JP2004362011A - User characteristic extracting system, information recommend system, information recommend method, program and recording medium with its program recorded - Google Patents

Abstract

<P>PROBLEM TO BE SOLVED: To extract personal characteristics for each item name, and to recommend contents based on the personal characteristics for each item name. <P>SOLUTION: As for a preliminarily set item name, the user item categorized intention information of a target user is acquired, and the item value of the number of times of appearance which is not less than a preliminarily set threshold is extracted as for the item name by referring to the acquired user item categorized intention information, and contents information including the item name as the value of the target item name is acquired, and the acquired contents information is recommended to the user. <P>COPYRIGHT: (C)2005,JPO&NCIPI

Description

[0001]
TECHNICAL FIELD OF THE INVENTION
The present invention relates to a user feature extraction system, an information recommendation system, an information recommendation method, a program, and a recording medium storing the program, which extract a feature of a content user and provide information in consideration of the feature. is there.
[0002]
[Prior art]
The information recommendation method using metadata includes the following.
Regarding the "combination" of the target attribute name and attribute value, how many times that "combination" appears in the metadata of all contents, and how long it appears in the metadata of the content watched by the user A method of estimating the user's preference for the "combination" by calculating the ratio and using a mechanism for recommending content (for example, see Non-Patent Document 1). With this method, it is possible to recommend a content including an attribute value (keyword) that the user prefers.
[0003]
[Non-patent document 1]
IPSJ Research Report, DBS-127, pp121-128, 2002
[0004]
[Problems to be solved by the invention]
However, the above prior art has the following problems. The recommendation of the content focuses only on the ratio between the metadata of the content viewed by the individual and the entire metadata, and determines what value the ratio is for the user (how does it compare to others)? Not considering information. Further, since different item names are compared by the same method, it cannot be used when there is a characteristic or the like for each item name. Further, it is an interest extraction in word units, and it is difficult to implement a recommendation such as a search function such as a search axis and a search function suitable for a user.
[0005]
Therefore, an object of the present invention is to (1) extract a feature for each search axis (item name) of an individual, (2) recommend a content based on a feature for each search axis (item name) of an individual, and (3) other contents. Feature extraction for each search axis (item name) taking into account the user's operation distribution tendency, (4) recommendation taking into account the operation distribution tendency of other users, (5) suitable for users To provide a search axis, a recommendation of a search function, a user feature extraction system, an information recommendation system, an information recommendation method, an information program, and a recording medium on which the program is made, which enables the above (1) to (5). is there.
[0006]
[Means for Solving the Problems]
In order to solve the above-mentioned problem, a user feature extraction system according to claim 1 stores and manages user history information storage management means for storing and managing a user's content operation history, and stores and manages content metadata information. A user based on metadata information storage management means, an operation history stored and managed by the user history information storage management means, and metadata information of content stored and managed by the metadata information storage management means; User item-oriented information calculating means for calculating item-specific intention information, and user item-oriented information storage management for storing and managing the user item-oriented information calculated by the user item-oriented information calculating means. And means.
[0007]
According to the user feature extraction system of the first aspect, it is possible to extract a feature in consideration of the user's item-specific intention information.
[0008]
Further, the information recommendation system according to claim 2, wherein the user history information storage management means for storing and managing the content operation history of the user, the metadata information storage management means for storing and managing the metadata information of the content, User item for calculating user-specific orientation information based on the operation history stored and managed by the user history information storage management means and the metadata information of the content stored and managed by the metadata information storage management means Separate intention information calculating means, user item-oriented information storage management means for storing and managing user-specific intention information calculated by the user item-specific intention information calculating means, and user item-specific intention information storage User item intention information stored and managed by the management means, and metadata of the content stored and managed by the metadata information storage and management means Wherein a content recommendation means for implementing the recommendation of the content based on the broadcast, that it consists.
[0009]
According to the information recommendation system of the second aspect, it is possible to recommend information based on user item-oriented information and content metadata information.
[0010]
The information recommendation system according to claim 3 is the information recommendation system according to claim 2, wherein the content recommendation unit is a user item stored and managed by the user item-oriented information storage management unit. Implementing a content recommendation based on intention information, metadata information of the content stored and managed by the metadata information storage management means, and user history information stored and managed by the user history information storage management means It is characterized by doing.
[0011]
According to the information recommendation system of the third aspect, information can be recommended based on user item-oriented information, content metadata information, and user history information.
[0012]
Further, the user feature extraction system according to claim 4 is a user history information storage management unit that stores and manages a content operation history of a user, a metadata information storage management unit that stores and manages metadata information of the content, Based on the operation history stored and managed by the user history information storage management means and the metadata information of the content stored and managed by the metadata information storage management means, item-specific intention information of the user is calculated. User item-specific intention information calculating means; user item-specific intention information storage managing means for storing and managing user item-specific intention information calculated by the user item-specific intention information calculating means; A user item-oriented distribution calculating means for creating a user item-oriented distribution based on user item-oriented information stored and managed by the intention information storing / managing means; Wherein the user itemized oriented distributed storage management means for storing managing calculated at the user item-by-item-oriented distribution calculating means user itemized oriented distribution, to become either.
[0013]
According to the user characteristic extracting system of the fourth aspect, it is possible to extract the characteristics of the user in consideration of the user item-oriented distribution.
[0014]
The information recommendation system according to claim 5, wherein a user history information storage management unit that stores and manages a user's content operation history; a metadata information storage management unit that stores and manages metadata information of the content; A user that calculates item-specific orientation information of a user based on the operation history stored and managed by the user history information storage management means and the metadata information of the content stored and managed by the metadata information storage management means Item-specific intention information calculating means, user item-specific intention information storage management means for storing and managing user-specific item-specific information calculated by the user item-specific intention information calculating means, and user-item-specific intention information A user item-oriented distribution calculating means for creating a user item-oriented distribution based on user item-oriented information stored and managed by the storage managing means; User item-oriented distribution storage management means for storing and managing the user item-oriented distribution calculated by the user item-oriented distribution calculation means, and use stored and managed by the user item-oriented intention information storage management means User item-oriented information, user item-oriented distribution stored and managed by the user item-oriented distribution storage management means, and metadata information stored and managed by the metadata information storage management means. Content recommendation means for implementing content recommendation.
[0015]
According to the information recommendation system described in claim 5, information can be recommended based on user item-oriented information, user item distribution, and content metadata information.
[0016]
The information recommendation system according to claim 6 is the information recommendation system according to claim 5, wherein the content recommendation unit is a user item stored and managed by the user item-oriented information storage management unit. The intention information, the user item-specific distribution stored and managed by the user item-oriented distribution storage management means, the content metadata information stored and managed by the metadata information storage management means, Content recommendation is performed based on the user history information stored and managed by the history information storage management means.
[0017]
According to the information recommendation system of the sixth aspect, information can be recommended based on user item-oriented information, content metadata information, user history information, and user history information.
[0018]
Further, a user feature extraction system according to claim 7, a user history information storage management unit that stores and manages a user's content operation history, a metadata information storage management unit that stores and manages metadata information of content, Based on the operation history stored and managed by the user history information storage management means and the metadata information of the content stored and managed by the metadata information storage management means, item-specific intention information of the user is calculated. User item-specific intention information calculating means; user item-specific intention information storage managing means for storing and managing user item-specific intention information calculated by the user item-specific intention information calculating means; A user item-oriented distribution calculating means for creating a user item-oriented distribution based on user item-oriented information stored and managed by the intention information storing / managing means; User item-oriented distribution storage management means for storing and managing the user item-oriented distribution calculated by the user item-oriented distribution calculation means, and usage stored and managed by the user item-oriented distribution distribution management means Item-by-item characteristic degree calculating means for calculating the user-specific characteristic degree based on the user item-oriented distribution, and item-by-item storing and managing the user item-specific characteristic degree calculated by the item-by-item characteristic degree calculating means. And a characteristic degree storage management means.
[0019]
According to the user feature extraction system according to the seventh aspect, it is possible to extract a user feature in consideration of the item-specific feature level.
[0020]
The information recommendation system according to claim 8, wherein the user history information storage management means for storing and managing the content operation history of the user, the metadata information storage management means for storing and managing the metadata information of the content, A user that calculates item-specific orientation information of a user based on the operation history stored and managed by the user history information storage management means and the metadata information of the content stored and managed by the metadata information storage management means Item-specific intention information calculating means, user item-specific intention information storage management means for storing and managing user-specific item-specific information calculated by the user item-specific intention information calculating means, and user-item-specific intention information A user item-oriented distribution calculating means for creating a user item-oriented distribution based on user item-oriented information stored and managed by the storage managing means; User item-oriented distribution storage management means for storing and managing the user item-oriented distribution calculated by the user item-oriented distribution calculation means, and usage stored and managed by the user item-oriented distribution distribution management means Item-by-item characteristic degree calculating means for calculating the user-specific characteristic degree based on the user item-oriented distribution, and item-by-item storing and managing the user item-specific characteristic degree calculated by the item-by-item characteristic degree calculating means. Characteristic degree storage management means, user item intention information stored and managed by the user item intention information storage management means, and user item characteristics stored and managed by the item characteristic degree storage management means And content recommendation means for performing content recommendation based on the metadata information stored and managed by the metadata information storage management means.
[0021]
According to the information recommendation system of the eighth aspect, information can be recommended based on user item-oriented information, user item-specific characteristics, and content metadata information.
[0022]
The information recommendation system according to claim 9 is the information recommendation system according to claim 8, wherein the content recommendation unit is a user item stored and managed by the user item-oriented information storage management unit. Intention information, user-specific characteristic degrees stored and managed by the item-specific characteristic storage management means, metadata information of contents stored and managed by the metadata information storage management means, and the user history The content is recommended based on the user history information stored and managed by the information storage management means.
[0023]
According to the information recommendation system of the ninth aspect, it is possible to recommend information based on user item-oriented information, user item-specific characteristics, content metadata information, and user history information.
[0024]
The information recommendation system according to claim 10, wherein a user history information storage management unit that stores and manages a user's content operation history; a metadata information storage management unit that stores and manages metadata information of the content; A user that calculates item-specific orientation information of a user based on the operation history stored and managed by the user history information storage management means and the metadata information of the content stored and managed by the metadata information storage management means Item-specific intention information calculating means, user item-specific intention information storage management means for storing and managing user-specific item-specific information calculated by the user item-specific intention information calculating means, and user-item-specific intention information A user item-oriented distribution calculating means for creating a user item-oriented distribution based on user item-oriented information stored and managed by the storage managing means; The user item-oriented distribution storage management means for storing and managing the user item-oriented distribution calculated by the user item-oriented distribution calculation means, and stored and managed by the user item-oriented distribution distribution managing means. Item-by-item characteristic degree calculating means for calculating a user-specific characteristic degree based on a user item-oriented distribution, and an item for storing and managing the user-specific characteristic degree calculated by the item-by-item characteristic degree calculating means. And a search axis recommendation unit for performing a recommendation of a search axis based on a user item characteristic degree stored and managed by the item-specific characteristic degree storage management means. I do.
[0025]
According to the information recommendation system of the tenth aspect, information can be recommended based on the user item-specific characteristic degrees.
[0026]
In the information recommendation method according to the eleventh aspect, the content recommendation unit acquires user item-oriented information of a target user from the user item-oriented information storage management unit with respect to a preset item name. A first step in which the content recommendation unit refers to the acquired user item-specific intention information and extracts an item value having an appearance count equal to or greater than a preset threshold value for the item name; A third step in which the means acquires the content information including the item value as the value of the target item name from the metadata information storage management means, and a fourth step in which the content recommendation means recommends the acquired content information to the user. , Consisting of
[0027]
According to the information recommendation method of the eleventh aspect, it is possible to recommend information satisfying a certain number of appearances.
[0028]
Also, in the information recommendation method according to the twelfth aspect, the content recommending means obtains user item-oriented information on all the item names of the target users from the user item-oriented information storage management means in the first step. And a second step in which the content recommendation means refers to and extracts user item-specific intention information that has acquired an item value having an appearance count equal to or greater than a preset threshold value for an uncalculated item name, A third step in which the means obtains content information including the extracted item value as the value of the target item name from the metadata information storage management means, and checks whether there is any other item name not calculated by the content recommendation means. Judge and return to the second step if there is an uncalculated item name, and go to the fifth step if there is no uncalculated item name. A fourth step of deciding to proceed to the next step, and a fifth step of comparing the content information extracted for each item name with the content recommendation means and preferentially recommending the content extracted with many item names. And step.
[0029]
According to the information recommendation method of the twelfth aspect, it is possible to preferentially recommend information starting from information having a large number of appearances.
[0030]
In the information recommendation method according to the thirteenth aspect, the content recommendation means obtains a user item-oriented distribution of a target user from a user item-oriented distribution storage management means for a preset item name. A first step, wherein the content recommending means extracts a mode value from the acquired user item-oriented distribution, and sets the threshold as a threshold value; and a content recommending means stores the user item-oriented information. A third step of acquiring user item-oriented information of the target item name of the target user from the management means and extracting an item value of the number of appearances equal to or greater than a threshold value; A fourth step of acquiring the content information including the content from the metadata information storage management means, and the content recommendation means A fifth step of recommending Ntsu information to the user, characterized in that it consists.
[0031]
According to the information recommendation method of the thirteenth aspect, it is possible to recommend information satisfying a certain number of appearances.
[0032]
The information recommendation method according to claim 14, wherein the content recommendation means obtains a user item-oriented distribution for all the item names of the target users from the user item-oriented distribution storage means; A second step in which the content recommendation unit extracts the number of appearances that is the final frequency value for each item name from the acquired user item-oriented distribution and sets a threshold value for each item name, and the content recommendation unit calculates A third step of acquiring the user item-oriented information of the target user from the user item-oriented information storage means for the items that have not been set, and the content recommendation means presetting the items that have not been calculated. A fourth step of extracting item values having the number of appearances equal to or greater than the threshold value by referring to the user's item-oriented information; Means for obtaining from the metadata information storage management means content information including the item value as the value of the target item name, and determining whether there is any other item name not calculated by the content recommendation means. If there is an item name that has not been calculated, the process returns to the third step. If there is no item name that has not been calculated, the sixth step of deciding to proceed to the seventh step. A seventh step of comparing the content information extracted for each name and preferentially recommending the content information extracted with a large number of item names.
[0033]
According to the information recommendation method of the fourteenth aspect, it is possible to preferentially make a recommendation from information having a large number of appearances.
[0034]
Also, in the information recommendation method according to the fifteenth aspect, the content recommending means obtains an item-specific characteristic of the target user from the item-specific characteristic storage managing means, and the content recommending means obtains the characteristic recommendation. A second step of extracting an item name as an axis of recommendation from the value of the item-specific feature degree, and the content recommending means for the user of the user targeted by the user item-oriented distribution storage management means for the extracted item name A third step of obtaining an item-oriented distribution, a fourth step in which the content recommendation unit extracts a mode value from the acquired user item-oriented distribution and sets a threshold value, and the content recommendation unit includes: Obtains user item-oriented information for the target item name of the target user from the user item-oriented information management means, and displays the number of occurrences equal to or greater than the threshold A fifth step of extracting the item value of the item information, a sixth step of the content recommending means acquiring the content information including the item value as the value of the target item name from the metadata information storage management means, And a seventh step of recommending the content information to the user.
[0035]
According to the information recommendation method according to the fifteenth aspect, information satisfying a certain number of appearances can be recommended.
[0036]
The information recommendation method according to claim 16, wherein the content recommendation means obtains a user item-oriented distribution for all of the target users from the user item-oriented distribution storage means, A second step in which the content recommendation unit extracts the number of appearances that is the most frequent value for each item name from the acquired user item-oriented distribution and sets a threshold value for each item name, and the content recommendation unit does not calculate A third step of acquiring user item-oriented information of the target user from the user item-oriented information storage means for the item, and a threshold value which is not set by the content recommendation means for an item not calculated in advance. A fourth step of extracting an item value having the number of appearances of the user by referring to the item-specific intention information of the user; A fifth step of acquiring content information including the item value as the value of the target item name from the metadata information storage management unit, and the content recommendation unit determines whether there is any other item name that has not been calculated. If there is an item name that has not been calculated, the process returns to the third step. If there is no other item name that has not been calculated, a sixth step of deciding to proceed to the seventh step; A seventh step of obtaining the item-specific feature level of the target user from the different-feature-level storage management means, and the content recommendation means comparing the content information extracted for each item name, and obtaining a large item-level feature level. An eighth step of preferentially recommending the content extracted by the item name.
[0037]
According to the information recommendation method of the sixteenth aspect, it is possible to preferentially recommend the information having a large number of appearances.
[0038]
In the information recommendation method according to the seventeenth aspect, the first step in which the search axis recommendation unit acquires the characteristic degree of the target user from the item-specific characteristic degree storage management unit, and the search axis recommendation unit acquires the characteristic degree And a second step of recommending a search function with a search axis having a large characteristic degree among the characteristic degrees.
[0039]
According to the information recommendation method of the seventeenth aspect, it is possible to recommend a search axis and a search function suitable for a user.
[0040]
A program according to claim 18 describes that the user feature extraction system, the information recommendation system, or the information recommendation method according to any one of claims 1 to 17 is described as a computer program and can be executed. Features.
[0041]
According to the program described in claim 18, a computer can realize the user feature extraction system, the information recommendation system, or the information recommendation method according to any one of claims 1 to 17.
[0042]
According to a nineteenth aspect of the present invention, there is provided a recording medium which records a program in which the user feature extraction system, the information recommendation system, or the information recommendation method according to any one of the first to seventeenth aspects is executable by a computer. Features.
[0043]
According to the recording medium of the nineteenth aspect, it is possible to realize the user feature extraction system, the information recommendation system, or the information recommendation method according to the first to seventeenth aspects by a computer.
[0044]
BEST MODE FOR CARRYING OUT THE INVENTION
Hereinafter, embodiments of the present invention will be described with reference to the drawings.
[0045]
(1st Embodiment)
FIG. 1 shows a system configuration of the first embodiment. The user history information storage management unit is realized as the user history information storage management unit 101. The metadata information storage management means is realized as the metadata information storage management unit 102. The user item-specific intention information calculation means is realized as the user item-specific intention information calculation unit 103. The user item-specific intention information storage management means is realized as the user item-specific intention information storage management unit 104. The content recommendation unit is realized as the content recommendation unit 107.
[0046]
The user history information storage management unit 101 stores and manages content operation history information for each user. The content operation history may be only the viewing history, only the search history, or a history combining the viewing history and the search history. The storage management format of the user history information storage management unit 101 is not particularly limited, and may be, for example, a general relational database. In the present embodiment, a case where the data is stored in the relational database shown in FIG. 2 will be described as an example.
[0047]
The metadata information storage management unit 102 stores and manages metadata information expressing the content of the content. The expression format of the metadata information is not particularly limited, and may be an XML format or a simple list of “item names” and “item values”. Similarly, the storage management format of the metadata information storage management unit 102 is not particularly limited, and may be a general relational database, a native XML database, or a general file system. is there. In the present embodiment, a case will be described as an example where the metadata information shown in FIG. 3 is in XML format.
[0048]
The user item-specific intention information calculation unit 103 acquires the operation history of the target user and the metadata information related thereto from the user history information storage management unit 101 and the metadata information storage management unit 102. Then, user item-specific intention information is calculated, and information of the user item-specific intention information storage management unit 104 is updated. FIG. 4 shows a flowchart of calculating the user-specific intention information.
[0049]
In FIG. 4, calculation of user item-specific information is started in S41. In S42, it is determined whether the target user has a new history. If there is a new history, the process proceeds to S43, and if not, the process proceeds to S47. In S43, one program ID of a new user history is extracted from the user history information storage management unit 101. In S44, the metadata information of the corresponding program ID is acquired from the metadata information storage management unit 102. In S45, the number of appearances of the target record is counted up with reference to the user item-specific intention information table for each item name. If the target record does not exist, a new record is created and the number of appearances is set to 1. In S46, it is determined whether or not there is a new history for the target user. If there is a new history, the process returns to S43. If not, the process proceeds to S47 to end the calculation of the user item-specific information. I do.
[0050]
The user item-specific intention information storage management unit 104 stores and manages the user item-specific intention information calculated by the user item-specific intention information calculation unit 103. The storage management format of the user item-specific intention information storage management unit 104 is not particularly limited, and may be, for example, a general relational database. In the present embodiment, a case where the data is stored in the relational database shown in FIG. 5 will be described as an example.
[0051]
In the content recommendation unit 107, user item-oriented information stored and managed in the user item-oriented information storage unit 104, metadata information stored and managed in the metadata information storage management unit 102, and user history The content is recommended based on the user history information stored and managed by the information storage management unit 101. There are various methods for recommending contents, and as an example, the following method is conceivable.
[0052]
First, with respect to the target user, user item-specific intention information of a specific item name set in advance is acquired from the user item-specific intention information storage management unit 104. The user item-oriented information acquired here can be considered as an item value indicating the user's interest more as the number of appearances increases. Next, as shown in FIG. 6, an item value having an occurrence count equal to or greater than a preset threshold value of the item name is extracted. Finally, it is conceivable that the content information including the item value as the value of the target item name is acquired from the metadata information storage management unit 102 and is recommended. At this time, by using the user history information, the content previously operated (viewed or the like) by the user can be excluded from the recommendation target. FIG. 7 shows a flowchart of the content recommendation in this example.
[0053]
In FIG. 7, first, content recommendation is started in S71. In S72, the user-item-oriented information of the target user is acquired from the user-item-oriented information storage management unit 104 for the preset item name. In S73, the obtained item value of the number of appearances equal to or greater than a preset threshold value for the item name is extracted with reference to the acquired user item-oriented information. In S74, the content information including the item value as the value of the target item name is acquired from the metadata information storage management unit 102. In S75, the acquired content information is recommended to the user, and in S76, the content recommendation ends.
[0054]
As another example, the following method can be considered. First, for the target user, user item-specific intention information on all item names is acquired from the user item-specific intention information storage management unit 104. Next, for each item name, an item value that is equal to or greater than the threshold value of the number of appearances set in advance for each item name is extracted, and the content including the item value as the value of the target item name is extracted from the metadata information storage management unit 102. get. Finally, the contents extracted for each item name are compared, and the content hit by many item names is recommended preferentially. At this time, by using the user history information, the content previously operated (viewed or the like) by the user can be excluded from the recommendation target. FIG. 8 shows a recommendation flowchart for content in this example.
[0055]
In FIG. 8, first, content recommendation is started in S81. In S82, user item-specific intention information on all item names of the target user is acquired from the user item-specific intention information storage management unit 104. In S83, with respect to the uncalculated item name, an item value having an appearance frequency equal to or greater than a preset threshold value is extracted with reference to the acquired user item-specific intention information. In S84, content information including the extracted item value as the value of the target item name is acquired from the metadata information storage management unit 102. In S85, it is determined whether there is another item name that has not been calculated. If there is an item name that has not been calculated, the process returns to S83, and if not, the process proceeds to S86. In S86, the content information extracted for each item name is compared, and the content extracted with many item names is preferentially recommended. In S87, the content recommendation ends.
[0056]
The system according to the first embodiment is roughly divided into “procedure for creating user item-oriented information, which is a characteristic level of a user, from user history information and metadata information”, and “procedure for performing content recommendation”. ]. The following briefly describes the implementation procedure for each.
[0057]
Regarding “Procedure for creating user item-specific intention information, which is a characteristic level of a user, from user history information and metadata information”, (1) user operation history is stored in user history information storage management unit 101 Have been. In addition, metadata information representing the content of the content is stored in the metadata information storage management unit 302. (2) The user item-specific intention information calculation unit 103 compares the user history information stored in the user history information storage management unit 101 with the metadata information stored in the metadata information storage management unit 102. Based on this, user item-specific intention information is calculated and stored in the user item-specific intention information storage management unit 104.
[0058]
Regarding the “procedure for performing content recommendation”, the content recommendation unit 107 stores and manages user item-oriented information stored and managed in the user item-oriented information storage unit 104 and the metadata information storage and management unit 102. The content is recommended based on the stored metadata information and the user history information stored and managed in the user history information storage management unit 101.
[0059]
As a modification of the first embodiment, a user history information storage management unit 101, a metadata information storage management unit 102, a user item-oriented information calculation unit 103, and a user item-oriented information shown in FIG. A system including a storage management unit 104, or a user history information storage management unit 101, a metadata information storage management unit 102, a user item-specific intention information calculation unit 103, and a user item-specific intention information storage management unit 104 shown in FIG. , And a content recommendation section 107. The content recommendation section 107 has a system for performing content recommendation based on user item-oriented information and metadata information.
[0060]
(2nd Embodiment)
FIG. 11 shows a system configuration of the second embodiment. The user history information storage management unit is realized as the user history information storage management unit 201. The metadata information storage management means is realized as the metadata information storage management unit 202. The user item-specific intention information calculation means is realized as the user item-specific intention information calculation unit 203. The user item-specific intention information storage management unit is realized as the user item-specific intention information storage management unit 204. The user item-specific intention distribution calculation means is realized as the user item-specific intention distribution calculation unit 205. The user item-oriented distribution storage management means is realized as the user item-oriented distribution storage management unit 206. The content recommendation unit is realized as the content recommendation unit 207. In the second embodiment, only portions different from the first embodiment will be described below.
[0061]
The user item-oriented distribution calculation unit 205 receives the user item-oriented information from the user item-oriented information storage management unit 204, calculates the user item-oriented distribution, and stores the user item-oriented orientation distribution storage / management unit. Pass the information to 206. FIG. 12 shows a flowchart of the user item-specific intention distribution calculation.
[0062]
In FIG. 12, first, in S121, the calculation of the intention distribution for each user item is started. At S122, attention is paid to user item-specific intention information. In S123, it is determined whether or not there is a change in the item-specific intention information of the target user. If there is a change, the process proceeds to S124, and if not, the process proceeds to S129. In S124, attention is paid to the changed item name. In S125, a graph (user ID = X: item name = Y) is created in which the abscissa represents “number of appearances” and the ordinate represents “item numerical value”. In S126, the user value of the user ID = X for each occurrence count is totaled with reference to the user item-by-user information table of the item name X, and plotted on a graph. In S127, it is determined whether there is another item name to be changed. If there is an item name to be changed, the process returns to S124. If not, the process proceeds to S128, and the calculation of the user-item-oriented distribution is terminated.
[0063]
The user item-oriented distribution storage management unit 206 stores and manages the user item-oriented distribution calculated by the user item-oriented distribution calculation unit 205. The storage management format of the user item-oriented distribution storage management unit 206 is not particularly limited, and may be, for example, a general relational database. FIGS. 13 and 14 show examples of storage formats and distribution diagrams of the user item-oriented distribution.
[0064]
In the content recommendation unit 207, user item-oriented information stored and managed in the user item-oriented information storage unit 204, and user item-oriented information stored and managed in the user item-oriented distribution storage management unit 206. The content is recommended based on the distribution, the metadata information stored and managed in the metadata borrowing information storage management unit 202, and the user history information stored and managed in the user history information storage management unit 201. . There are various methods for recommending content. As an example, the following method can be considered.
[0065]
First, for a target user, user item-oriented information and a user item-oriented distribution for a specific item name set in advance are stored in a user item-specific intention information storage management unit 204 and a user item, respectively. It is obtained from the separate-oriented distribution storage management unit 206. As in the first embodiment, the user item-oriented information and the item-oriented distribution acquired here can be considered as item values indicating the user's interest more as the number of appearances increases. Next, as shown in FIG. 15, a value (for example, mode) based on a statistical processing method is extracted from the acquired user item-oriented distribution, and the value is used as a threshold. Obtained from user item-oriented information. Finally, it is conceivable to obtain the content including the item value as the value of the target item name from the metadata information storage management unit 202 and make a recommendation. At this time, by using the user history information, the content previously operated (viewed or the like) by the user can be excluded from the recommendation target. FIG. 16 shows a flowchart of the content recommendation in this example.
[0066]
In FIG. 16, first, content recommendation is started in S161. In step S162, the user item-oriented distribution of the target user is acquired from the user item-oriented distribution storage management unit 206 for the preset item name. In S163, the mode value, which is the mode value, is extracted from the acquired user item-oriented distribution, and is set as a threshold value. In S164, the user item-oriented information of the target item name of the target user is acquired from the user item-oriented information storage management unit 204, and the item value of the number of appearances equal to or greater than the threshold is extracted. In S165, the content information including the item value as the value of the target item name is acquired from the metadata information storage management unit 202. In S166, the obtained content information is recommended to the user, and the process advances to S167 to end the content recommendation.
[0067]
As another example, the following method can be considered. First, for the target user, the user item-oriented information and the user item-oriented distribution for all item names are stored in the user item-oriented information storage manager 204 and the user item-oriented distribution storage manager 206, respectively. And get from. Next, for each item name, a value (for example, mode) based on a statistical processing method is extracted for each item from the user item-oriented distribution, and the value is used as a threshold. The item value is extracted, and the content including the item value as the value of the target item name is acquired from the metadata information storage management unit 202. It is also conceivable to compare the content groups extracted for each item name and preferentially recommend content hit by many item names. At this time, by using the user history information, the content previously operated (viewed or the like) by the user can be excluded from the recommendation target. FIG. 17 shows a flowchart of the content recommendation in this example.
[0068]
In FIG. 17, first, content recommendation is started in S171. In S <b> 172, the user item-oriented distribution for all the item names of the target user is acquired from the user item-oriented distribution storage unit 206. In S173, the number of appearances that is the final frequency value for each item name is extracted from the acquired user item-oriented distribution, and is set as a threshold value for each item name. In S 174, the user-information-oriented information of the target user is acquired from the user-item-oriented information storage unit 204 for the item that has not been calculated. In S175, with respect to the items that have not been calculated, the item values having the number of appearances equal to or greater than the preset threshold value are extracted with reference to the item-specific intention information of the user. In S176, the content information including the item value as the value of the target item name is acquired from the metadata information storage management unit 202. In S177, it is determined whether there is another item name that has not been calculated. If there is an item name that has not been calculated, the process returns to S174; otherwise, the process proceeds to S178. In S178, the content information extracted for each item name is compared, and the content information extracted with many item names is preferentially recommended. In S179, the content recommendation ends.
[0069]
The system according to the second embodiment is roughly divided into “procedure for creating user item-oriented information and user item-oriented distribution information from user history information and metadata information” and “procedure for executing content recommendation. ]. The following briefly describes the implementation procedure for each.
[0070]
Regarding “Procedure for creating user item-oriented information and user item-oriented distribution information from user history information and metadata information”, (1) user operation history is stored in user history information storage management unit 201 Has been accumulated. Also, metadata information representing the content of the content is stored in the metadata information storage management unit 202. (2) The user item-specific intention information calculation unit 203 compares the user history information stored in the user history information storage management unit 201 with the metadata information stored in the metadata information storage management unit 202. Based on this, user item-specific intention information is calculated and stored in the user item-specific intention information storage management unit 204. (3) The user item-oriented distribution calculation unit 205 acquires user item-oriented information from the user item-oriented information storage management unit 204, creates a user item-oriented distribution, and creates a user item-oriented orientation. The information is stored in the distribution storage unit 206.
[0071]
Regarding the “procedure for executing a content recommendation”, the content recommendation unit 207 stores the user-oriented information stored in the daily-oriented information storage unit 204 and the user-item-oriented distribution storage management unit 206. , The metadata distribution stored and managed in the metadata information storage management unit 202, and the user history stored and managed in the user history information storage management unit 201. Recommend content based on information.
[0072]
As a modification of the second embodiment, a user history information storage management unit 201, a metadata information storage management unit 202, a user item-specific intention information calculation unit 203, and a user item-specific intention information storage shown in FIG. A system including a management unit 204, a user item-oriented distribution calculation unit 205, and a user item-oriented distribution storage management unit 206, or a user history information storage administration unit 201 and a metadata information storage administration unit 202 shown in FIG. A user item-specific intention information calculation unit 203, a user item-specific intention information storage management unit 204, a user item-specific intention distribution calculation unit 205, a user item-specific intention distribution storage management unit 206, and a content recommendation unit 207. The content recommendation unit 207 performs content recommendation based on user item-oriented information, user item-oriented distribution, and metadata information. There is a system to carry out the command.
[0073]
(Third embodiment)
FIG. 20 shows a system configuration according to the third embodiment. The user history information storage management unit is realized as the user history information storage management unit 301. The metadata information storage management means is realized as the metadata information storage management unit 302. The user item-specific intention information calculation means is realized as a user item-specific intention information calculation unit 303. The user item-specific intention information storage management means is realized as a user item-specific intention information storage management unit 304. The user item-oriented distribution calculation means is realized as a user item-oriented distribution calculation unit 305. The user item-oriented distribution storage management means is realized as the user item-oriented distribution storage management unit 306. The item-by-item characteristic degree calculation unit is realized as the item-by-item characteristic degree calculation unit 307. The item-by-item characteristic degree storage management unit is realized as the item-by-item characteristic degree storage management unit 308. The content recommendation unit is realized as the content recommendation unit 309. In the third embodiment, only portions different from the second embodiment will be described below.
[0074]
The item-specific characteristic degree calculation unit 307 calculates the item-specific characteristic degree for each user based on the user item-oriented distribution stored and managed in the user item-oriented distribution storage management unit 306. FIG. 21 shows a flowchart of the item-specific feature level calculation.
[0075]
In FIG. 21, first, in S211 the calculation of the characteristic degree is started. In S212, it is determined whether or not the item-oriented distribution of the target user is changed. If there is a change, the process proceeds to S213, and if not, the process proceeds to S216. In S213, the distribution status of the user item-oriented distribution of the target user and the item-oriented distribution of the segments classified from a certain viewpoint or the entire user are compared by a certain statistical process. In S214, it is determined whether there is any other item name that has not been counted. If so, the process returns to S213; otherwise, the process proceeds to S215. In S215, the degree of bias for each item name obtained in the above flow is considered as the user's characteristic degree vector, and is stored in the item-specific characteristic degree storage management unit 308. In S216, the characteristic degree calculation ends.
[0076]
In addition, the expression for calculating the degree of characteristic used in this flowchart is based on statistical processing, and uses various methods such as “average”, “median value”, and “mode value” to calculate the difference between the calculated values. It can be used as it is, or a normalized value can be used. Here, an example of the relationship between the characteristic degree calculated based on the statistical processing and the degree of interest in the item name of the user is shown.
[0077]
For example, when an average value is used as the statistical processing method, if the average value of the user-oriented distribution of a certain user for a certain item name is significantly larger than the average value for the entire item name, This means that the item value of the item name of the content operated by the user overlaps more than the average of the user, and it can be predicted that the user is particular about the item. Therefore, it is considered that the recommendation based on the item name is effective. FIG. 22 shows an example of the relationship between the feature level of each item of the user (the relationship between the user and the average in the user item-oriented distribution) and the interest of the user. In FIG. 22, in pattern 1, the item values in this item overlap more than the user average. In other words, this item has particular attention, and it is effective to present information centering on this item. Further, in the pattern 2, the user average and the overlap of the item values in this item are the same. In other words, the commitment to this item is the average level. Further, in pattern 3, the overlap of the item values in this item is smaller than the user average. In other words, attention to this item is below the average level, and presentation of information based on this item is not effective.
[0078]
The item-by-item characteristic degree storage management unit 308 stores and manages the item-by-item characteristic degree calculated by the item-by-item characteristic degree calculation unit 307. The storage management format of the item-specific feature storage buried section 308 is not particularly limited, and may be, for example, a general relational database. In the third embodiment, a case where the data is stored in the relational database shown in FIG. 23 will be described as an example.
[0079]
In the content recommendation unit 309, user item-oriented information stored and managed in the user item-oriented information storage unit 304, item-specific feature degrees stored and managed in the item-specific feature storage management unit 308, and meta information The content is recommended based on the metadata information stored and managed in the data information storage management unit 302 and the user history information stored and managed in the user history information storage management unit 301. There are various methods for recommending content. As an example, in addition to the content recommendation shown in FIG. 16 of the second embodiment, it is conceivable to select a target item name for calculating the content to be recommended based on the value of the characteristic degree. FIG. 24 shows a flowchart of the content recommendation in this example.
[0080]
In FIG. 24, first, content recommendation is started in S241. In S <b> 242, the itemized characteristic degree of the target user is acquired from the itemized characteristic degree storage management unit 308. In S243, an item name as a recommendation axis is extracted from the acquired values of the item-specific feature degrees. In step S244, the user item-oriented distribution of the target user is acquired from the user item-oriented distribution storage management unit 306 for the extracted item name. In S245, the appearance frequency value that becomes the mode is extracted from the acquired user item-oriented distribution, and is set as a threshold value. In S246, the user-item-oriented information of the target item name of the target user is acquired from the user-item-oriented information management unit 304, and the item value of the occurrence count equal to or larger than the threshold is extracted. In S247, the content information including the item value as the value of the target item name is acquired from the metadata information storage management unit 302. In S248, the obtained content information is recommended to the user, and in S249, the content recommendation ends.
[0081]
As another example, as an improvement of the content recommendation shown in FIG. 17 of the second embodiment, instead of simply recommending content that hits with many item names, a characteristic degree is used as a coefficient for each item name. A method is also conceivable in which content that hits many item names with a large value of the degree of characteristic is preferentially recommended. FIG. 25 shows a flowchart of the content recommendation in this example.
[0082]
In FIG. 25, first, content recommendation is started in S251. In S <b> 252, the user item-oriented distribution of all the target users is obtained from the user item-oriented distribution storage unit 306. In S253, the number of appearances, which is the mode value for each item name, is extracted from the acquired user item-oriented distribution, and is set as a threshold value for each item name. In S 254, regarding the items that have not been calculated, the user item-oriented information of the target user is acquired from the user item-oriented information storage unit 304. In S255, with respect to the items that have not been calculated, item values having the number of appearances equal to or greater than a preset threshold value are extracted with reference to the item-oriented information of the user. In S256, the content information including the item value as the value of the target item name is acquired from the metadata information storage management unit 302. In S257, it is determined whether there is another item name that has not been calculated. If there is, the process returns to S254, and if not, the process proceeds to S258. In S <b> 258, the itemized characteristic degree of the target user is acquired from the itemized characteristic degree storage management unit 308. In S259, the content information extracted for each item name is compared, and the content extracted by the item name having the large item-by-item characteristic is recommended with higher priority. In S260, the content recommendation ends.
[0083]
The system described in the third embodiment is roughly divided into “procedure for creating user item-oriented information and user item-oriented distribution information from user history information and metadata information”, and “content recommendation”. Procedures to be implemented ”. The following briefly describes the implementation procedure for each.
[0084]
Regarding “Procedure for creating user item-oriented information and user item-oriented distribution information from user history information and metadata information”, (1) User operation history is stored in user history information storage section 301 Has been accumulated. Also, one piece of metadata information representing the content of the content is stored in the metadata information storage management unit 302. (2) The user item-specific intention information calculation unit 303 compares the user history information stored in the user history information storage management unit 301 with the metadata information stored in the metadata information storage management unit 302. Based on this, user item-specific intention information is calculated and stored in the user item-specific intention information storage management unit 304. (3) The user item-oriented distribution calculation unit 305 acquires the user item-oriented information from the user item-oriented information storage management unit 304, creates a user item-oriented distribution, and creates a user item-oriented. It is stored in the distribution storage unit 306. (4) The item-specific characteristic degree calculation unit 307 acquires the user item-specific intention distribution from the user item-specific intention distribution storage management unit 306, calculates the item-specific characteristic degree, and sets the item-specific characteristic degree storage management unit 308. Store.
[0085]
Regarding the “procedure for executing a content recommendation”, the content recommendation unit 309 stores and manages the user item-oriented information stored in the user item-oriented information storage unit 304 and the item-specific feature degree storage management unit 308. Item-based characteristic, metadata information stored and managed in the metadata information storage management unit 302, and user history information stored and managed in the user history loan information storage management unit 301. Conduct content recommendations.
[0086]
As a modification of the third embodiment, a user history information storage management unit 301, a metadata information storage management unit 302, a user item-specific intention information calculation unit 303, and a user item-specific intention information storage shown in FIG. FIG. 27 shows a system including a management unit 304, a user item-oriented distribution calculation unit 305, a user item-oriented distribution storage management unit 306, an item-based characteristic degree calculation unit 307, and an item-based characteristic degree storage management unit 308. User history information storage management unit 301, metadata information storage management unit 302, user item-oriented information calculation unit 303, user item-oriented information storage management unit 304, user item-oriented distribution calculation unit 305, usage User-oriented distribution storage management section 306, item-specific characteristic degree calculation section 307, item-specific characteristic degree storage management section 308, and content recommendation section 309. Tsu recommendation unit 309 is a system for implementing a user itemized oriented information, and user itemized features of the content recommendation based on the metadata information.
[0087]
(Fourth embodiment)
FIG. 28 shows a system configuration of the fourth embodiment. The user history information storage management means is realized as the user history loan information storage management unit 401. The metadata information storage management means is realized as the metadata information storage management unit 402. The user item-specific intention information calculation means is realized as the user item-specific intention information calculation unit 403. The user item-specific intention information storage management unit is realized as a user item-specific intention information storage management unit 404. The user item-oriented distribution calculation means is realized as the user item-oriented distribution calculation unit 405. The user item-oriented distribution storage management means is realized as the user item-oriented distribution storage management unit 406. The item-by-item characteristic degree calculation unit is realized as the item-by-item characteristic degree calculation unit 407. The item-by-item characteristic degree storage management unit is realized as the item-by-item characteristic degree storage management unit 408. The search axis recommendation unit is realized as the search axis recommendation unit 409. In the fourth embodiment, only portions different from the third embodiment will be described below.
[0088]
The search axis recommendation unit 409 recommends a search axis and a search function based on the item-by-item characteristic degree stored and managed in the item-by-item characteristic degree storage management unit 408. The recommendation of search axis and search function is to recommend a search axis specialized for items such as director search, actor search, and age search, and is based on axes such as recommended director content of this week and recommended actor content of this week. You can also consider the recommendation that was made. FIG. 29 shows a flowchart of the search axis recommendation.
[0089]
In FIG. 29, first, the search axis recommendation is started in S291. In step S292, the feature level of the target user is acquired from the item-specific feature level storage management unit 408. In S293, the search axis and the search function with the large characteristic degree are recommended, and the search axis recommendation is ended in S294.
[0090]
The system described in the fourth embodiment is roughly divided into “procedure for creating user item-oriented information and user item-oriented distribution information from user history information and metadata information”, and “search axis recommendation”. Procedure for implementing the process ”. Here, only portions different from the third embodiment will be described below.
[0091]
Regarding the “procedure for executing a search axis recommendation”, the search axis recommendation unit 409 recommends the search axis based on the item-specific feature degrees stored and managed in the item-specific feature degree storage management unit 408.
[0092]
In the present invention, a part or all of the processing functions of the system shown in FIGS. 1, 9, 10, 11, 18, 19, 20, 26, 27, and 28 are configured as a program and realized using a computer. In addition, the processing procedures shown in FIGS. 4, 7, 8, 12, 16, 17, 21, 24, 25, and 29 can be configured as a program and executed by a computer. Also, a computer-readable recording medium such as a flexible disk, an MO, a ROM, and a memory card may be used to execute a program for realizing the processing function of each unit in the computer or a program for causing the computer to execute the processing procedure. , A CD, a DVD, a removable disk, or the like, and can be stored or provided, and can be distributed via a communication network such as the Internet.
[0093]
【The invention's effect】
According to the present invention, in recommendation using metadata, (1) extraction of a feature for each individual search axis (item name), and (2) content recommendation based on a feature for each individual search axis (item name). (3) feature extraction for each individual search axis (item name) taking into account the operation distribution tendency of other users; (4) recommendation taking into account the operation distribution tendency of other users; 5) Search axis suitable for user, recommendation of search function, user feature extraction system, information recommendation system, information recommendation method, information recommendation method, program, and recording medium on which the program can be performed (1) to (5) Can be provided.
[Brief description of the drawings]
FIG. 1 is a system configuration diagram of a first embodiment.
FIG. 2 is a user history information table.
FIG. 3 is an example of metadata information.
FIG. 4 is a flowchart example of user item-specific intention information calculation.
FIG. 5 is a user item-specific intention information table.
FIG. 6 is an example of extracting recommendation item values.
FIG. 7 is a flowchart example of content recommendation.
FIG. 8 is a flowchart example of content recommendation.
FIG. 9 is a system configuration diagram of the first embodiment.
FIG. 10 is a system configuration diagram of the first embodiment.
FIG. 11 is a system configuration diagram of a second embodiment.
FIG. 12 is a flowchart of calculating a user item-oriented distribution.
FIG. 13 is a user item-oriented distribution table.
FIG. 14 is an intentional distribution by user item.
FIG. 15 is an example of item value extraction in content recommendation.
FIG. 16 is a flowchart example of a content recommendation.
FIG. 17 is a flowchart example of a content recommendation.
FIG. 18 is a system configuration diagram of the second embodiment.
FIG. 19 is a system configuration diagram of the second embodiment.
FIG. 20 is a system configuration diagram of a third embodiment.
FIG. 21 is a flowchart of characteristic degree calculation.
FIG. 22 is an example of a relationship between a feature level and a user interest level.
FIG. 23 is an item-by-item characteristic degree table.
FIG. 24 is a flowchart example of content recommendation.
FIG. 25 is a flowchart example of content recommendation.
FIG. 26 is a system configuration diagram of the third embodiment.
FIG. 27 is a system configuration diagram of the third embodiment.
FIG. 28 is a system configuration diagram of a fourth embodiment.
FIG. 29 is a flowchart of a search axis recommendation.
[Explanation of symbols]
101: User history information storage management unit
102: Metadata information storage management unit
103 ... user item-specific intention information calculation unit
104: User item-oriented information storage management unit
107: Content recommendation section
201: User history information storage management unit
202: Metadata information storage management unit
203: User item-oriented information calculation unit
204: intention information storage management unit by user item
205: User item-oriented distribution calculation unit
206: User item-oriented distribution storage management unit
207 ... Content recommendation section
301: User history information storage management unit
302: Metadata information storage management unit
303: User item-specific intention information calculation unit
304: User item oriented information storage management unit
305: User item-oriented distribution calculation unit
306: User item-oriented distribution storage management unit
307: item-specific feature degree calculation unit
308... Item-specific feature storage management unit
309 ... Content recommendation section
401: User history information storage management unit
402: Metadata information storage management unit
403: User item-specific intention information calculation unit
404: User item-oriented information storage management unit
405: User item-oriented distribution calculation unit
406: User item-oriented distribution storage management unit
407: item-specific feature degree calculation unit
408: Item-specific feature storage management unit
409: Search axis recommendation part

Claims (19)

User history information storage management means for storing and managing the content operation history of the user;
Metadata information storage management means for storing and managing content metadata information; operation history stored and managed by the user history information storage management means; and content management information stored and managed by the metadata information storage management means. User item-oriented information calculating means for calculating user-oriented information based on metadata information, and user item-oriented information calculated by the user item-oriented information calculating means are stored and managed. And a user item-specific intention information storage management means. User history information storage management means for storing and managing the content operation history of the user;
Metadata information storage management means for storing and managing content metadata information; operation history stored and managed by the user history information storage management means; and content management information stored and managed by the metadata information storage management means. User item-oriented information calculating means for calculating user-oriented information based on the metadata information;
User item-specific intention information storage management means for storing and managing the user item-specific intention information calculated by the user item-specific intention information calculation means;
Implementing a content recommendation based on user item-specific intention information stored and managed by the user item-specific intention information storage management unit and content metadata information stored and managed by the metadata information storage management unit An information recommendation system comprising: In the information recommendation system according to claim 2,
The content recommendation means, user item-oriented information stored and managed by the user item-oriented information storage management means, and metadata information of the content stored and managed by the metadata information storage management means, An information recommendation system for recommending content based on user history information stored and managed by the user history information storage management means. User history information storage management means for storing and managing the content operation history of the user;
Metadata information storage management means for storing and managing content metadata information; operation history stored and managed by the user history information storage management means; and content management information stored and managed by the metadata information storage management means. User item-oriented information calculating means for calculating user-oriented information based on metadata information, and user item-oriented information calculated by the user item-oriented information calculating means are stored and managed. User item-oriented information storage management means
User item-oriented distribution calculating means for creating a user item-oriented distribution based on user item-oriented information stored and managed by the user item-oriented information storage managing means;
A user item-oriented distribution storage managing means for storing and managing the user item-oriented distribution calculated by the user item-oriented distribution calculating means; User history information storage management means for storing and managing the content operation history of the user;
Metadata information storage management means for storing and managing content metadata information; operation history stored and managed by the user history information storage management means; and content management information stored and managed by the metadata information storage management means. User item-oriented information calculating means for calculating user-oriented information based on metadata information, and user item-oriented information calculated by the user item-oriented information calculating means are stored and managed. User item-oriented information storage management means
User item-oriented distribution calculating means for creating a user item-oriented distribution based on user item-oriented information stored and managed by the user item-oriented information storage managing means;
User item-oriented distribution storage managing means for storing and managing the user item-oriented distribution calculated by the user item-oriented distribution calculating means;
User item intention information stored and managed by the user item intention information storage management means, user item distribution stored and managed by the user item intention distribution storage management means, and the metadata An information recommendation system, comprising: content recommendation means for performing content recommendation based on metadata information stored and managed by the information storage management means. The information recommendation system according to claim 5,
The content recommendation means comprises user item-oriented information stored and managed by the user item-oriented information storage management means, and user item-oriented information stored and managed by the user item-oriented distribution storage management means. Implementing a content recommendation based on the distribution, the metadata information of the content stored and managed by the metadata information storage management means, and the user history information stored and managed by the user history information storage management means An information recommendation system characterized by: User history information storage management means for storing and managing the content operation history of the user;
Metadata information storage management means for storing and managing content metadata information; operation history stored and managed by the user history information storage management means; and content management information stored and managed by the metadata information storage management means. User item-oriented information calculating means for calculating user-oriented information based on metadata information, and user item-oriented information calculated by the user item-oriented information calculating means are stored and managed. User item-oriented information storage management means
User item-oriented distribution calculating means for creating a user item-oriented distribution based on user item-oriented information stored and managed by the user item-oriented information storage managing means;
User item-oriented distribution storage managing means for storing and managing the user item-oriented distribution calculated by the user item-oriented distribution calculating means;
An item-by-item characteristic degree calculating unit that calculates an item-by-item characteristic degree of a user based on the user item-by-user-oriented distribution stored and managed by the user item-by-user-oriented distribution storage managing unit;
A user-specific feature level storage management means for storing and managing the user-specific feature levels calculated by the item-specific feature level calculation means. User history information storage management means for storing and managing the content operation history of the user;
Metadata information storage management means for storing and managing content metadata information; operation history stored and managed by the user history information storage management means; and content management information stored and managed by the metadata information storage management means. User item-oriented information calculating means for calculating user-oriented information based on metadata information, and user item-oriented information calculated by the user item-oriented information calculating means are stored and managed. User item-oriented information storage management means
User item-oriented distribution calculating means for creating a user item-oriented distribution based on user item-oriented information stored and managed by the user item-oriented information storage managing means;
User item-oriented distribution storage managing means for storing and managing the user item-oriented distribution calculated by the user item-oriented distribution calculating means;
An item-by-item characteristic degree calculating unit that calculates an item-by-item characteristic degree of a user based on the user item-by-user-oriented distribution stored and managed by the user item-by-user-oriented distribution storage managing unit;
Item-by-item characteristic degree storage management means for storing and managing the user-specific characteristic degrees calculated by the item-by-item characteristic degree calculation means;
User item-oriented information stored and managed by the user-item-oriented information storage management means; user-item-specific characteristic degrees stored and managed by the item-specific characteristic degree storage management means; and the metadata information An information recommendation system, comprising: content recommendation means for performing content recommendation based on metadata information stored and managed by the storage management means. In the information recommendation system according to claim 8,
The content recommendation means includes user item intention information stored and managed by the user item intention information storage management means, and user item characteristic degrees stored and managed by the item characteristic degree storage management means. And recommending the content based on the metadata information of the content stored and managed by the metadata information storage management means and the user history information stored and managed by the user history information storage management means. An information recommendation system characterized by the following. User history information storage management means for storing and managing the content operation history of the user;
Metadata information storage management means for storing and managing content metadata information; operation history stored and managed by the user history information storage management means; and content management information stored and managed by the metadata information storage management means. User item-oriented information calculating means for calculating user-oriented information based on metadata information, and user item-oriented information calculated by the user item-oriented information calculating means are stored and managed. User item-oriented information storage management means
User item-oriented distribution calculating means for creating a user item-oriented distribution based on user item-oriented information stored and managed by the user item-oriented information storage managing means;
User item-oriented distribution storage managing means for storing and managing the user item-oriented distribution calculated by the user item-oriented distribution calculating means;
An item-by-item characteristic degree calculating unit that calculates an item-by-item characteristic degree of a user based on the user item-by-user-oriented distribution stored and managed by the user item-by-user-oriented distribution storage managing unit;
Item-by-item characteristic degree storage management means for storing and managing the user-specific characteristic degrees calculated by the item-by-item characteristic degree calculation means;
An information recommendation system comprising: search axis recommendation means for performing a recommendation of a search axis based on a user item characteristic degree stored and managed by the item-specific characteristic degree storage management means. A first step in which the content recommendation means obtains user item-oriented information of a target user from the user item-oriented information storage management means with respect to a preset item name;
A second step in which the content recommendation means refers to the acquired user item-specific intention information and extracts an item value having an appearance count equal to or greater than a preset threshold value for the item name;
A third step in which the content recommendation unit acquires content information including the item value as a value of the target item name from the metadata information storage management unit;
An information recommendation method, comprising: a fourth step in which the content recommendation means recommends the acquired content information to the user. A first step in which the content recommendation means obtains user item-oriented information on all item names of the target user from the user item-oriented information storage management means;
A second step in which the content recommendation means refers to and extracts user item-specific intention information that has acquired an item value having an appearance count greater than or equal to a preset threshold value for an uncalculated item name,
A third step in which the content recommendation unit acquires from the metadata information storage management unit content information including the extracted item value as a value of the target item name;
The content recommendation unit determines whether there is any other item name that has not been calculated. If there is any item name that has not been calculated, the process returns to the second step. A fourth step of deciding to proceed to a step;
A content recommendation means for comparing content information extracted for each item name, and recommending content extracted with many item names preferentially, a fifth step of preferentially recommending the content information. . A first step in which the content recommendation unit obtains a user item-oriented distribution of a target user from the user item-oriented distribution storage management unit with respect to a preset item name;
A second step in which the content recommendation means extracts a mode value, which is a mode value, from the acquired user item-oriented distribution and sets the value as a threshold value;
A third step in which the content recommendation means obtains user item-specific intention information of the target item name of the target user from the user item-specific intention information storage management means, and extracts an item value having an appearance count equal to or greater than a threshold value;
A fourth step in which the content recommendation unit acquires content information including the item value as a value of the target item name from the metadata information storage management unit;
An information recommendation method, comprising: a fifth step in which the content recommendation means recommends the acquired content information to the user. A first step in which the content recommendation means obtains a user item-oriented distribution for all item names of the target user from the user item-oriented distribution storage means;
A second step in which the content recommendation means extracts the number of appearances that is the final frequency value for each item name from the acquired user item-oriented distribution, and sets a threshold value for each item name;
A third step in which the content recommendation unit obtains user item-oriented information of the target user from the user item-oriented information storage unit for items that have not been calculated;
A fourth step in which the content recommendation means extracts an item value having an appearance count equal to or greater than a preset threshold value for an uncalculated item by referring to the item-specific intention information of the user;
A fifth step in which the content recommendation unit acquires content information including the item value as a value of the target item name from the metadata information storage management unit;
The content recommendation unit determines whether there is another item name that has not been calculated. If there is an item name that has not been calculated, the process returns to the third step. A sixth step of deciding to proceed to a step;
A seventh step of comparing the content information extracted for each item name by the content recommendation means and preferentially recommending the content information extracted by many item names. . A first step in which the content recommendation means obtains the item-specific characteristic of the target user from the item-specific characteristic storage management means;
A second step in which the content recommendation means extracts an item name as an axis of the recommendation from the acquired value of the characteristic degree for each item;
A third step in which the content recommendation means obtains the user item-oriented distribution of the target user from the user item-oriented distribution storage management means with respect to the extracted item name;
A fourth step in which the content recommendation means extracts a mode value, which is the mode, from the acquired user item-oriented distribution, and sets the frequency as a threshold value;
A fifth step in which the content recommendation means obtains user item-oriented information of the target item name of the target user from the user item-specific intention information management means, and extracts an item value having an appearance count equal to or greater than a threshold value;
A sixth step in which the content recommending means obtains content information including the item value as a value of the target item name from the metadata information storage management means;
An information recommendation method, comprising: a seventh step in which the content recommendation means recommends the acquired content information to the user. A first step in which the content recommendation means obtains a user item-oriented distribution for all of the target users' item names from the user item-oriented distribution storage means;
A second step in which the content recommendation means extracts the number of appearances, which is the most frequent value for each item name, from the acquired user item-oriented distribution and sets it as a threshold value of each item name;
A third step in which the content recommendation unit obtains user item-oriented information of the target user from the user item-oriented information storage unit for items that have not been calculated;
A fourth step in which the content recommendation means extracts an item value having an appearance count equal to or greater than a preset threshold value for an uncalculated item by referring to the item-specific intention information of the user;
A fifth step in which the content recommendation unit acquires content information including the item value as a value of the target item name from the metadata information storage management unit;
The content recommendation unit determines whether there is any other item name that has not been calculated. If there is another item name that has not been calculated, the process returns to the third step. If there is no other item name that has not been calculated, A sixth step of deciding to proceed to a seventh step, and a seventh step in which the content recommendation means obtains the item-specific characteristic of the target user from the item-specific characteristic storage management means,
An eighth step in which the content recommendation means compares the content information extracted for each item name, and preferentially recommends the content extracted by the item name having a large item-by-item characteristic degree. Information recommendation method to be. A first step in which the search axis recommendation means obtains the characteristic degree of the target user from the item-specific characteristic degree storage management means;
An information recommendation method, characterized in that the search axis recommendation means comprises: a search axis having a large feature level among the acquired feature levels; and a second step of recommending a search function. 18. A program, wherein the user feature extraction system, the information recommendation system, or the information recommendation method according to any one of claims 1 to 17 is described in a computer program so as to be executable. 18. A recording medium on which is recorded a program in which the user characteristic extraction system, the information recommendation system, or the information recommendation method according to any one of claims 1 to 17 is executable by a computer.

Images