividing the sum of the time differences by the number of browsed users is obtained, and the obtained quotient is set as a time difference index of the provided information (S183). The information index calculation unit 12B repeats such calculation for all the remaining individual provision information recorded in the browsing history storage unit 11 (S181s, S181e).

Finally, the information index calculation means 12B sets the information identifier of the information and the calculated time difference index as a set and stores them in the index storage unit 14B (S173).

Next, referring to the flowchart of FIG. 19, the user index calculation unit 13B acquires the browsing history from the browsing history storage unit 11B (S191), and further browses individual users (S192s) from the index storage unit 14B. The information index of the obtained information is acquired (S193), and the user index is calculated (S194).

In this embodiment, the user index is composed of a time difference index. The time difference index of the user index represents a browsing tendency indicating how quickly each user is browsing the provided information. The time difference index of a certain user is calculated as an average value of the time difference indices of the provided information browsed by the user.

The user index calculation means 13B sets the user identifier of the user and the user index of the user as a set and stores them in the index storage unit 14B (S195).

The index matching unit 16B performs the same operation as the index matching unit 16 in the first embodiment except for the following points. The index matching means 16B uses the formula shown in FIG. 20 when calculating the distance between the user index and the information index in step S64 of FIG.

Next, the effect of this embodiment will be described.

According to the present embodiment, it is possible to recommend the information in consideration of the positioning of the individual provided information for the user group and the browsing tendency of the individual users. The reason for this is that the time difference index, which is how the individual provided information is browsed, is calculated from the browsing history of the provided information of the user group as an information index, and the average of the time difference indices of one or more provided information viewed by each user This is because the time difference index of the user is calculated as the user index, and the provided information having the information index similar to the user index is used as the recommended information.

In this embodiment, a time difference index is used as an information index and a user index. The time difference index of information is generally an index indicating how early the information is browsed, and the time difference index of the user index generally tends to browse information browsed early. It is an index indicating whether or not it tends to browse exclusively information that has not been browsed early. Therefore, for users who tend to browse information browsed early, the information provided by the information providing server 3 recommends information that has been browsed early after it has been published or since it was first viewed. On the contrary, for users who tend to browse information that has not been browsed at an early stage, the information provided by the information providing server 3 has not been browsed at an early stage after being published or first viewed. It becomes possible to recommend information.

“Fourth embodiment”
Referring to FIG. 21, the information recommendation system 104 according to the fourth exemplary embodiment of the present invention is compared with the information recommendation system 101 according to the first exemplary embodiment illustrated in FIG. Instead, the information recommendation server 1C is provided. Further, the terminal 4 includes an information reference unit 41C instead of the information reference unit 41.

The information reference unit 41C of the terminal 4 is different from the information reference unit 41 in that a keyword serving as a recommendation condition is further included in the information recommendation request in addition to the user identifier. The keyword is specified by the user, for example.

Compared with the information recommendation server 1 of the information recommendation system 101 according to the first embodiment, the information recommendation server 1C includes an information query unit 15C and an index matching unit 16C instead of the information query unit 15 and the index matching unit 16. And the difference is that an index storage unit 17 is newly provided.

The index storage unit 17 stores an index of provided information provided by the information providing server 3. Specifically, for example, a list of keywords included in the provided information is stored in association with the information identifier of the provided information.

The information query unit 15C and the index matching unit 16C extract the recommended information based on the suitability of the keyword in addition to the similarity between the information index and the user index, and thus the information query unit 15 according to the first embodiment. And the index matching means 16 is different.

Referring to the flowchart of FIG. 22, upon receiving an information recommendation request including a user identifier and a keyword from the terminal 4 (S221), the information query unit 15C obtains the information identifier of the provided information including the keyword from the index storage unit 17. Extraction is performed (S222), and the received user identifier and the list of extracted information identifiers are transmitted to the index matching means 16C.

When the index matching unit 16C receives the user identifier and the information identifier list transmitted from the information query unit 15C (S223), the user index of the user specified by the received user identifier from the index storage unit 14 is obtained. And the information index of the information specified by the received information identifier is further extracted from the index storage unit 14 (S224). Next, the index matching unit 16C calculates the distance between the user index of the user and the information index of the extracted information (S225s, S226, S225e). Here, the distance between the user index and the information index is calculated by the equation shown in FIG. 8 as in the first embodiment. Further, the index matching unit 16C returns a list of information identifiers as an information recommendation result to the information query unit 15C in order of increasing distance, and the information query unit 15C returns the information recommendation result to the terminal 4 (S227).

Unlike the process of FIG. 22, the information query unit 15C performs a process of removing the information identifier of information that does not include a keyword from the list of information identifiers returned from the index matching unit 16 of the first embodiment. May be. Further, as in the first embodiment, the information identifier of the provided information that has already been browsed by the user may be removed from the list of information identifiers.

According to the present embodiment, it is possible to obtain the same effects as those of the first embodiment, and at the same time, it is possible to narrow down the recommended candidates by the keyword specified by the user of the terminal 4.

Although this embodiment is based on the first embodiment, the index storage unit 17 is added to the second or third embodiment to narrow down recommendation candidates by keywords as in this embodiment. An embodiment in which it is performed is also conceivable.

Next, Example 1 of the present invention will be described using specific examples. Example 1 corresponds to the first embodiment shown in FIG.

In this embodiment, there are five users of the information recommendation system, user identifiers 000001 to 000005, and the period d at the time of calculating the information index is 3 days from September 23, 2007 to the same day 25, and the unit The time is one day.

In the browsing history storage unit 11, the information browsing history shown in FIG. 7 is stored in advance.

The information index calculation means 12 acquires the information browsing history of FIG. 7 from the browsing history storage unit 11 (S21), and calculates the information index (S22). First, a recognition index is calculated for the provided information specified by the information identifier http://aaa.jp (S31s). Since the users who have browsed the information specified by the information identifier http://aaa.jp in the period d are four users specified by the user identifiers 000001, 000002, 000003, and 000004 (S32), the information identifier http The information index of information specified by: //aaa.jp is 4/5 = 0.8 (S33). Similarly, since the users who have browsed the information specified by the information identifier http://bbb.jp (S31s) are the three names specified by the user identifiers 000001, 000002, and 000003 (S32), the information identifier The information index of information specified by http://bbb.jp is 3/5 = 0.6 (S33).

Next, the fixing index is calculated in the order of provision information specified by the information identifier http://aaa.jp and provision information specified by the information identifier http://bbb.jp (S41s). Since the user specified by the user identifier 000001 is browsing the provided information specified by the information identifier http://aaa.jp (S42s) on September 24, 2007 and the same day (S43). The total number of unit times that the user specified by the user identifier 000001 browses the provided information specified by the information identifier http://aaa.jp is 2 (S44). Therefore, the fixing index for the provided information specified by the user information identifier http://aaa.jp specified by the user identifier 000001 is 2/3 = 0.67 (S45). Similarly, the fixing index for the provided information specified by the information identifier http://aaa.jp of the user (S42s) specified by the user identifier 000002 is 2/3 = 0.67 (S43, S44, S45), the fixing index for the provided information specified by the information identifier http://aaa.jp of the user (S42s) specified by the user identifier 000003 is 1/3 = 0.33 (S43, S44). , S45), the fixing index for the provided information specified by the information identifier http://aaa.jp of the user (S42s) specified by the user identifier 000004 is also 1/3 = 0.33 (S43, S44, S45). Since the user specified by the user identifier 000005 is not browsing the provision information specified by the information identifier http://aaa.jp in the period d, the user is specified by the information identifier http://aaa.jp. The calculation of the fixing index for the provided information is excluded from the calculation target (S46).

When an average value is calculated for the fixing index for each user for the provided information specified by the calculated information identifier http://aaa.jp (S47), (0.67 + 0.67 + 0.33 + 0.33) / 4. = 0.5, and this is set as a fixing index of the provided information specified by the information identifier http://aaa.jp (S23).

Similarly, the fixing index of information specified by the information identifier http://bbb.jp is ((2/3) + (2/3) + (2/3)) / 3 = 0.66 (FIG. 4).

Finally, the information identifier, the calculated recognition index, and the fixing index are paired and stored in the index storage unit 14 (S24). An example of the contents of the index storage unit 14 at this time is shown in FIG.

Next, calculate the user index for each individual user. The user index calculation unit 13 acquires the information browsing history shown in FIG. 7 from the browsing history storage unit 11.

First, for the user specified by the user identifier 000001 (S52s), the calculated information index shown in FIG. 23 is acquired from the index storage unit 14, and the recognition index and the fixing index are calculated. The information identifiers of the information browsed by the user specified by the user identifier 000001 in the period d are http://aaa.jp and http://bbb.jp, and the respective information indices (recognition index and establishment index) Are (0.8, 0.5) and (0.6, 0.66) (S53). When the average values of the recognition index and the fixing index are obtained for the information index, (0.7, 0.58) is obtained (S54). The calculated set of average values is stored in the index storage unit 14 as a user index (S24).

Similarly, an information index is calculated for other information, a user index is calculated for other users, and stored in the index storage unit 14. An example of the contents of the index storage unit 14 at this time is shown in FIGS.

Next, a case where the user specified by the user specified by the user identifier 000001 receives information recommendation will be described.

The user specified by the user identifier 000001 uses the information reference means 41 of the terminal 4 to transmit an information recommendation request including its own user identifier 000001 to the information recommendation server 1.

The information query unit 15 receives the information recommendation request from the terminal 4 and transmits the user identifier 000001 included in the request to the index matching unit 16.

When the index matching unit 16 receives the user identifier 000001 from the information query unit 15 (S61), the user index (0.7, 0.58) of the user specified by the user identifier 000001 from the index storage unit 14 is obtained. Is acquired (S62).

The distance between the acquired user index (0.7, 0.58) and the information index (0.8, 0.5) of the information specified by the information identifier http://aaa.jp is shown in FIG. When calculated according to the above formula, 0.13 is obtained. Similarly, the distance between the information index (0.6, 0.66) of the information specified by the information identifier http://bbb.jp is 0.13, and the information specified by the information identifier http://ccc.jp The distance from the information index (0.1, 0.9) is 0.68 (S63s, S64, S63e). Therefore, the index matching unit 16 arranges information identifiers in the order of {http://aaa.jp, http://bbb.jp, http://ccc.jp}, and sends the information identifier to the terminal 4 via the information query unit 15. return.

Although the embodiments and examples of the present invention have been described above, the present invention is not limited to the above examples, and various other additions and modifications can be made. For example, as an information index and a user index, a combination of a recognition index and a fixing index in the first embodiment, a pair of a recognition index, a fixing index, and a staying index in the second embodiment, a third embodiment In this example, the time difference index is used. However, any one of the recognition index, the fixation index, the staying index, and the time difference index can be used alone or in any combination.

In addition, the information recommendation server of the present invention can be realized by a computer and a program as well as the functions of the information recommendation server by hardware. The program is provided by being recorded on a computer-readable recording medium such as a magnetic disk or a semiconductor memory, and is read by the computer at the time of starting up the computer, etc. In the form of an information recommendation server.

This application claims priority based on Japanese Patent Application No. 2008-55752 filed on Mar. 6, 2008, the entire disclosure of which is incorporated herein.

The present invention can be applied to a search result narrowing system and a ranking system in a search engine. In particular, when applied to a search server for a department in a company, the user group's preferences are relatively similar, so that a good recommendation result can be obtained.

Claims (25)

Information index calculation means for calculating an information index that is an index based on a browsing tendency, which is how the information is browsed, for each individual information from the browsing history of information by a plurality of users;
For each user, from the information index calculated for the information browsed by the user, user index calculation means for calculating a user index based on a browsing tendency that is a way of browsing the information of the user;
An information recommendation server device comprising: index matching means for extracting information having an information index similar to the calculated user index for the user from a recommendation target information group. An index storage means for storing an information index for each individual information calculated by the information index calculation means and a user index for each individual user calculated by the user index calculation means. Item 4. The information recommendation server device according to Item 1. An information recommendation request including a user identifier is received from a terminal via a network, and information having an information index similar to the user index calculated by the user index calculation means for the user specified by the user identifier The information recommendation server apparatus according to claim 1, further comprising an information query unit that extracts from the recommendation target information group by the index matching unit and transmits the extraction result to the terminal. Index storage means for recording keywords included in the recommendation target information group;
An information recommendation request including a user identifier and a keyword is received from a terminal via a network, information including the keyword is extracted from the recommendation target information group with reference to the index storage unit, and the use is extracted from the extracted information. An information query for extracting information having an information index similar to the user index calculated by the user index calculating means for the user specified by the user identifier by the index matching means and transmitting the extraction result to the terminal The information recommendation server device according to claim 1, further comprising: means. The information index calculation means calculates a recognition index indicating how many users have been viewed as one of the information indexes, and the user index calculation means is calculated for information viewed by the user. 5. The information recommendation server device according to claim 1, wherein an average value of the recognition indices is calculated as one of the user indices. The information index calculation means calculates a fixing index indicating how continuously browsed as one of the information indices, and the user index calculation means calculates the information calculated by the user 5. The information recommendation server device according to claim 1, wherein an average value of a fixing index is calculated as one of the user indices. The information index calculating means calculates a staying index indicating a browsing time per time as one of the information indices, and the user index calculating means calculates the staying index calculated for the information viewed by the user. The information recommendation server device according to claim 1, wherein an average value is calculated as one of the user indexes. The information index calculation means calculates a time difference index indicating how quickly the information is viewed as one of the information indices, and the user index calculation means calculates the time difference calculated for information viewed by the user. The information recommendation server device according to any one of claims 1 to 4, wherein an average value of an index is calculated as one of the user indices. The information recommendation server device according to any one of claims 1 to 8, and information including an information identifier returned as a response to an information recommendation request including a user identifier transmitted to the information recommendation server device An information recommendation system comprising: a terminal device that receives a recommendation result; and an information providing server device that provides the terminal device with information specified by the information identifier. a) a step of calculating an information index, which is an index based on a viewed tendency, which is how the information is browsed, for each piece of information, from information browsing history of information by a plurality of users;
b) For each user, the user index calculation means calculates a user index based on a browsing tendency, which is a method of browsing the user's information, from the information index calculated for the information browsed by the user. A calculating step;
and c) an index recommendation unit including, from the recommendation target information group, extracting information having an information index similar to the calculated user index for the user. An index storage means for storing an information index for each individual information calculated by the information index calculation means and a user index for each individual user calculated by the user index calculation means. Item 13. The information recommendation method according to Item 10. d) Information query means accepts an information recommendation request including a user identifier from a terminal via a network, and is similar to the user index calculated by the user index calculation means for the user specified by the user identifier The information recommendation method according to claim 10 or 11, further comprising: extracting information having an information index to be extracted from the recommendation target information group by the index matching unit and transmitting the extraction result to the terminal. d) An information query means receives an information recommendation request including a user identifier and a keyword from a terminal via the network, extracts information including the keyword from the recommendation target information group, and extracts the user from the extracted information. Extracting information having an information index similar to the user index calculated by the user index calculation means for the user specified by the identifier by the index matching means, and transmitting the extraction result to the terminal 12. The information recommendation method according to claim 10 or 11, characterized in that: The information index calculation means calculates a recognition index indicating how many users have been viewed as one of the information indexes, and the user index calculation means is calculated for information viewed by the user. The information recommendation method according to claim 10, wherein an average value of the recognition indices is calculated as one of the user indices. The information index calculation means calculates a fixing index indicating how continuously browsed as one of the information indices, and the user index calculation means calculates the information calculated by the user The information recommendation method according to claim 10, wherein an average value of the fixing index is calculated as one of the user indices. The information index calculating means calculates a staying index indicating a browsing time per time as one of the information indices, and the user index calculating means calculates the staying index calculated for the information viewed by the user. 14. The information recommendation method according to claim 10, wherein an average value is calculated as one of the user indexes. The information index calculation means calculates a time difference index indicating how quickly the information is viewed as one of the information indices, and the user index calculation means calculates the time difference calculated for information viewed by the user. The information recommendation method according to claim 10, wherein an average value of an index is calculated as one of the user indices. Computer
Information index calculation means for calculating an information index that is an index based on a browsing tendency, which is how the information is browsed, for each individual information from the browsing history of information by a plurality of users;
For each user, from the information index calculated for the information browsed by the user, user index calculation means for calculating a user index based on a browsing tendency that is a way of browsing the information of the user;
A program for causing the user to function as an index matching unit that extracts information having an information index similar to the calculated user index from a recommendation target information group. The computer comprises index storage means for storing an information index for each individual information calculated by the information index calculation means and a user index for each individual user calculated by the user index calculation means. The program according to claim 18, wherein The computer further
An information recommendation request including a user identifier is received from a terminal via a network, and information having an information index similar to the user index calculated by the user index calculation means for the user specified by the user identifier 20. The program according to claim 18 or 19, wherein the program is made to function as an information query unit that extracts from the recommendation target information group by the index matching unit and transmits the extraction result to the terminal. The computer further
Index storage means for recording keywords included in the recommendation target information group;
An information recommendation request including a user identifier and a keyword is received from a terminal via a network, information including the keyword is extracted from the recommendation target information group with reference to the index storage unit, and the use is extracted from the extracted information. An information query for extracting information having an information index similar to the user index calculated by the user index calculating means for the user specified by the user identifier by the index matching means and transmitting the extraction result to the terminal 20. The program according to claim 18 or 19 for functioning as a means. The information index calculation means calculates a recognition index indicating how many users have been viewed as one of the information indices, and the user index calculation means is calculated for information viewed by the user. The program according to any one of claims 18 to 21, wherein an average value of the recognition indices is calculated as one of the user indices. The information index calculation means calculates a fixing index indicating how continuously browsed as one of the information indices, and the user index calculation means calculates the information calculated by the user The program according to any one of claims 18 to 21, wherein an average value of a fixing index is calculated as one of the user indices. The information index calculation means calculates a staying index indicating a browsing time per time as one of the information indices, and the user index calculation means calculates the staying index calculated for the information viewed by the user. The program according to any one of claims 18 to 21, wherein an average value is calculated as one of the user indexes. The information index calculation means calculates a time difference index indicating how quickly the information is viewed as one of the information indices, and the user index calculation means calculates the time difference calculated for information viewed by the user. The program according to any one of claims 18 to 21, wherein an average value of an index is calculated as one of the user indices.

Images