JP2007058562A - Content classification device, content classification method, content classification program and recording medium - Google Patents

Abstract

<P>PROBLEM TO BE SOLVED: To classify content into groups without imposing any labor to classify content on a user. <P>SOLUTION: When meta information showing the contents of content is at least partially matched with the other meta information showing the contents of the other content which is different from the content, a content relevancy judging part 12 judges that the content and the other content are relevant to each other. A content classifying part 13 classifies the content and the other content whose relevancy is judged by the content relevancy judging means into the same group. Therefore, it is possible to classify the content into groups without imposing any labor to classify the content on a user. <P>COPYRIGHT: (C)2007,JPO&INPIT

Description

  The present invention relates to a content classification device, a content classification method, a content classification program, and a recording medium that can automatically classify content into groups without giving users the trouble of classifying content.

  Televisions, hard disk recorders, and personal computers are known as devices that can receive broadcast programs. These apparatuses record the program by storing the received program in a predetermined recording medium such as a hard disk. In the previous device, when recording a program, the contents of each program were simply arranged in the root directory sequentially as a predetermined file. Therefore, the user has not been able to quickly search for desired content from among many contents.

  There is Patent Document 1 as a document disclosing a technique for solving this problem. This document includes an acquisition unit that acquires provided information, a selection unit that selects information to be recorded from the information acquired by the acquisition unit, and a first unit that records information selected by the selection unit. An information processing apparatus comprising: one recording unit; a determination unit that determines a classification of the information selected by the selection unit; and a second recording unit that records the classification determined by the determination unit Is disclosed.

  In the apparatus of Patent Document 1, program information is acquired together with a broadcast program at the time of recording reservation. A program attribute is extracted from the acquired program information, and a user selects a predetermined character string from the program attribute. A directory is created using the selected character string as a directory name, and a program is recorded under the directory. Thereby, the program is classified. Further, when the directory of the selected character string already exists, the program is recorded under the existing directory.

A similar technique is also disclosed in Patent Document 2.
JP 2000-285638 A (released on October 13, 2000) JP 2000-285647 A (released on October 13, 2000)

However, the techniques disclosed in Patent Documents 1 and 2 require time and effort for the user to select a character string as a directory name for each program. As a character string that can be a directory name, there are many options such as a genre, a day of the week, a title, a performer, and a keyword among program attributes. Therefore, every time a user reserves a program, the user has to select from among them.

  In addition, the user must consider the directory structure at the reservation stage so as to obtain an appropriate classification situation. Appropriate classification status refers to a situation where content is moderately dispersed. That is, there is no situation where content is biased to a certain directory, or the number of directories is close to the number of content and does not make sense for the directory. Also, when searching for a program, the directory name must be such that the contents of the directory can be easily guessed. A directory structure that is appropriate while content is small can become inappropriate as content increases or decreases. This is because the content may be biased in one directory. Therefore, the user needs to optimize the classification configuration with a heavy burden in order to make it easy to search for programs.

  Furthermore, it is easier to search for certain specific contents stored in the root directory. For example, content that is scheduled to be erased immediately, content that has just been recorded and that has a new date, and content that has a small number of contents in the same directory even if a directory is created. Therefore, if all contents are classified into the same directory, the user must select the directory once at the time of search, which is not convenient.

  When based on program information such as a title, the content may not be correctly classified. This is because the title may be omitted, or there may be fluctuations in the number of broadcasts and subtitles. If the selected character string exists in an existing directory, a new directory is not created and is recorded under the existing directory. However, conversely, if it does not exist, a directory of the newly selected character string is created.

  For example, assume that a title is selected in the program attributes and a directory whose directory name is the title is created. Here, it is assumed that the user wants to record the same program in the same directory. Next, when reserving the same program on another day, the user selects a title from the program attributes in the same way at the time of reservation. At this time, the titles of the previous reservation and the current reservation may be the same notation. However, the program information data may have fluctuations in the notation. Therefore, in this case, the correspondence with the existing directory cannot be taken, and another new directory is created. Therefore, the program may not be classified according to the user's expectation.

  The present invention has been made to solve the above-described problems, and its purpose is to provide a content classification device, a content classification method, and a content classification method capable of automatically classifying content into groups without giving users the trouble of classifying content. To provide a content classification program and a recording medium.

In order to solve the above problems, the content classification device according to the present invention provides:
It is determined that the content and the other content are related to each other when the additional information representing the content and the other additional information representing the content of other content different from the content are at least partially matched. Content relevance determining means for
Content classification means for classifying the content determined to be related to each other by the content relevance determination means and the other content into the same group is provided.

  According to the above configuration, in the content classification device, the content relevance determination unit compares the additional information indicating the content details with other additional information indicating the details of other content. The “content” here is, for example, a predetermined program being broadcast. The “additional information” is, for example, an electronic program guide that represents the contents of a program.

  The content classification device can acquire content that includes additional information therein, can acquire additional information added to the content, or can acquire content and additional information separately. it can. In any case, the content classification device compares the additional information acquired by some means with other additional information. At this time, if the additional information and the other additional information match at least partially, the content relevance determining unit determines that the content and the other content are related to each other.

  In the content classification device, the content classification unit further classifies the content determined to be related to each other by the content relevance determination unit and the other content into the same group. For example, when it is determined that the information related to the same performer is included in each electronic program guide, the content classification unit classifies the two programs into groups related to the performer.

  In this way, the content classification device automatically classifies a plurality of contents determined to be related to each other into a common group. Therefore, the content classification device has an effect of automatically classifying content into groups without giving time and effort to classify content to users.

In order to solve the above problems, the content classification method according to the present invention provides:
It is determined that the content and the other content are related to each other when the additional information representing the content and the other additional information representing the content of other content different from the content are at least partially matched. A content relevance determining step,
When it is determined that the content and the other content are related to each other, a content classification step of classifying the content and the other content into the same group is included.

  According to said structure, there exists an effect similar to the content classification device based on this invention.

In the content classification device according to the present invention,
The additional information includes a keyword representing the content content,
The other additional information includes other keywords representing the contents of the other content,
The content relevance determining means is
When it is determined that the keyword and the other keyword match at least partially, it is preferable to determine that the content and the other content are related to each other.

  According to said structure, the additional information showing the content content contains the predetermined keyword. This keyword represents the content. The additional information includes text representing the title of the content, for example, as a keyword. Furthermore, other additional information representing the contents of other contents also includes other predetermined keywords. The other keywords represent the contents of other contents. Other additional information includes text representing the genre of other content as another keyword, for example.

  Here, the content relevance determining unit determines that the content and the other content are related to each other when the keyword and the other keyword are at least partially matched. As a result, the content classification means classifies the content in which the keywords match each other and other content into a common group.

  Therefore, it is possible to determine whether the content and other content are related by determining only the keyword without determining the entire additional information. As described above, the content classification device has an effect of further increasing the processing speed when the content is classified into groups.

In the content classification device according to the present invention,
The content relevance determining means is
When the distance between the keyword and the other keyword is smaller than a predetermined threshold, it is preferable to determine that the content and the other content are related to each other.

  According to said structure, a content relevance determination means calculates the distance between two keywords. At this time, for example, this distance is calculated by DP matching. The content relevance determining unit determines that the content and the other content are related to each other when the calculated distance is smaller than a predetermined threshold. Thereby, the content classification means classifies a plurality of contents whose distance between keywords is smaller than a threshold value into a common group. Therefore, the content classification device has an effect of classifying content into groups based on a more objective index.

In the content classification device according to the present invention,
The name of the same group is preferably a character string common to the keyword and the other keyword.

  According to said structure, a content classification | category means classify | categorizes a content and another content into the group by which the character string common to a keyword and another keyword is set to the name. Thereby, there exists an effect which can classify | categorize a content into the group with a close relationship with a content.

In the content classification device according to the present invention,
The category to which the keyword belongs is preferably the same as the category to which the other keyword belongs.

  According to the above configuration, the content relevance determining unit determines that the content and the other content are related to each other when the keyword and the other keyword belonging to the same category at least partially match each other. . As a result, the content classification means can classify content matching keywords belonging to the same category into a common group. Therefore, the content classification device has an effect that the accuracy when the content is classified into groups can be further increased.

In the content classification device according to the present invention,
The content relevance determination means further includes a keyword belonging to another category different from the category included in the additional information and another keyword belonging to the other category included in the other additional information. When at least partially matching, it is preferable to determine that the content and the other content are related to each other.

  According to the above configuration, regarding two contents matching by a keyword belonging to a certain category, it is further determined whether or not to match again by a keyword belonging to another category different from this category. That is, when two types of keywords belonging to different categories match each other, the content relevance determination unit determines that the two contents are related to each other. Thereby, the content classification means classifies these two contents into a common group. That is, the contents to be classified are narrowed down.

  In this way, the content classification device has an effect of preventing excessive content from being classified into a specific group.

In the content classification device according to the present invention,
The keyword is preferably a keyword for which a higher priority is set.

  According to said structure, a content relevance determination means uses the keyword by which the higher priority was set for the content matching. A high priority means that it is preferably used for content matching in preference to other keywords. Thereby, the content classification device has an effect that the content can be classified into a more appropriate group into which the content should be classified.

In the content classification device according to the present invention,
Keyword selection means for accepting selection of the keyword by the user;
It is preferable to further include priority setting means for setting a higher priority for the keyword selected by the user.

  According to said structure, a priority setting means sets a higher priority to the keyword which the user selected than the keyword which the user did not select. Here, the keyword selected by the user is a keyword that is highly likely to be liked by the user. As a result, the content relevance determining unit determines that the two contents that match the keywords that are likely to be liked by the user are related to each other. Therefore, the content classification device has an effect that the classification configuration for classifying content into groups can be made closer to what the user wants.

In the content classification device according to the present invention,
A group name selection means for receiving the name of the group selected by the user;
It is preferable to further include priority setting means for setting a higher priority for the keyword including the name of the group received by the group name selection means.

  According to said structure, a priority setting means sets a higher priority to the keyword containing the name of the group which the user selected. Here, the name of the group selected by the user is likely to be preferred by the user. As a result, the content relevance determination unit classifies the two contents matched by the keyword including the character string that the user is likely to like into the same group. Therefore, the content classification device has an effect that the classification configuration for classifying content into groups can be made closer to what the user wants.

In the content classification device according to the present invention,
User information input means for receiving input of information about the user;
Preferably, the apparatus further includes priority setting means for setting a higher priority to a keyword including a character string included in the information related to the user received by the user information input means.

  According to said structure, a priority setting means sets a higher priority to the keyword containing the character string contained in the information regarding a user. Therefore, the content classification means classifies two contents that match with keywords that are closely related to information about the user into the same group.

  Therefore, the content classification device can make the classification configuration for classifying content into groups closer to what the user wants. Further, such processing can be realized without inputting the keyword or group name that the user likes.

In the content classification device according to the present invention,
The content classification means
It is preferable to set a keyword having a higher priority and a character string common to other keywords to the name of the group located in a higher hierarchy.

  According to said structure, the character string which is common to the keyword in which a higher priority is set, and another keyword is set to the name of the group which classifies content and other content. Further, this group is positioned in a higher hierarchy. Therefore, since the group more closely related to the user is positioned higher, it is possible to achieve a group configuration in which the user can easily search for content.

In the content classification device according to the present invention,
When a hierarchical group exists in advance,
It is preferable to further include priority setting means for setting a higher priority for a keyword belonging to the same category as the category to which the keyword including the name of the group located in a higher hierarchy belongs.

  According to the above configuration, the priority setting means, based on the hierarchical structure of the existing group, assigns the keyword belonging to the same category to the category to which the keyword including the name of the group located higher in this hierarchy belongs Set a higher priority. Therefore, the content classification device can resemble the hierarchical structure of groups that classify content and other content with the hierarchical structure of existing groups. Thereby, since a sense of unity occurs in the hierarchical structure of the group, there is an effect that it is possible to make a group configuration that makes it easy to search for contents classified into groups.

In the content classification device according to the present invention,
It is preferable to further include group deletion means for deleting the group designated by the user.

  According to said structure, the content classification device can delete the group which a user feels unnecessary. Thereby, there exists an effect which can optimize the structure of the group currently managed.

In the content classification device according to the present invention,
A group deleting means for deleting the group instructed by the user;
It is preferable to include priority setting means for setting the priority of the keyword including the name of the group deleted by the group deleting means to a lower level.

  According to said structure, the content classification device can delete the group which a user feels unnecessary. Therefore, first, the configuration of the managed group can be optimized. In the content classification device, the priority setting unit further sets the priority of the keyword including the name of the deleted group to a lower level.

  Here, the group that the user requests to delete is a group that the user feels unnecessary. That is, the name of this group is likely not particularly important to the user. At this time, the content classification apparatus sets a lower priority for keywords that the user does not feel particularly important. Therefore, it becomes difficult for the content classification means to use this keyword when determining whether the content and other content are related to each other.

  Thereby, the content classification device has an effect of preventing the configuration when the content is classified into groups from being undesired by the user.

In the content classification device according to the present invention,
It is preferable to further include display means for displaying the result of classification by the content classification means on a predetermined display.

  According to the above configuration, the display unit displays the result of the content classification unit classifying the content into the group on the predetermined display. For example, a new classification configuration of content is displayed on the display. Thus, the user can easily grasp the content classification status managed by the content classification device.

A preference learning unit for learning the keyword selected by the user;
It is preferable to further include content recommendation means for recommending content to the user based on the keyword learned by the preference learning unit.

  According to said structure, a preference learning part learns the keyword which the user selected as a keyword which the user likes, for example. At this time, the learned keyword is stored in, for example, a predetermined preference information database. Accordingly, the content recommendation unit recommends the content represented by the keyword to the user based on the keyword learned by the preference learning unit. For example, the content is recommended to the user by displaying information on the content on a predetermined display.

  In this way, the content classification device has an effect of recommending to the user content that is likely to be liked by the user simply by selecting the keyword used for content matching.

In the content classification device according to the present invention,
Content adding means for adding new content to the group based on user instructions;
When the new content is added to the group by the content adding means, the maximum of the distance between the keyword representing the content of the new content and the keyword representing the content of the content included in the group in advance Keyword distance measuring means for calculating the distance of
When the maximum distance is larger than a predetermined threshold, it is preferable to further include threshold setting means for setting the maximum distance as a new threshold.

  According to the above configuration, in the content classification device, the content adding unit adds the content specified by the user to the group. Further, the keyword distance measuring means calculates the maximum distance among the distances between the keyword representing the content of the new content and the keyword representing the content of the content previously included in the group into which the content is classified. Next, when the maximum distance calculated by the keyword distance measuring means is larger than a threshold value predetermined for the group, this maximum distance is set as a new threshold value for the group.

  It is highly likely that the user instructs the content classification apparatus to add a certain content to the group because the range of the content classified into the group is desired to be wider. Therefore, the threshold value setting means changes a predetermined threshold value to a larger value so that more contents can be classified into this group. As described above, the content classification apparatus has an effect that the range of content newly added to the group can be automatically expanded only by adding the content requested by the user to the group.

In the content classification device according to the present invention,
Content deletion means for removing the content instructed by the user from the group;
When the content is removed from the group by the content deletion means, a keyword representing the content of the first content arbitrarily selected from the group and a first content different from the first content arbitrarily selected from the group A keyword distance measuring means for calculating the maximum distance among the distances to other keywords representing the contents of the two contents;
When the maximum distance is smaller than the predetermined threshold, it is preferable to further include threshold setting means for setting the maximum distance as a new threshold.

  According to the above configuration, in the content classification device, the content deleting unit removes the content designated by the user from the group. Further, the keyword distance measuring means includes a keyword representing the content of the first content arbitrarily selected from the group after the content is removed and another keyword representing the content of the second content also arbitrarily selected. Of the distances, the maximum distance is calculated. Next, when the calculated maximum distance is smaller than a predetermined threshold, the threshold setting unit sets the maximum distance as a new threshold.

  It is highly likely that the user instructs the content classification apparatus to remove a certain content from the group because the range of the content to be classified into the group is desired to be narrower. Therefore, the threshold value setting means adjusts the predetermined threshold value to a smaller value so that the number of contents classified next in this group becomes smaller. As described above, the content classification device has an effect of automatically narrowing the range of content newly added to the group only by removing the content requested by the user from the group.

In the content classification device according to the present invention,
A content adding means for adding new content designated by the user to the group;
The content classification means
When the new content is added to the group by the content adding means, a character string common to the keyword representing the content of each content included in the group is set as a new name of the group. Is preferred.

  According to said structure, a content addition means adds the content which the user instruct | indicated to the group. Thereby, the content classification means changes the name of the group to which new content is added. Specifically, the content classification unit extracts a character string common to each keyword representing the content of each content classified into a group, and sets this character string as the name of the group. Thereby, the name set to the group reflects the keyword of the new content added to the group.

  In this way, the content classification device has an effect that a more appropriate name that reflects the content newly added to the group can be set for the group.

In the content classification device according to the present invention,
A content deleting means for removing the content instructed by the user from the group;
The content classification means
When the content is removed from the group by the content deletion unit, it is preferable to newly set a character string common to the keyword representing the content of each content newly included in the group as the name of the group .

  According to said structure, a content deletion means removes the content which the user instruct | indicated from the group. Thereby, the content classification means changes the name of the group. Specifically, the content classification unit extracts a character string common to each keyword representing the content of each content remaining in the group, and sets this character string as the name of the group. Therefore, the name set for the group reflects the content remaining in the group, that is, does not reflect the content removed in the group.

  As described above, the content classification apparatus has an effect that a more appropriate name can be newly set for a group reflecting content remaining in the group after a certain content is removed.

In the content classification device according to the present invention,
Program receiving means for receiving a predetermined program as the content;
It is preferable to further include program information acquisition means for acquiring program information representing the contents of the program as the additional information.

  According to the above configuration, the content classification device has an effect of automatically classifying a received program into a group without giving the user the trouble of classifying the program into a group.

In the content classification device according to the present invention,
The additional information includes a predetermined keyword representing the content of the program,
The other additional information includes a predetermined other keyword representing the content of the other program,
The content relevance determining means is
When it is determined that the keyword and the other keyword are at least partially matched, it is preferable to determine that the program and the other program are related to each other. The program can be classified into a more appropriate group.

In the content classification device according to the present invention,
Program information storage means for storing the program information representing the program requested by the user for at least one of recording and viewing on a predetermined recording medium;
It is preferable to further include priority setting means for setting a higher priority for the keyword including the character string included in the program information stored in the recording medium.

  According to said structure, a program information preservation | save means memorize | stores in the predetermined recording medium the program information which the program which the user requested | required at least any one of recording and viewing. That is, the program information storage means stores the history of the program designated by the user in the recording medium. As a result, the priority setting means sets a higher priority for the keyword including the character string included in the program information stored in the recording medium. That is, the priority setting means sets a higher priority to the keyword representing the program to be viewed than the program recorded by the user. Therefore, the content classifying unit classifies programs into groups based on keyword matching results representing programs that the user is usually familiar with.

Thereby, in the content classification device, the program stored in the recording medium has an effect of automatically classifying the program into a group according to the user's preference without the user inputting information regarding the program preference. A preference learning unit for learning the keyword including the character string included in the information;
It is preferable to further include program recommendation means for recommending a program to the user based on the keyword learned by the preference learning unit.

  According to said structure, a preference learning part learns a keyword based on the log | history of the program designated by the user. At this time, the learned keyword is stored in, for example, a predetermined preference information database. Thereby, the program recommendation means recommends a program represented by the keyword to the user based on the keyword learned by the preference learning unit. For example, a program is recommended to the user by displaying information about the program on a predetermined display.

  As described above, the content classification device has an effect of automatically recommending to the user a program that is likely to be liked by the user only by viewing or recording the program.

Keyword selection means for accepting selection of the keyword by the user;
A preference learning unit for learning the keyword selected by the user;
It is preferable to further include program recommendation means for recommending a program to the user based on the keyword learned by the preference learning unit.
x According to said structure, a preference learning part learns the keyword which the user selected as a keyword which the user likes, for example. At this time, the learned keyword is stored in, for example, a predetermined preference information database. Thereby, the program recommendation means recommends the program represented by the keyword to the user based on the keyword learned by the preference learning unit. For example, the program is recommended to the user by displaying information about the program on a predetermined display.

  As described above, the content classification device has an effect of automatically recommending to the user a program that is likely to be liked by the user only by selecting the keyword used for program matching.

  The content classification apparatus may be realized by a computer. In this case, a content classification program for realizing the content classification device in a computer by operating the computer as each of the above means and a computer-readable recording medium recording the content classification program are also included in the scope of the present invention.

  As described above, since the content classification device according to the present invention includes content classification means for classifying content determined to be related to each other and other content into the same group, the content classification device classifies the content. There is an effect that contents can be automatically classified into groups without giving users trouble.

An embodiment of the present invention will be described below with reference to FIGS.
(Content classification device 1)
First, the configuration of the content classification device 1 according to the present embodiment will be described below with reference to FIG. FIG. 1 is a block diagram showing details of a content classification apparatus 1 according to an embodiment of the present invention. According to this figure, the content classification device 1 includes an input unit 2, a display 4, a content database 5, a classification database 6, and a content management unit 10.
(Input unit 2)
The input unit 2 receives input of predetermined content. This content is, for example, a web page or an electronic book. The input unit 2 further receives input of meta information (additional information) representing the contents. The input unit 2 outputs the input content and meta information to the content management unit 10.
(Display unit 3 and display 4)
The display unit 3 displays predetermined information related to content classification on the display 4. For example, the current content classification configuration is displayed on the display 4. In addition, a list of keywords to be selected by the user when the contents are classified into groups is displayed.
(Content database 5)
The content database 5 is a database that stores content input to the content classification device 1. The database further stores meta information representing the contents. The content database 5 is formed in a predetermined memory (not shown) in the content classification device 1.
(Classification database 6)
The classification database 6 is a database that stores display data that defines which group the content is classified into. Details of the classification data will be described later. The classification database 6 is formed in a predetermined memory (not shown) in the content classification apparatus 1.
(Content Management Unit 10)
The content management unit 10 manages content input through the input unit 2 in the content database 5 and the classification database 6. The content management unit 10 includes a keyword extraction unit 11, a content relevance determination unit 12 (content relevance determination unit), a content classification unit 13 (content classification unit), and a priority setting unit 14 (priority) as submodules. A setting unit), a group creation unit 15, a group deletion unit 16 (group deletion unit), and a preference learning unit 17 (preference learning unit).

Next, each of these members will be described.
(Keyword extraction unit 11)
The keyword extraction unit 11 extracts a predetermined keyword from meta information representing the content content. The keyword extraction unit 11 can extract, for example, a keyword represented by predetermined text data from meta information representing newly input content. Further, a keyword representing the content content is extracted from the meta information representing the content classified into the existing group. At this time, the keyword extraction unit 11 extracts keywords from the meta information stored in the content database 5.

The keyword extraction unit 11 outputs the extracted keyword to the content relevance determination unit 12.
(Content relevance determination unit 12)
The content relevance determination unit 12 determines whether the content and other content different from the content are related to each other. Specifically, the content relevance determination unit 12 determines whether or not the meta information representing the content and the other meta information representing other content different from the content match at least partially. To do. When determining that they match, the content relevance determination unit 12 determines that the content and other content are related to each other.

  The content relevance determination unit 12 further associates the content with the other content even when the keyword extracted from the meta information and the other keyword extracted from the other meta information at least partially match each other. It is determined that That is, the content relevance determination unit 12 determines whether two content items are related by matching keywords representing the content content.

The content relevance determination unit 12 classifies the determination result in the content classification unit 13.
(Content classification unit 13)
The content classification unit 13 classifies content into groups based on the determination result by the content relevance determination unit 12. For example, content newly input to the content classification device 1 is classified into a created group. In addition, a plurality of newly input contents can be classified into a newly created group.
(Priority setting unit 14)
The priority setting unit 14 sets a priority for the keyword. Details of this will be described later.
(Group creation unit 15)
The group creation unit 15 creates a group for classifying content. The group creation unit 15 can set a character string common to two content keywords as the name of the group. Details of this will be described later.
(Group deletion unit 16)
The group deletion unit 16 deletes an existing group designated by the user. That is, the information regarding the group designated by the user is deleted from the classification data stored in the classification database 6.
(Preference learning section 17)
The preference learning unit 17 learns the keyword selected by the user. At this time, the preference learning unit 17 stores information on the learned keywords in a preference information database (not shown). Details will be described later.

In the content classification device 1, the keyword extraction unit 11, the content relevance determination unit 12, the content classification unit 13, the group creation unit 15, and the group deletion unit 16 are combined into each other as a single member. May be implemented. In that sense, for example, the content classification unit 13 may have a function as the group creation unit 15.
(Computer system configuration)
The content classification device 1 according to the present embodiment is substantially realized by computer hardware, a program executed by the computer hardware, and data stored in the computer hardware.

  A computer system 200 that implements the content classification device 1 will be described below with reference to FIG. FIG. 2 is a diagram illustrating a computer system 200 that implements the content classification apparatus 1.

  According to this figure, the computer system 200 includes a computer 220 having an optical disk drive 201 and a magnetic disk drive 202, a monitor 203, and a remote control 204.

  The computer 220 includes a remote control interface (I / F) 205 that receives a signal from the remote control 204 in addition to the optical disc drive 201 and the magnetic disc drive 202. Further, a CPU (Central Processing Unit) 206 is provided. Further, a bus 207 connected to the remote control interface 205, the CPU 206, the optical disk drive 201, and the magnetic disk drive 202 is provided. Furthermore, a read-only memory (ROM) 208 is provided which is connected to the bus 207 and stores a bootup program and the like. Further, a random access memory (RAM) 209 is also connected to the bus 207 and stores program instructions, system programs, work data, and the like.

  Although not shown in FIG. 2, the computer 220 may further include a network adapter board that provides a connection to a local area network (LAN).

  Both a program for causing the computer system 200 to realize the functions of the information processing apparatus and input information are stored in the optical disc 210 inserted into the optical disc drive 201. Alternatively, it is stored in the magnetic disk 211 inserted into the magnetic disk drive 202. The content classification program stored in this recording medium is transferred to the hard disk 212. Alternatively, the content classification program may be transmitted to the computer 220 through a network (not shown) and stored in the hard disk 212. The content classification program is executed by being loaded into the RAM 209. Note that the program may be loaded directly into the RAM 209 from the optical disk 210, from the magnetic disk 211, or via a network.

  This content classification program includes a plurality of instructions for causing the computer 220 to realize the functions of the content classification apparatus 1 according to the present embodiment. Some of the basic functions necessary to realize these functions are provided by operating system (OS) or third-party programs running on the computer 220 or modules of various toolkits installed on the computer 220. . Therefore, this content classification program does not necessarily include all the functions necessary for realizing the functions of the content classification device 1 according to the present embodiment. That is, the content classification program includes only an instruction for executing control of the information processing apparatus described above by calling an appropriate function or “tool” in a controlled manner so as to obtain a desired result. It only has to be. The operation of computer system 200 is well known and will not be repeated here.

Note that recording media storing programs for realizing the functions of the information processing apparatus are CD-ROM (Compact Disc Read Only Memory), MO (Magneto-Optical disc), MD (Mini Disc), and DVD ( The present invention is not limited to the optical disk 210 such as Digital Versatile Disc), and the magnetic disk 211 such as FD (flexible disk) and hard disk. Tapes such as magnetic tape and cassette tape, card-type recording media such as IC (Integrated Circuit) card and optical card, and semiconductors such as mask ROM, EPROM (Erasable Programmable ROM), EEPROM (Electrically Erasable Programmable ROM), and flash ROM It can be either memory. However, the computer system 200 needs to be equipped with a reading device for reading from these recording media.
(Data structure of content and meta information)
The data processed by the content classification device 1 and the data structure of meta information representing the content will be described below with reference to FIG. FIG. 3 is a diagram illustrating content data representing content and meta information. According to this figure, the content data processed by the content classification device 1 has a structure in which a content area representing content and a meta information area representing meta information are integrated.

  The content data is literally data constituting the content. The meta information is predetermined information that represents the content. As shown in FIG. 3, the meta information includes a predetermined keyword. Each keyword represents the content. Each of these keywords belongs to a predetermined category. For example, the keyword “content ABC part 4” belongs to the category “title”. Further, the keyword “Ichiro Tanaka” belongs to the category “creator”.

Examples of categories to which keywords belong include creator, creation date, content size, and content description. Each of these categories is a predetermined attribute that expresses a specific feature related to the content.
(Details of classification database 6)
As described above, the content classification device 1 uses the classification database 6 to manage which group of content is to be classified. The details of the classification database 6 will be described below with reference to FIG. FIG. 4 is a diagram showing details of the classification database 6.

According to this figure, the classification database 6 is a database representing a correspondence relationship between contents and groups into which the contents are classified. That is, the classification database 6 stores the content title and the group name in association with each other. Furthermore, for each group, the name of the group and the category to which the name belongs are associated and stored. The “category” to which the group name belongs here is the same as the category to which the keyword representing the content content used when the group creating unit 15 determines the group name.
(Group classification flag)
The classification database 6 further stores a predetermined group classification flag. This group classification flag indicates whether or not a certain content is classified into a certain group. The group classification flag is associated with the title of the content and the name of the group into which the content is classified in the classification database 6.

When the group classification flag is “1”, the content is classified into groups. For example, the group classification flag associated with the content title “content ABC” and the group name “content ABC” is “1”. That is, the content “content ABC” is classified into the group “content ABC”. When the group classification flag is “0”, the content is not classified into groups.
(Keyword storage)
The classification database 6 further stores keywords used when it is decided to classify a certain content into a certain group. This keyword is associated in the classification database 6 with the title of the content and the name of the group into which the content is classified. The category to which the keyword stored in the classification database 6 belongs is the same as the category to which the name of the group to which this keyword is associated belongs. That is, the keywords are stored in association with groups having the same category.

  In the example shown in FIG. 4, the three contents whose titles are content ABC, content ABC part 2 and content ABC part 3 are all classified into one group whose name is content ABC. The name of the group “content ABC” belongs to the category “title” as shown in FIG. Therefore, the content classification device 1 determines whether or not to classify newly input content into this group, and selects a keyword belonging to the category “title” from among keywords included in the meta information representing the content. Determine by using for matching. That is, the content relevance determination unit 12 performs keyword matching by using keywords belonging to the same category.

As shown in FIG. 4, each of the contents “data 1”, “data 2”, “information 1”, and “information 2” is classified into the group “Yamada”. The name “Yamada” of this group belongs to the category “Creator”. Therefore, the content classification device 1 determines whether or not to classify newly input content into this group, and selects keywords belonging to the category “creator” from among the keywords included in the meta information representing the content. Determine by using for matching.
(Keyword extraction)
In the content classification device 1, the keyword extraction unit 11 extracts a predetermined keyword from the meta information. Furthermore, predetermined other keywords are extracted from other meta information. In the following, for convenience of explanation, the expression “keyword” includes meanings that are other keywords. The same applies to content, meta information, and programs.

  The keyword extraction unit 11 extracts all keywords belonging to each category from one piece of meta information. For example, when extracting keywords from the meta information shown in FIG. 3, the keyword extracting unit 11 extracts keywords belonging to the title, creator, creation date, content size, genre, and description category, respectively.

  The keyword extraction unit 11 may further extract a shorter partial character string such as a word from a character string extracted as a keyword belonging to a certain category. For example, there may be a plurality of content creators, not just one. In content created by multiple creators, the meta information representing the content includes multiple characters that represent each creator, such as “Taro Yamada, Ichiro Yamada” as keywords belonging to the category “creator”. Contains columns. That is, a plurality of information is included in one category. In such a case, the keyword extracting unit 11 may extract a plurality of pieces of information belonging to one category as one keyword.

  The keyword extraction unit 11 further converts only character strings belonging to the respective categories, for example, character strings belonging to the category “title” or character strings belonging to the category “contents description” to only nouns or only verbs by morphological analysis. For example, it may be extracted as a shorter partial character string.

At this time, the content relevance determination unit 12 may use each extracted partial character string as one keyword. In the following, an example will be described in which character strings belonging to each category extracted from meta information are used as keywords as they are.
(Determination of content relevance)
In the content classification device 1, the content relevance determination unit 12 determines whether or not the content newly input to the content classification device 1 is classified into a created group. At this time, the content relevance determination unit 12 compares the keyword extracted from the meta information of the input content with the keyword extracted from the meta information of each content included in the existing group. Thereby, it is determined whether the input content and the content existing in the group (other content) are related to each other.

  At this time, the content relevance determination unit 12 uses a keyword belonging to the same category as the category to which the name of the existing group belongs. For example, when the group name belongs to the category “title”, the keywords (for example, content ABC part 4) belonging to the category “title” extracted from the meta information of the input content and the existing group are classified. Compared with a keyword (for example, content ABC part 2) belonging to the category “title” extracted from the meta information of other content. When these keywords match each other, the content relevance determination unit 12 determines to classify the additional content into the group.

Based on this determination, the content classification unit 13 classifies the content into groups. That is, the group title extracted from the meta information of the content to be added and the keyword used for matching are stored in the classification database 6. Further, a group classification flag is set on the name of the group to be classified.
(Calculation of distance)
As described above, when the content relevance determination unit 12 at least partially matches the keyword of the content of the newly input content and the keyword indicating the content of the content classified into the existing group, It is determined that this content and other content are related to each other. Here, the content relevance determination unit 12 may determine that the two contents are related to each other when the distance between the two keywords is smaller than a predetermined threshold.

At this time, the content relevance determination unit 12 calculates the distance between the keyword and another keyword. Furthermore, this distance is compared with a predetermined threshold value. This threshold value is stored in advance in a memory (not shown) in the content classification device 1. When the calculated distance is smaller than a predetermined threshold, the content relevance determination unit 12 determines that the two compared keywords are matched with each other. When it is determined that the two keywords match, the content relevance determination unit 12 determines that the content and other content are related to each other.
(DP matching)
The content relevance determination unit 12 calculates the distance between the keyword and another keyword, for example, by DP matching. That is, based on this DP matching, it is determined whether or not the keyword matches another keyword. At this time, the content relevance determination unit 12 calculates the number of characters constituting the keyword as the number of characters N1. Further, the number of characters constituting another keyword is calculated as the number of characters N2.

  When using DP matching, the content relevance determination unit 12 calculates the distance between a keyword and another keyword using the following formula.

(Detailed example)
An example in which the content classification unit 13 classifies the content ABC part 4 into the group “content ABC” will be described below.

  First, when determining whether to classify the content ABC part 4 into the group “content ABC”, the content relevance determination unit 12 checks the category to which the name of this group belongs. The name of the group “content ABC” belongs to the category “title”. Therefore, the content relevance determination unit 12 extracts keywords belonging to the category “title” from the meta information representing the content of the additional content ABC part 4. In the meta information of the content ABC part 4, the keyword belonging to the category “title” is “content ABC part 4”. That is, the keyword extraction unit 11 extracts the character string “content ABC part 4” as a keyword.

In the content classification device 1, the content ABC, the content ABC part 2, and the content ABC part 3 are classified into the existing group “content ABC”. Therefore, the keyword extraction unit 11 extracts the character string “content ABC” as the title keyword from the meta information of the content ABC. Further, the character string “content ABC part 2” is extracted from the meta information of the content ABC part 2 as the keyword of the title. Further, the character string “content ABC part 3” is extracted from the meta information of the content ABC part 3 as a title keyword.
(Example of calculated distance)
The content relevance determination unit 12 sets the distance between the keyword “content ABC part 4” and the title keyword extracted from the meta information of each content included in the group “content ABC”, using the formula (1). ) By DP matching shown in FIG. Specifically, the distance between the keyword “content ABC part 4” and the keyword “content ABC” is calculated. Further, the distance between the keyword “content ABC part 4” and the keyword “content ABC part 2” is calculated. Further, the distance between the keyword “content ABC part 4” and the keyword “content ABC part 3” is calculated. Further, the distance between the keyword “content ABC part 4” and the character string “content ABC” which is the name of the group is calculated.

An example of the distance calculated by the content relevance determination unit 12 will be described below with reference to FIG. FIG. 5 is a diagram illustrating an example of the distance between keywords calculated by the content relevance determining unit 12. According to this figure, the content relevance determining unit 12 includes the keyword “content ABC part 4” and the name “content ABC”. ”Is calculated to be 0.345. On the other hand, the distance between the keyword “content ABC part 4” and the name “content ABC” is calculated to be 0.345. Further, the distance between the keyword “content ABC part 4” and the keyword “content ABC part 2” is calculated to be 0.15. Further, the distance between the keyword “content ABC part 4” and the keyword “content ABC part 3” is calculated to be 0.15.

(Comparison between judgment distance and reference maximum distance)
As described above, the content relevance determination unit 12 calculates the distance between the keyword extracted from the input meta information of the content and the name of the existing group to be 0.345. The content relevance determination unit 12 calculates this distance as a “determination distance”. On the other hand, the content relevance determination unit 12 calculates the largest distance among the distances between the keyword extracted from the meta information of the content to be added and the keyword extracted from the meta information of the content included in the existing group. , 0.345. The content relevance determination unit 12 calculates this distance as a “reference maximum distance”.

  As described above, the “determination distance” is the same as the “reference maximum distance”. That is, the content relevance determination unit 12 calculates the determination distance to be within the reference maximum distance. Thereby, the content relevance determination unit 12 determines to classify the content ABC part 4 into the group “content ABC”. Therefore, the content classification unit 13 classifies the content ABC part 4 into the group “content ABC”.

Thus, when it is determined that the determination distance is within the reference maximum distance, the content classification unit 13 adds the content to the group. That is, the content relevance determination unit 12 determines that the keyword is matched with another keyword when the calculated distance is smaller than a predetermined threshold.
(Update of classification database 6)
The content classification unit 13 adds the title of the content ABC part 4 to be added to the group to the classification database 6. That is, the character string “content ABC part 4” extracted from the meta information of the content ABC part 4 and used for content matching is stored in association with the group “content ABC”. Further, a group classification flag indicating that the content ABC part 4 is classified into the group “content ABC” is set in the field of the group “content ABC”. That is, this group classification flag is set to “1”.
(Use substring)
As described above, the content classification device 1 extracts the keyword “content ABC part 4” belonging to the category “title” from the input meta information of the content, and uses it for keyword matching. Here, the content relevance determination unit 12 may extract the partial character string from the character string “content ABC part 4” and use it for matching. For example, the partial character string “content” or the partial character string “ABC4” may be extracted from the content ABC part 4 and used for matching.
(Create a new group)
As described above, the content classification device 1 can classify input content into existing groups. In addition to this, the content classification device 1 can also create a new group and classify the content into this group. At this time, the content classification device 1 extracts one content from a plurality of contents, and determines whether or not the extracted content and other content are related to each other. Thereby, a plurality of contents are extracted from the contents related to each other, and these are classified into one newly created group. A specific example of this process will be described below.

FIG. 6 is a diagram illustrating an example of a group managed by the content classification device 1 and content classified into each group, (a) is a diagram illustrating an example of a group, and (b) is a content. FIG. 6C is a diagram illustrating an example for explaining the content in more detail. In the example illustrated in FIG. 6A, the content classification device 1 creates two groups. That is, the group “content ABC” and the group “Yamada”. The creation date of the group “content ABC” is January 21, 2005. The name of the group “content ABC” belongs to the category “title”. That is, the content classification device 1 creates this group by using keywords belonging to the category “title” for matching.

As shown in FIG. 6, the creation date of the group “Yamada” is February 28, 2005. The name of the group “Yamada” belongs to the category “Creator”. That is, the content classification device 1 creates the group “Yamada” by using keywords belonging to the category “creator” for matching.
(Example of content and group)
As shown in FIG. 6B, the content classification device 1 stores six contents in the content database 5. That is, three content “music α”, one content “shock video scoop”, one content “singing request”, and one content “music B” are stored.

  The dates when the content classification device 1 added the three contents “music α” to the content database 5 are February 14, 2005, February 21, 2005, and February 28, 2005, respectively. The date when the content “impact video scoop” was added to the content database 5 is February 23, 2005. Furthermore, the date when the content “kayo request” is added to the content database 5 is February 25, 2005. Furthermore, the date on which the content “kayo request” is added to the content database 5 is March 31, 2005.

  The content classification device 1 also stores meta information representing the content content in the content database 5 together with each content.

  In the example shown in FIG. 6C, the user selects the content “music α” whose creation date is February 14, 2005. Information on the content selected by the user is received by the input unit 2 and output to the content management unit 10. As a result, the keyword extraction unit 11 extracts keywords from the meta information of the content “music α”. That is, the character string “music α” is extracted as a keyword belonging to the category “title”. Further, the character string “music” is extracted as a keyword belonging to the category “genre”. Further, the character string “Ichiro Tanaka” is extracted as a keyword belonging to the category “creator”.

The keyword extraction unit 11 also extracts keywords from the meta information of the other five contents.
(Create group “Music α”)
The content relevance determination unit 12 performs DP matching by using keywords belonging to the same category. As shown in FIG. 6C, the keyword “music α” belonging to the category “title” has three keywords that match perfectly, that is, the keyword “music α”. That is, there are three contents whose title is music α. Therefore, the content relevance determination unit 12 determines that these three contents are related to each other.

The group creation unit 15 sets the matching keyword, that is, the character string “music α” as the name of the group to be created. That is, the group creation unit 15 creates a group whose name is music α. Accordingly, the content classification unit 13 classifies the three content “music α” into the group “music α” created by the group creation unit 15.
(Create group “music”)
In the example shown in FIG. 6C, there are further five keywords that perfectly match the keyword “music” belonging to the category “genre”. That is, there are five contents whose genre is music. Therefore, the content relevance determination unit 12 determines that these five contents are related to each other.

The group creating unit 15 sets the matching keyword, that is, the character string “music” as the name of the group to be created. That is, a group whose name is music is created. The content classification unit 13 classifies the five contents into the group “music α” created by the group creation unit 15.
(Created by the group “Ichiro Tanaka”)
In the example shown in FIG. 6C, there are two more keywords that completely match the keyword “Ichiro Tanaka” belonging to the category “creator”. That is, there are two contents whose creator is Ichiro Tanaka. Thereby, the content relevance determination unit 12 determines that these two contents are related to each other.

  The group creation unit 15 sets the matching keyword, that is, the character string “Ichiro Tanaka” as the name of the group to be created. That is, a group whose name is Ichiro Tanaka is created. The content classification unit 13 classifies the two contents into the group “music α” created by the group creation unit 15.

In this way, the content classification unit 13 classifies content and other content into groups in which character strings common to the keyword and other keywords are set as names. As a result, the content can be classified into groups that are closely related to the content.
(Determined by comparing with threshold)
The content relevance determination unit 12 may determine that two contents are related to each other when the distance between the keyword to be compared and another keyword is smaller than a predetermined threshold. For example, the content relevance determination unit 12 calculates the distance between the keyword “music α” and the keyword “music B” to be 0.28 by DP matching. At this time, it is assumed that the predetermined threshold is 0.35. This threshold value is stored in advance in a memory (not shown).

  The content relevance determination unit 12 compares the calculated distance (0.28) with a threshold value (0.35) stored in advance in a memory (not shown) to thereby determine the keyword “music α” and the keyword “music B”. Is determined to be smaller than a predetermined threshold. Thereby, the content relevance determination unit 12 determines that the keyword “music α” and the keyword “music B” are not perfect but match each other. Therefore, it is determined that the content whose title is music α and the content whose title is music B are related to each other.

Thereby, the content classification unit 13 can classify the content “music B” into the group “music α”. That is, more content can be classified into groups than when the keywords match perfectly.
(Use of substring)
As described above, the content relevance determination unit 12 can use the keyword extracted by the keyword extraction unit 11 as it is. However, the following problems may occur at this time. For example, when the keyword extraction unit 11 extracts keywords and the content relevance determination unit 12 searches for matching content, the matching content may be less than a predetermined number. Furthermore, the keyword extracted by the keyword extraction unit 11 may be longer than necessary and is not suitable for matching.

  In order to avoid these problems, the keyword extraction unit 11 may further extract at least a part of the extracted keywords as a partial character string. At this time, the content relevance determination unit 12 may use the partial character string extracted by the keyword extraction unit 11 for matching. At this time, the content relevance determination unit 12 uses a shorter keyword for content matching.

  The keyword extraction unit 11 may extract the character string “music” and the character string “α” as partial character strings from the character string “music α”. Furthermore, the content relevance determination unit 12 may use these partial character strings for matching.

For example, the content relevance determination unit 12 determines that the text “music α” and the text “music B” do not completely match each other.
At this time, the keyword extraction unit 11 extracts “music” and “α” as partial character strings from “music α”. Further, “music” and “β” are extracted as partial character strings from “music B”. The content relevance determination unit 12 uses these partial character strings extracted by the keyword extraction unit 11 as keywords for matching. At this time, the content relevance determination unit 12 determines that “music” extracted from “music α” and “music” extracted from “music B” completely match each other.
(Set the substring to the name of the group)
The group creation unit 15 may set partial character strings that completely match each other as the group name. That is, you may create the group which has the partial character string which matches completely as a name. For example, “music” extracted from “music α” and “music” extracted from “music B” are completely matched. Therefore, the group creation unit 15 may set the character string “music” as the name of the group to be created.
(Substring extraction method)
The keyword extraction part 11 should just extract a partial character string from a keyword based on a predetermined algorithm. For example, a character string of a predetermined length separated by different characters such as hiragana, kana, kanji, and symbols included in the keyword may be extracted as a partial character string. Alternatively, a partial character string included in this dictionary may be extracted from the keyword by referring to a keyword dictionary in which partial character strings used as keywords are stored in advance. Further, the keyword extraction unit 11 may perform morphological analysis on the keyword and extract, for example, only a noun as a partial character string.
(When content is not classified into groups)
In some cases, the content classification unit 13 may classify almost all of the content stored in the content database 5 into one group. This makes no sense to classify content into groups. Therefore, the content classification unit 13 does not have to classify content into groups even when the keyword of the content matches the name of the group. That is, even if it is determined that the contents are related to each other, the group creating unit 15 does not have to create a group to classify these contents. It should be noted that the group creation unit 15 determines whether to create the group depending on how much of the content stored in the content database 5 is included in the newly created group. May be determined.
(Categorize content into multiple groups)
The content relevance determination unit 12 may determine whether the input content and the content already classified into the existing group are related to each other. Accordingly, the content classification unit 13 can classify one content into a plurality of different groups.
(Keyword priority)
As described above, when the keywords included in the meta information match each other, the content relevance determination unit 12 determines that the two contents are created with each other. When the keywords match each other, the content classification unit 13 classifies these two contents into a common group. Here, the content relevance determination unit 12 may determine the keyword used for matching based on the priority set for the keyword.

  First, as keywords that can be set as the name of the group to be created, it is assumed that there is a text “music” belonging to the title category, a text “Tanaka” belonging to the creator category, and a text “music” belonging to the genre category. At this time, the priority setting unit 14 determines the priority set for the keyword candidate. This priority is a value that serves as an index of which keyword candidate to use as a keyword among a plurality of keyword candidates. The priority setting unit 14 outputs information on the priority to be determined to the content relevance determination unit 12.

  In the example shown in FIG. 6B, the group creation unit 15 creates a group in which the keyword “music” is set as the name. The content classifying unit 13 classifies the three contents whose title is music α and the one content whose title is music B into this group. That is, the group creating unit 15 extracts text “music” that perfectly matches music α and music B, and sets this text as the name of the group.

  Further, the group creation unit 15 creates a group in which the keyword “Tanaka” is set as the name. The content classification unit 13 classifies into this group one content whose creator is Ichiro Tanaka, two contents whose creator is Shiro Tanaka, and one content whose creator is Goro Tanaka. ing. That is, the keyword extraction unit 11 extracts the text “Tanaka” that perfectly matches Ichiro Tanaka, Shiro Tanaka, and Goro Tanaka. In this example, the distance between two keywords arbitrarily selected from Ichiro Tanaka, Shiro Tanaka, and Goro Tanaka is assumed to be smaller than a predetermined threshold.

  The priority setting unit 14 may set a higher priority for the keyword selected by the user. Or you may set a higher priority to the keyword candidate arbitrarily selected from the keyword candidates by the content classification device 1.

An example in which the content classification device 1 provides keyword candidates to the user and a higher priority is set for the keyword candidate selected by the user will be described below with reference to FIG. FIG. 8 is a diagram illustrating a display screen for allowing the user to select a keyword.
(Example of user selecting a keyword)
According to this figure, the display unit 3 displays music, Tanaka, and music on the display 4 as keyword candidates. Furthermore, the titles of the contents currently classified into each group are displayed at the same time. Thereby, the display unit 3 notifies the user of the name of the content classified into each group.

At this time, the display unit 3 may display the category to which the text belongs on the display 4 together with the keyword candidate text. The user selects an arbitrary keyword candidate as the name of the group to be created from the keyword candidates displayed on the display 4. That is, the user selects one keyword candidate. Alternatively, a plurality of keyword candidates may be selected. In this example, the user selects music as one keyword candidate. That is, the display unit 3 inverts the “select” button selected by the user and displays it on the display 4.
(Priority setting rules)
As described above, the content classification device 1 may determine the priority of keywords based on a predetermined rule. This rule is stored in advance in a memory (not shown) in the content classification device 1. As this rule, for example, a higher priority is set for a keyword belonging to a title category, and a lower priority is set for a keyword belonging to a genre category.

  The priority setting unit 14 may determine the priority to be set based on the length of the keyword candidate. At this time, a higher priority may be set for a longer keyword candidate, that is, a keyword candidate having a larger number of characters. Or, when the keyword candidate is set to the name of a certain group, the priority setting unit 14 sets the same priority as the priority set to the keyword representing the content classified into this group, This keyword candidate may be set. At this time, a higher priority may be set for a keyword that is the same as the name of a group into which more content is classified.

The priority setting unit 14 determines the priority for setting the priority for each keyword candidate based on the selection of keyword candidates by the user or the selection of keyword candidates by the content classification device 1. For example, when the user selects a plurality of keyword candidates, the same priority is set for the plurality of keyword candidates.
(Set priority based on keyword learning)
The priority setting unit 14 may use the keyword information learned by the preference learning unit 17 when selecting a keyword. As described above, the preference learning unit 17a learns keyword candidates and groups selected by the user, and groups deleted by the user. Thereby, the priority setting part 14 may select the keyword which the user likes as a keyword which sets a priority higher. Or the priority setting part 14 may select keywords other than the keyword which a user does not like as a keyword which sets a priority higher.

  The user inputs information about the user through the input unit 2. As this information, for example, the user inputs profile information such as the user's name, the user's affiliation, and information indicating the user's preferences. For example, the user inputs information indicating that the user likes music. Thus, the content classification device 1 stores and manages this information in a memory (not shown).

For example, the priority setting unit 14 sets the priority of the keyword “music” to be higher than the priority of other keywords based on the user information stored in the content classification device 1. That is, the priority of the keyword that matches the keyword that the user likes is set higher than the priority of other keywords.
(Example of setting priority based on classification configuration)
The priority setting unit 14 may determine the priority of the keyword based on the existing group classification configuration. Thereby, the group creation unit 15 can resemble the classification configuration of the group to be created with the classification configuration of another group that has already been created. This example will be described below with reference to FIG. FIG. 12 is a diagram illustrating a hierarchical structure of groups.

  According to this figure, the group creation unit 15 creates a group “Yamada” in the classification database 6. The name of this group belongs to the creator category. Further, the group creation unit 15 creates a group “information” under this group. The name of this group belongs to the title category. That is, when these two groups are created, the group creating unit 15 first creates a group for classifying content that matches the content by the creator's keyword, and among the contents classified into this group, A group is created to classify the matching content by the keyword of the title.

  The group creation unit 15 creates a group for classifying new content based on this classification configuration. The content classification unit 13 classifies the content into groups with the classification configuration created by the group creation unit 15. In other words, the group creating unit 15 first creates a group for classifying content that matches with the creator's keyword. As a result, the content classification unit 13 classifies content matched by the creator's keyword into this group. Next, the group creating unit 15 creates a group for classifying content matched by the keyword of the title among the contents classified into this group. As a result, the content classification unit 13 classifies the content matched by the keyword of the title into this group.

A group having a similar classification structure means that a category to which a group name belongs matches among higher groups, and further matches between groups.
(Priority setting)
For example, it is assumed that the priority setting unit 14 processes, as keyword candidates, Tanaka belonging to the creator category, music belonging to the title category, and music belonging to the genre category. As described above, in the group classification configuration that has already been created, a group in which “Yamada” belonging to the creator category is set as the name is present higher than other groups. At this time, the priority setting unit 14 sets the priority set for the keyword candidate “Tanaka” higher than the priority set for the keyword candidates “music α” and “music”.

When the priority setting unit 14 determines the priority of the keyword based on the group configuration already created, the display unit 3 selects the keyword with the higher priority on the display screen. Display it above.
(Summary)
Thus, the content management unit 10 unifies the group classification configuration. Thus, the user can easily grasp the group classification configuration when searching for content. Therefore, it is easy to search for content. As described above, it is very significant for the user that the priority setting unit 14 determines the priority to be set as the keyword candidate based on the existing group classification configuration.

  Hereinafter, an example in which the user selects the keyword “music” belonging to the title category will be described. When the user selects “music”, the priority setting unit 14 makes the priority of “music” higher than other keywords. The priority setting unit 14 updates the priority information set for the keyword in the classification database 6.

  The group creation unit 15 preferentially creates a group in which a keyword with a higher priority is set in the name. At this time, the group creation unit 15 updates the classification data stored in the classification database 6 shown in FIG. For example, a group in which music belonging to the title category is set as a name is created. Further, a flag is specified for the content belonging to this group. FIG. 7 shows the classification data after the update. FIG. 7 is a diagram showing details of the classification database.

  The display unit 3 creates display data representing information to be displayed on the display 4 based on the classification data stored in the classification database 6. The display unit 3 displays predetermined information on the display 4 based on the created display data. An example of information displayed on the display 4 by the display unit 3 will be described below with reference to FIG. FIG. 10 is a diagram illustrating an example of a display screen displaying groups and contents.

According to this figure, the display unit 3 displays three groups. That is, the group “content ABC”, the group “Yamada”, and the group “music” are displayed. Further, two contents not classified into these three groups are displayed. That is, the content “shock video scoop” and the content “song request” are displayed. The user can select these two contents directly from the displayed contents.
(Keyword learning)
The keyword selected by the user may represent the user's preference. Therefore, the preference learning unit 17 learns the keyword selected by the user. This example will be described.

  For example, it is assumed that the user can select music belonging to the title category, Tanaka belonging to the creator category, and music belonging to the genre category. The user selects “music” from these keywords. Accordingly, the preference learning unit 17 stores the keyword “music” selected by the user in a memory (not shown) as a keyword that the user likes. That is, the preference learning unit 17 learns the keyword “music”. Thereby, when classifying the content confirmed in the content database 5 into a group, the content classification device 1 uses this keyword in preference to other keywords.

When the user selects “music”, the priority setting unit 14 determines that the user likes the content whose title is music among the music genres. As a result, the priority setting unit 14 stores the keyword “music α” in a memory (not shown).
(Priority management database)
An example of a priority management database for managing priorities set for keywords will be described below with reference to FIG. FIG. 9 is a diagram illustrating an example of priorities set for each keyword. When the user selects music, the priority setting unit 14 adds a keyword whose keyword is music and whose category is genre. Furthermore, the point of this keyword is 1. This point represents the priority of the keyword.

  When the user selects the title “music” from among the music genres, the priority setting unit 14 adds a keyword whose genre is music, whose keyword is music, and whose category is title. Further, this keyword point is set to plus one. In the example shown in FIG. 9, the genre is music, the keyword is Tanaka, and the creator's keyword shown is also added. Further, the keyword point is set to minus one.

The priority setting unit 14 stores information on the keyword selected by the user in the priority management database. Furthermore, when the information input by the user includes information related to the user's preference, this information may also be stored in the priority management database.
(Example of deleting a group)
The content classification device 1 can delete the group designated by the user from the classification database 6. Therefore, an example in which the content classification device 1 deletes a group designated by the user will be described below.

  The user looks at the classification configuration displayed on the display 4 and instructs the content classification apparatus 1 to delete a group that is felt unnecessary. The group deletion unit 16 deletes the group designated by the user.

  Here, when a higher priority is set for the keyword selected by the user, the user will not feel particularly dissatisfied with the created group. However, when the keyword priority is automatically set, the group creation unit 15 may create a group in which a keyword unnecessary for the user is set as a name. Or, there is a possibility that a group that makes it difficult for the user to imagine the classified content is created.

  For example, it is assumed that there are a keyword “music” belonging to the category “title” and a keyword “Tanaka” belonging to the category “creator” as keywords of the group that can be created. Accordingly, it is assumed that the group creating unit 15 creates the group “music” and the group “Tanaka”. Here, the user determines that the group “Tanaka” is unnecessary. For example, in the display screen shown in FIG. 10, the group “Tanaka” is selected as the group to be deleted, and the delete button is further pressed.

  In the content classification device 1, the input unit 2 receives information regarding a group to be deleted designated by the user and outputs the information to the content management unit 10. In the content management unit 10, the group deletion unit 16 deletes the group designated by the user from the classification database 6. Further, the preference learning unit 17 learns the name of the group designated by the user as a character string that the user does not like.

Referring to the example shown in FIG. 9, when the user selects to delete the group “Tanaka”, the preference learning unit 17 uses the keyword that the genre is music, the keyword is Tanaka, and the category is creator. Is stored in the priority management database as minus one. The priority management database stores keywords with negative points as keywords that the user does not like.
(Example of group reorganization)
Furthermore, the content classification device 1 can remove the content specified by the user from the existing group. In addition, content specified by the user can be added to an existing group. These examples will be described next.

  The user may feel that the classification configuration of a plurality of contents classified into groups by the content classification unit 13 is not the classification configuration intended by the user. At this time, the user wants to delete unnecessary contents classified into the group from the group. Furthermore, content that is not classified into a group may want to be classified into this group. The content classification device 1 realizes these requests by the user by the following method.

  In the content classification device 1, the display unit 3 displays the current classification configuration on the display 4. The user who sees this display selects the content to be removed from the group from the contents displayed on the display 4 through the input unit 2. The input unit 2 receives the content “information” selected by the user and outputs information related to the received content to the content management unit 10. The content classification unit 13 removes the content selected by the user from the group in which the content is classified.

When the content removed from a certain group is not classified into any other group different from the group, the content classification unit 13 stores this content in the root directory. The display unit 3 displays on the display screen that this content is stored in the root directory. Further, the content classification unit 13 deletes the information on the content to be deleted from the group from the classification database 6.
(Add content to group)
When the user wants to add a certain content to the group, the user selects the content to be added on the display screen displayed on the display 4 and presses the group addition menu. Thereby, the display unit 3 displays a list of groups that can classify the content selected by the user on the display 4. The user selects a group into which content is to be classified from the displayed list.

  The content classification unit 13 adds the content selected by the user to the group selected by the user. At this time, content information to be added is added in the classification database 6. Specifically, a group classification flag is set for the group to be added. Further, the category to which the name of the group to be added belongs is checked, and keywords belonging to this category are extracted from the meta information of the category to be added. In addition, keywords extracted from the meta information are added to the classification database 6.

As described above, the content classification device 1 changes the configuration of the content classified into groups based on the designation by the user. At this time, the content classification device 1 may change the keyword matching threshold predetermined for the group to a new value that reflects the new group classification configuration. Alternatively, keyword matching may be newly executed between contents newly classified into a certain group, and the name of the group may be changed to a name that reflects the new content.
(Example of changing the threshold)
First, an example in which the content classification device 1 changes the keyword threshold will be described. The content classification device 1 stores a threshold related to distance in the classification database 6 for each group. That is, as shown in FIG. 4, the classification database 6 stores a threshold value set in advance for a group in addition to a group name, a group classification flag, and a matching keyword.

  As described above, the group deletion unit 16 removes the content designated by the user from the group in which the content is classified. As a result, the content classification unit 13 calculates L1 which is the minimum distance between the removed content keyword and the content keyword remaining in the group. Furthermore, L2 which is the maximum distance among the distances between the keywords of the content remaining in this group is calculated.

  Here, when the calculated L1 is larger than L2, the content classification unit 13 changes the threshold set for this group to L2. It should be noted that the content relevance determination unit 12 selects keywords belonging to the same category as the category used for keyword matching when performing keyword matching, when creating a group, or when adding content to the group. use.

  Since it is determined that unnecessary content is classified in the group, the user requests the content classification device 1 to remove the unnecessary content from the group. Further, L1 being larger than L2 means that an excessive content that should not be classified because the threshold used for keyword matching is too large is classified into a group. Therefore, the priority setting unit 14 changes the group threshold to a more appropriate value. As a result, the content classification unit 13 prevents the unnecessary content from being classified into groups when the content is next classified into groups.

  As described above, the group creation unit 15 creates a group “music”. In the group “Music”, three contents whose title is “Music α” and one content whose title is Music B are classified. FIG. 11 shows the distance between the keywords of these contents. FIG. 11 is a diagram showing a range of contents that can be classified into groups before and after removing the contents.

  Here, it is assumed that the user instructs the content classification apparatus 1 to remove the content “Music B” from the group “Music”. As a result, the group deletion unit 16 removes the content “Music B” from the group “Music”. At this time, three contents “music α” are left in the group “music” from which the content “music B” is removed. In these three remaining contents, the distance between keywords is calculated as “0”, which is smaller than L1. At this time, the content classification unit 13 calculates L2 as “0”. Since L1 is the same as L2, the content relevance determination unit 12 changes the threshold value of this group to “0”.

  FIG. 11 shows the range of content classified into a group before the content “Music B” is removed from the group and the range of content classified into a group after the content is removed. Even if the content “music B” is newly input to the content classification device 1, the content classification unit 13 does not classify the content into the group “music α”. Since the threshold value of the group is “0”, the content classification device 1 classifies only the content whose keyword completely matches, that is, the content “music α”, into the group “music α”.

  If the distance between the keyword “music γ” and the keyword “music α” is L1, the content classification device 1 determines that the group “music” before the music B is removed is the content “music γ”. Can be classified. However, the content “music γ” cannot be classified into the group “music α” after the content “music γ” is removed.

  When L1 and L2 are the same, the content relevance determination unit 12 may leave the threshold value as it is without changing the threshold value. However, the content classification unit 13 may remove a plurality of contents from one group at a time. At this time, the content relevance determination unit 12 calculates L2 ', which is the second largest after L2, among the distances between the keywords of the content remaining in the group. At this time, the calculated L2 'may be set as a new threshold for the group. Furthermore, the content classification unit 13 removes the content from the group when there is content related to the keyword whose distance from the other content is all L2 (L1) among the content remaining in the group. Also good.

  As described above, the group creation unit 15 created a group whose name is “music”. In the group “music”, the content “music α” and the content “music B” are classified. The content classification unit 13 removes the content “Music B” from the group “Music” based on the user's designation. At this time, if the content “music B” exists in the content remaining in the group, L1 and L2 are the same. That is, the distance between the keyword “music B” and the keyword “music α” is L2. This L2 is the same as L1.

As a result, the content classification unit 13 removes the content “Music B” from the group “Music”. Furthermore, the content relevance determination unit 12 changes the threshold value of this group to “0” smaller than L2. In this way, the content classification device 1 can automatically remove other content different from this content from the group based on the content specified by the user.
(Example of adding content to a group)
Next, an example in which the content classification unit 13 adds content specified by the user to a predetermined group will be described below.

  When the content classification unit 13 removes content specified by the user that is not classified into an existing group from the group, the content relevance determination unit 12 classifies the content keyword added to the group and the group in advance. L3 which is the maximum distance among the distances to the keyword of the content being calculated is calculated. Further, the calculated L3 is compared with a threshold value M set in advance for this group. When L3 is larger than the threshold value M, the content relevance determination unit 12 changes the threshold value of this group to the calculated L3. At this time, the content relevance determination unit 12 uses a keyword belonging to the same category as the category to which the group name belongs for matching.

It is highly likely that the user instructs the content classification apparatus 1 to add a certain content to the group because the range of the content to be classified into the group is desired to be wider. Therefore, when the calculated L3 is greater than the predetermined threshold M, the content relevance determination unit 12 increases the threshold set for the group. This broadens the range of content keywords that match the group name. Therefore, when a new content is added to the content classification device 1, if the distance between the keyword of the content classified into the group and the keyword of the added content is less than L3, the content classification unit 13 Can add this content to a group.
(Changing the classification structure)
The content relevance determination unit 12 may search the content database 5 for content that can be classified into this group when the predetermined threshold is changed to a new value. At this time, the content classification unit 13 calculates the distance between the keyword of the searched content and the keyword of the content classified into the group. When the distance calculated at this time is smaller than the threshold newly set for the group, the content classification unit 13 adds the content searched from the content database 5 to the group.

  As described above, the content classification device 1 removes one content from the group or adds one content to the group, thereby executing another content not classified in this group. The classification configuration can also be changed as appropriate.

When adding content to the group, the content classification unit 13 updates the classification database 6. That is, first, the content to be added is associated with this group in the classification database 6. Further, a group classification flag indicating that the content is classified into this group is set on the added content. Further, from the added content keywords, keywords belonging to the same category as the category to which the group name belongs are extracted and added to the classification database 6.
(Rename group)
When the content classification device 1 changes the number of contents to be classified into a group, the content classification apparatus 1 extracts a more appropriate keyword by executing matching between keywords of the content currently classified into this group, and the name of this group Can be set. This example will be described below.

  Hereinafter, an example will be described in which the content classification device 1 changes the name of the group after removing the content designated by the user from the group. The content classification unit 13 determines whether there is a character string that matches the keywords related to the content that remains in the group. When determining that there is a matching character string, the group creating unit 15 extracts this character string and adds it to the name of the current group. At this time, the group creation unit 15 adds the matched character string to the name of the group when complete matching is performed. Therefore, the group creation unit 15 sets a longer name for the group.

  The content classification device 1 uses a keyword belonging to the same category as the category to which the group name belongs for matching. By removing one content from a group, a longer character string may be matched when a keyword of content remaining in the group is matched. At this time, the content classification device 1 may change the name of the group to a matched character string.

  For example, the group creation unit 15 creates a group whose name is “music”. In the group “Music”, the content “Music α” and the content “Music B” are classified. Here, the content classification unit 13 removes the music B designated by the user from the group “music”. At this time, the content relevance determination unit 12 uses “music α”, which is a keyword of the content “music α” remaining in the group, for matching.

  As a result, the keyword extraction unit 11 extracts “music α”, which is a character string longer than “music”, which is the name of the current group, from the keyword. Therefore, the content classification unit 13 changes the name of the group from “music” to “music α”.

  The group creating unit 15 may connect a newly extracted character string to a name of an existing group via a predetermined symbol. Specifically, “α” is connected to “music” via “−”, and the group name is set to “music-α”. Thereby, even when the connection between the name of the original group and the newly extracted character is unnatural, a more natural name can be set for the group. Further, the user can know that a plurality of keywords are used when the name of the group is determined by looking at the name of the group. Specifically, it is known that “music” and “α” are associated with this group as keywords.

  Before and after removing the content from the group, the content relevance determination unit 12 may calculate an average value of the distances between the keywords of the content classified into the group. At this time, when the average value calculated after removing the content from the group is larger than the average value calculated before removing the content, the keywords of the content remaining in the group are classified into the group before removing the content. It is determined that there is a higher possibility that long characters match than the keywords of the content in question. Therefore, at this time, the content relevance determination unit 12 may determine whether there is a newly matching character in the content keyword.

  The content relevance determination unit 12 determines whether or not the partial character string of the keyword of the added content matches the name of the group when content specified by the user and not currently classified into the group is added to the group. You may judge. If it is determined to match, the group creating unit 15 may leave the group name unchanged. At this time, the content relevance determination unit 12 may use a keyword belonging to the same category as the category to which the group name belongs for matching. Matching here means matching in advance.

  When the partial character string of the content keyword and the name of the group do not match, the content relevance determination unit 12 shares the keyword of the content newly classified into the group in common before adding the content. It may be determined whether or not it matches a partial character string of the name. When determining that matching is to be performed, the group creating unit 15 may change the group name to a partial character string that matches the group name.

  When the content relevance determination unit 12 determines that there are no matching keywords, the content relevance determination unit 12 compares the keywords belonging to other categories among the content keywords to determine whether there are other matching keywords. It may be determined. When determining that it exists, the group creation unit 15 may set a newly matched character string as the name of the group.

As described above, when changing the name of the group, the content classification device 1 also changes the name of the group managed in the classification database 6.
(Program classification device 1a)
The details of the content classification apparatus 1 shown in FIG. 1 have been described above. Next, a program classification apparatus 1a that acquires a program as contents and further acquires program information as meta information and automatically classifies the received programs into groups based on the acquired program information will be described. The program classification device 1a has a function of acquiring program information and a function of recording a program, such as a hard disk recorder. Further, it has a function of recommending to the user a program determined to be suitable for the user.

  First, details of the program classification device 1a will be described below with reference to FIG. FIG. 13 is a block diagram showing details of the program classification device 1a according to the embodiment of the present invention.

  According to this figure, the program classification device 1a includes an input unit 2a, a display 4a, a program database 5a, a classification database 6a, a communication antenna 7a, and a program management unit 10a.

  Further, the program management unit 10a includes a keyword extraction unit 11a (keyword extraction unit), a program relevance determination unit 12a, a program classification unit 13a (program classification unit), a priority setting unit 14a (priority setting unit), and a group creation unit. 15a, a group deletion unit 16a (group deletion unit), a preference learning unit 17a (preference learning unit), and a program recommendation unit 18a (program recommendation unit).

Among these members, those having the same functions as those provided in the content classification device 1 shown in FIG. The program classification device 1a processes a program as content. Further, program information is processed as meta information.
(Program receiver 8a)
The program receiving unit 8a receives a broadcast program and stores it in the program database 5a. Program information representing this program will be described below with reference to FIG. FIG. 14 is a diagram illustrating an example of program information. As shown in this figure, the program information stores keywords belonging to a certain category as text information for each program. For example, the category includes genre, title, broadcast station name, performer, and content description.

  The program receiving unit 8a receives program information of a program of an arbitrary number of days from a broadcasting station in a specific time zone, and stores it in the program database 5a. When receiving a program broadcast according to a broadcasting system such as digital satellite broadcasting, the program receiving unit 8a can simultaneously receive program information added to the program. The program receiving unit 8a may receive program information separately from the program through the Internet or the like.

  The user inputs information related to program recording through the input unit 2a. The display unit 3a classifies and displays the recorded programs on the display 4a. The program classification unit 13a classifies programs designated by the user into existing groups. At this time, program information representing the contents of the program is used.

  The program management unit 10a may use program information stored in the classification database 6a when classifying programs into groups. At this time, the user inputs a program ID (program identification information) that identifies the program. The program management unit 10a specifies a program designated by the user in the program database 5a based on the input program ID. The program management unit 10a classifies the programs into groups based on the program information of the specified programs.

Note that the processing when the program classification device 1a classifies the program into groups is substantially the same as the processing when the program classification device 1a classifies the content into groups, and thus detailed description thereof is omitted.
(Program recommendation unit 18a)
The program recommendation unit 18a recommends a program to the user. At this time, the user inputs a program recommendation request through the input unit 2a. The program recommendation unit 18a specifies a program recommended to the user from the program database 5a based on information related to the user's preference stored in a preference information database (not shown).

  The display unit 3a displays on the display 4a information related to the program recommended to the user specified by the program recommendation unit 18a.

  When determining the priority of a keyword, the user selects one keyword from a plurality of keywords displayed on the display 4a. For example, “Actor B” is selected from the keywords “Actress A”, “Actor B”, and “Actress C”. Thereby, the group creation unit 15a creates a group whose name is the actor B. The program classification unit 13a classifies programs into this group.

At this time, the preference learning unit 17a stores the keyword “actress B” in the preference information database as a keyword that the user likes. The program recommendation unit 18a extracts the keyword “Actor B” stored in the preference information database, and recommends the program represented by the program information including this character string as a keyword to the user.
(Display recommended programs)
The display unit 3a creates classification data for displaying the program extracted by the program recommendation unit 18a, and displays information on the program recommended to the user on the display 4a based on the created classification data. For example, the display screen shown in FIG. 15 is displayed on the display 4a.

In this manner, the preference learning unit 17a learns information selected by the user when classifying programs into groups. Based on the information learned by the preference learning unit 17a, the program recommendation unit 18a estimates a program that the user likes and recommends it to the user. Thereby, the program classification device 1a can recommend a program closely related to the user without the user inputting the user's preference in advance.
(Learning based on recording or viewing)
The preference learning unit 17a may learn that the program recorded by the user is a program that the user likes. At this time, the preference learning part 17a should just extract a keyword from the program information of the program designated to be recorded by the user, and store the extracted keyword in the preference information database as a favorite keyword.

  The preference learning unit 17a may learn that the program viewed by the user is a program that the user likes. At this time, the preference learning unit 17a may extract a keyword from the program that the user is viewing, that is, the program information of the viewing program, and store the extracted keyword in the preference information database as a favorite keyword.

  When learning the user's preference based on the viewing program or the recorded program, the preference learning unit 17a includes keywords included in the program information of the program viewed by the user or the program information of the program to be recorded at a very high rate. You may learn as a favorite keyword.

  As shown in FIG. 9, the priority setting unit 14a may set a priority for a keyword. That is, every time a user views or records a program, the keyword points included in the program information of the program are increased. At this time, the program recommendation unit 18a determines that a program with a higher point is a program that the user likes more.

  When the number of keyword points is increased each time a user views, a higher priority is set for a keyword of a program that is viewed more frequently, which is advantageous. That is, the program recommendation unit 18a extracts a keyword having a higher point, and recommends a program including the extracted keyword to the user.

  In this way, the preference learning unit 17a learns the user's preference based on the history of the program viewed by the user or the history of the recorded program. At this time, the preference learning unit 17a can learn the information preferred by the user more accurately by using the keyword selected by the user as information preferred by the user when selecting a program as a group.

  For example, it is assumed that the user has recorded the program represented by the program information including the keywords “Actor A” and “Actor B” belonging to the performer category a plurality of times. The title of the program A is “program A” and the genre is drama. Each time the user instructs recording of the program A, the preference learning unit 17a adds points to the title program A, the genre drama, and actors A and B of the performers.

  The preference learning unit 17a increases the priority set for the program and the keyword of the program in the same manner. Thereby, the program recommendation unit 18a recommends both a program including the actor A as a keyword and a program including the actor B as a keyword to the user.

Here, it is assumed that the user likes the actor A but does not like the actor B. At this time, the program in which the actor A appears is likely to be a program that the user does not like. However, based on the history of the program recorded by the user and the history of the program viewed, the preference learning unit 17a learns which keyword the user likes when the program information includes a plurality of keywords. ,Have difficulty.
(Keyword priority and program recommendation)
When a plurality of programs including “program A” as a keyword are stored in the content database 5, the user selects the program A from these programs and instructs the program classification apparatus 1 a to classify it into a group. And At this time, keyword candidates that can be selected by the user include “actor A”, “actor B”, the title of program A, and the genre (drama) of program A. At this time, the user selects actor B as a keyword candidate, for example. Thereby, the priority setting unit 14a makes the point of “Actor B” higher than the point of “Actor A”. In the program classification device 1a, when the user selects the actor B, a value to be added to the points is determined in advance.

  When the user selects “actor B” as a keyword candidate, the preference learning unit 17a can make the actor B's point higher than the actor A's point. At this time, the program classification device 1a recommends a program with a higher point to the user. At this time, the program recommendation unit 18a preferentially recommends the program represented by the program information including the actor B as a keyword to the user over the program represented by the program information including the actor A as a keyword. Thereby, it can suppress recommending to the user the program in which actor A appears.

In this way, the preference learning unit 17a learns the keyword specified by the user when the program classification device 1a classifies the program in addition to the history of the program recorded by the user and the history of the program viewed by the user. Therefore, the program recommendation unit 18a can recommend the program with higher accuracy by recommending the program based on the learned information.
(Example of display screen)
An example of a display screen displayed on the display 4a when the content classification device 1 classifies programs into groups will be described below with reference to FIG. FIG. 16 is a diagram illustrating an example of a display screen displayed on the display 4a when the user selects a program to be classified into groups. FIG. 16A is a display screen displaying a list of recorded programs. (B) is a diagram showing a display screen on which the user has selected program A, (c) is a diagram showing a display screen for allowing the user to select a group to be created, and (d) is a diagram showing the user It is a figure showing the display screen which displays that the program selected by was classified into the group which the user selected.

  According to this figure, first, the display unit 3a displays a list of recorded programs on the display 4a. At this time, the keyword of the program is extracted from the program information of each program and displayed simultaneously. For example, the program title, recording date / time, genre, performer, and description of contents are displayed.

  At this time, when the user selects the classification menu, the display unit 3a displays the display screen shown in FIG. 16B on the display 4a. The user selects a program to be classified into a group from the programs displayed on the display screen. In the example of FIG. 16B, the user has selected “program A”.

  The keyword extraction unit 11a extracts a plurality of keywords that can be set as the group name from the program information of the program A selected by the user. Thereby, the display part 3a displays the display screen shown in FIG.16 (c) on the display 4a. The user selects a keyword to be set as the group name from the keywords selected on the display screen. In the example of FIG. 16C, the user has selected the keyword “actor B”.

When the user selects a keyword, the display unit 3a displays a display screen shown in FIG. 16D on the display 4a. That is, the display unit 3a displays a display screen indicating that the program A is classified into the group “actor B”.
(Use two keywords with different categories)
Hereinafter, an example in which program information of one program includes a program title, a program description, and a program content description as keywords will be described. The program classification device 1a processes a plurality of programs. Each of the plurality of program information representing each of the plurality of programs includes “team Al” indicating a professional baseball team as a keyword. Thereby, the program relevance determination unit 12a determines matching using the team A as a keyword. By doing so, it is determined that a plurality of programs represented by the program information including the keyword are matched.

  The program relevance determination unit 12a further narrows down contents to be matched by another keyword from among a plurality of programs matched by the keyword “Team A”. At this time, keywords belonging to other categories, for example, program genres, are extracted from a plurality of pieces of program information representing these programs. The group creation unit 15a creates a group in which a character string common to the keywords extracted from the genre is set as the name.

  As described above, the program relevance determination unit 12a determines that the programs matched by the keyword “team A” and further matched by the keyword “baseball” are related to each other.

  Thereby, the group creation unit 15a creates a group whose program genre is “baseball” and whose program description is “team A”. At this time, the group creating unit 15a may create a group whose name is “Team A”. Alternatively, the group creation unit 15a may create a group whose name is “baseball-team A”. That is, the group creation unit 15a may create a group in which any one of keywords used for matching is set as a name. Or you may create the group which set the text which connected each keyword used for matching by predetermined symbols, such as a hyphen, to the name.

The content classification unit 13 classifies only programs that match the created group with both the keyword “team A” and the keyword “baseball”. Therefore, neither a program matching only by team A nor a program matching only by baseball is classified into this group. Furthermore, when a plurality of programs are matched only by team A, the group creation unit 15a does not create a group in which team A is set as a name.
(Example of not creating a group)
When a user who uses the program classification device 1a reserves a program, it does not have to specify a group for classifying the program to be reserved. On the other hand, in the prior art apparatus, the user needs to specify which group to classify each time a program is reserved. In other words, the program classification device 1a can automatically classify programs into groups without allowing the user to designate groups.

  Further, the program classification device 1a classifies programs into existing groups based on keywords representing program contents included in program information representing programs. Alternatively, a new group is created and programs are classified into the created group. That is, the group creation unit 15a creates only a group necessary for classifying programs. Therefore, programs that are not classified into the created group are placed in the root directory. Thereby, the user can directly search for programs classified into groups without going through the groups.

If all programs are classified into groups, the user must always designate a group into which the programs are classified once when selecting a program. In the content classification device 1, the user can omit the procedure and can directly select a program. For example, it is not necessary to classify programs that are deleted immediately after being recorded by the user. Therefore, the content classification device 1 does not classify such programs into groups.
(Example of classifying programs into multiple groups)
The program classification device 1a can classify one program into a plurality of different groups. At this time, the program classification device 1a uses a plurality of keywords representing programs for matching, and when matching by a certain keyword, the program classification device 1a classifies the program into a group in which the keyword is set as a name, When matching by, this program is classified into other groups in which this keyword is set in the name. At this time, the user can search for one program from a plurality of groups. Therefore, it is easier to search for programs than when one program is classified into one group.

The content classification device 1 extracts keywords from a plurality of programs currently stored in the content database 5 and creates groups in which characters to be compared with these keywords are set in names. As a result, the program classification device 1a can create an optimal group for classifying the programs currently stored in the content database 5.
(Problem of conventional example)
On the other hand, in the conventional example, when a user reserves a movie program, the user specifies that the program is stored in the group “movie” in the item of genre. Thereafter, if a plurality of movie programs are reserved and stored in the group “movie” in the same manner, many programs are stored in the group “movie”. In the conventional example, if an item such as a day of the week is selected by the user at the time of reservation, the apparatus creates a small group for “day of the week” in the “movie” group. However, if this is not the case, the device stores many programs in the “movie” group. Therefore, it is difficult for the user to search for a program.

Conversely, if the user selects all items (title, day of the week, time, etc.), the device will change from “movie” to “day of the week” while there are few programs in the “movie” group. You must first select a group. This increases the number of search procedures. As described above, in the method in which the user selects an item at the time of reservation, it is not possible to know how much more programs will enter the “movie” group at the time of reservation. Therefore, it is difficult for the user to select an item accurately at the time of reservation.
(Advantages of the present device)
The program classification device 1a can cope with a case where the number of programs in the “movie” group increases and the user wants to create a small group in the “movie”. When a user selects a certain program in “movie”, the program classification device 1a extracts a program having a keyword common to the program selected in the “movie” and the program data, and creates a group. The program data includes information such as performers and movie directors. Therefore, for example, if the user often records movies of the same director, the program classification device 1a can create a group of movies of the director. Therefore, the program classification device 1a can create a group in accordance with the programs stored in the classification database 6.

  When matching a keyword, the program classification device 1a calculates a distance between the keywords by extracting a portion where characters match between the keywords. Whether or not to match is determined by whether or not the calculated distance is closer than a threshold value. By doing so, it is possible to deal with fluctuations in the notation between keywords.

  For example, a series number such as “criminal suspense drama 1” or “criminal suspense drama 2” may be given to a program title such as a series program. Even in such a case, the program classification device 1a can create groups with titles to be compared by regarding the titles of these groups as the same titles. Further, the program title included in the program information includes omitted notations and notation fluctuations, but the program classification device 1a can cope with such cases.

  Whether or not keywords match each other depends on a threshold value of the distance between keywords. As the threshold value is decreased, keywords with more common characters are matched. The program classification device 1a automatically changes the threshold for the created group in accordance with the correction performed by the user. For example, an example of a group that classifies programs that match each other with the title “criminal suspense drama” will be described. It is assumed that a program titled “Suspense Theater” is classified into this group because the distance between keywords is within the threshold.

  In this case, when the program titled “Suspense Theater” designated by the user is removed from the group, the program classification device 1a makes the threshold of the group smaller than the threshold before the program is left. . Thereby, the program classification device 1a can prevent the program titled “Suspense Theater” from being reclassified into the group.

  When a program is classified, if there are a plurality of groups that can be created, the program classification device 1a can create a group designated by the user. The program classification device 1a learns the keywords of the group selected by the user as keywords that the user likes. Thereby, the program classification | category apparatus 1a can recommend a program to a user based on the learned keyword.

On the other hand, if information related to user preferences is stored in advance, when the user selects one group from the groups that can be created, the program classification device 1a uses this information. Thereby, the content classification device 1 can create a group having the keyword that the user likes as a name.
(Flow of each process executed by the content classification device 1)
The details of the content classification device 1 and the program classification device 1a have been described above. Next, the flow of processing executed by the content classification device 1 will be described below with reference to the flowcharts shown in FIGS.

  First, an example in which the content classification device 1 adds input content to an existing group will be described below with reference to FIG. FIG. 17 is a flowchart showing the flow of processing when the content classification device 1 classifies the input content into groups.

  The user inputs content into the content classification device 1 (step S11). The keyword extraction unit 11 extracts a predetermined keyword from the meta information representing the content of the input content (Step S12). The content relevance determination unit 12 determines whether or not the keywords extracted by the keyword extraction unit 11 match each other (step S13). When the keywords match, the group creation unit 15 classifies the contents into groups (step S14). The group creation unit 15 further updates the classification data in the classification database 6. That is, information regarding the added content is stored in the classification database 6. Further, a group classification flag is set for the group into which the contents are classified.

  The flow of processing when the content classification device 1 deletes the group specified by the user will be described below with reference to FIG. FIG. 18 is a flowchart showing the flow of processing when the content classification device 1 deletes a group designated by the user and further changes the name of the group. According to this figure, the user inputs an instruction for requesting deletion of a group to the content classification apparatus 1 (step S21). The group deletion unit 16 deletes the group designated by the user (step S22). The preference learning unit 17 learns the name of the group deleted by the group deletion unit 16 (step S24). Further, the content management unit 10 updates the classification data in the classification database 6.

  The flow of processing when the content classification device 1 learns keywords will be described below with reference to FIG. FIG. 19 shows a flow of processing when the content classification device 1 sets a higher priority to the keyword selected by the user or the content classification device 1 and the content classification device 1 learns the keyword selected by the user. FIG.

  According to this figure, the user or content classification device 1 selects content to be classified into groups (step S31). The keyword extraction unit 11 extracts a predetermined keyword from the meta information representing the content of the selected content (Step S32). The content relevance determination unit 12 is based on the priority set for the extracted keyword. A keyword used for matching is determined (step S33). The content relevance determination unit 12 determines whether or not a plurality of keywords match (step S34). The preference learning unit 17 determines whether or not the user has selected a keyword (). When the determination result is “true” in step S35 (Yes), the preference learning unit 17 learns the keyword selected by the user (step S36). Further, the content management unit 10 updates the classification data in the classification database 6.

  The flow of processing when the content classification device 1 changes the threshold predetermined for the group to a more appropriate value will be described below with reference to FIG. FIG. 20 is a flowchart showing the flow of processing when the content classification device 1 changes the threshold value predetermined for the group.

According to this figure, the user instructs the content classification device 1 to make a request to change the group threshold (). The content relevance determination unit 12 calculates the distance between the keywords by using the keywords of the content classified into the group for matching, and changes the threshold value of the group. Furthermore, if necessary, the group creation unit 15 changes the name of the group (step S43). Further, the content management unit 10 updates the classification data in the classification database 6.
The flow of processing when the content classification device 1 displays the current classification configuration of content on the display 4 will be described below with reference to FIG. FIG. 21 is a flowchart showing the flow of processing when the content classification device 1 displays the content classification status on the display 4 in accordance with a user instruction.

  According to this figure, the user requests the content classification apparatus 1 to display the content and group on the display 4 (step S51). Thereby, the display part 3 acquires the present classification structure of a group by accessing the classification database 6, and displays it on the display 4 (step S52).

Further, the present invention can also be realized with the following configuration.
(First configuration)
A storage unit that stores a plurality of contents and content meta information; a selection unit that selects content stored in the storage unit; and a search for content having content meta information common to the content meta information of the selected content And an information processing apparatus that classifies a plurality of contents according to content having the common content meta information.
(Second configuration)
In the first configuration, an information processing apparatus having keyword extracting means for extracting a keyword from content meta information, and determining whether or not the content meta information is common depending on whether the keyword can be matched.
(Third configuration)
In the second configuration, whether matching is possible is determined by extracting a portion where characters match between keywords, calculating a distance between keywords, and determining that keywords that are closer than a certain threshold match each other. A characteristic information processing apparatus.
(Fourth configuration)
In the third configuration, the information processing apparatus is characterized in that, after content is classified as being capable of being matched, the threshold is corrected when the classification is corrected by the user.
(Fifth configuration)
An information processing apparatus according to the second configuration, wherein keywords are classified according to a keyword having a high priority among keywords commonly included in a plurality of content meta information.
(Sixth configuration)
An information processing apparatus according to the fifth configuration, characterized in that the user has means for selecting a keyword, and the priority of the keyword is determined based on the selection result of the user.
(Seventh configuration)
In the fifth configuration, an information processing apparatus having user information storage means and determining keyword priorities based on user information.
(Eighth configuration)
The information processing apparatus according to the fifth configuration, wherein the priority of the keyword is determined by a classification configuration of other contents.
(Ninth configuration)
In the second configuration, when a keyword 1 is matched with content meta information of a plurality of contents, a condition is added to the keyword such that the keyword 1 can be used for classification only when matching with another keyword 2 is possible at the same time. An information processing apparatus characterized by the above.
(Tenth configuration)
An information processing apparatus having display means in the second configuration.
(Eleventh configuration)
In the second configuration, an information processing apparatus having preference learning means and learning a keyword used for classification as a preference keyword.
(Twelfth configuration)
In the second configuration, an information processing apparatus capable of canceling a generated classification.
(13th configuration)
In the twelfth configuration, an information processing apparatus that learns a keyword that generates a canceled classification as a non-preferred keyword.
(14th configuration)
In the second configuration, after classifying content, when the classification is corrected by the user, the name of the classification is changed.
(15th configuration)
An information processing apparatus having a program acquisition means, a program information acquisition means, and a means for classifying a program, and a display means for displaying a classification status in the first configuration.
(Sixteenth configuration)
In the fifteenth configuration, the information processing has a history storage unit, has a preference learning unit that learns a preference from a history, and determines the priority of keywords commonly included in program information based on the learned preference apparatus.
(17th configuration)
The information processing apparatus according to the sixteenth configuration, wherein the program is recommended with the learned preference information.
(18th configuration)
A step of storing a plurality of contents and content meta information, a step of selecting content stored in the storage means, and a step of searching for content having content meta information common to the content meta information of the selected content; An information processing method, wherein a plurality of contents are classified according to the common content meta information.
(19th configuration)
A program for realizing the eighteenth configuration.
(20th configuration)
A recording medium storing a program having an eighteenth configuration.

  The present invention can be used as an apparatus that receives a program and automatically classifies the received program into a predetermined group, such as a television, a hard disk recorder, and a personal computer.

It is a block diagram showing the detail of the content classification apparatus which concerns on one Embodiment of this invention. It is a figure which shows the computer system which implement | achieves a content classification apparatus. It is a figure explaining the content data showing a content and meta information. It is a figure showing the detail of a classification database. It is a figure which shows the example of the distance between the keywords which a content relevance determination part calculates. It is a figure showing the example of the group which the content classification apparatus manages, and the content classified into each group. It is a figure showing the detail of a classification database. It is a figure showing the display screen for making a user select a keyword. It is a figure showing the example of the priority set to each keyword. It is a figure showing the example of the display screen which is displaying the group and the content. It is a figure showing the range of the content which can be classified into a group before and after removing a content. It is a figure showing the hierarchical structure of a group. It is a block diagram showing the detail of the program classification | category apparatus which concerns on one Embodiment of this invention. It is a figure showing an example of program information. It is a figure showing the list of the programs recommended to a user currently displayed on the display. It is a figure showing the example of the display screen displayed on a display when a user selects the program classified into a group, (a) is a display screen which displays the list of the recorded programs, (b) Is a diagram showing a display screen on which the user has selected program A, (c) is a diagram showing a display screen for allowing the user to select a group to be created, and (d) is a program selected by the user. It is a figure showing the display screen which displays that it classified into the group which the user selected. It is a flowchart showing the flow of a process when a content classification device classify | categorizes the input content into a group. It is a flowchart showing the flow of processing when a content classification device deletes a group designated by a user and further changes the name of the group. It is a figure showing the flow of a process when a content classification device sets a higher priority to the keyword which the user or the content classification device selected, and also the content classification device learns the keyword which the user selected. It is a flowchart showing the flow of a process when a content classification apparatus changes the threshold value predetermined for the group. It is a flowchart showing the flow of a process when a content classification apparatus displays the classification status of a content on a display according to a user's instruction | indication.

Explanation of symbols

1 content classification device 2 input unit (keyword selection means, user information input means, group name selection means)
3 Display unit 4 Display 5 Content database 6 Classification database 10 Content management unit 11 Keyword extraction unit (keyword extraction means)
12 Content relevance determination unit (content relevance determination means, keyword distance measurement means)
13 Content classification section (content classification means)
14 Priority setting section (priority setting means)
15 Group creation section 16 Group deletion section (content deletion means)
17 Preference learning part (preference learning means)
1a Program classification device 2a input unit (keyword selection means, user information input means, group name selection means)
3a display unit 4a display 5a program database 6a classification database 7a communication antenna 8a program receiving unit (program receiving means)
10a Program management unit 11a Keyword extraction unit (keyword extraction means)
12a Program relevance determining unit (content relevance determining means)
13a Program classification part (content classification means)
14a Priority setting unit (priority setting means)
15a Group creation unit 16a Group deletion unit (group deletion means)
17a Preference learning unit (preference learning means)
18a Program recommendation section (preference learning means)

Claims (28)

It is determined that the content and the other content are related to each other when the additional information representing the content and the other additional information representing the content of other content different from the content are at least partially matched. Content relevance determining means for
A content classification apparatus comprising: content classification means for classifying the content determined to be related to each other by the content relevance determination means and the other content into the same group. The additional information includes a keyword representing the content content,
The other additional information includes other keywords representing the contents of the other content,
The content relevance determining means is
2. The content classification apparatus according to claim 1, wherein when it is determined that the keyword and the other keyword are at least partially matched, it is determined that the content and the other content are related to each other. . The content relevance determining means is
The content according to claim 2, wherein when the distance between the keyword and the other keyword is smaller than a predetermined threshold, it is determined that the content and the other content are related to each other. Classification device.   The content classification apparatus according to claim 2, wherein the name of the same group is a character string common to the keyword and the other keyword.   The content classification apparatus according to claim 2, wherein the category to which the keyword belongs is the same as the category to which the other keyword belongs.   The content relevance determination means further includes a keyword belonging to another category different from the category included in the additional information and another keyword belonging to the other category included in the other additional information. 6. The content classification apparatus according to claim 5, wherein the content classification device determines that the content and the other content are related to each other when at least partially matching.   The content classification apparatus according to claim 2, wherein the keyword is a keyword having a higher priority. Keyword selection means for accepting selection of the keyword by the user;
The content classification apparatus according to claim 2, further comprising priority setting means for setting a higher priority for the keyword selected by the user. A group name selection means for receiving the name of the group selected by the user;
3. The content classification apparatus according to claim 2, further comprising priority setting means for setting a higher priority for the keyword including the name of the group received by the group name selection means. . User information input means for receiving input of information about the user;
Claims further comprising priority setting means for setting a higher priority to a keyword including a character string included in the information related to the user received by the user information input means. Item 3. The content classification device according to Item 2. The content classification means
8. The content classification apparatus according to claim 7, wherein a character string common to a keyword set with a higher priority and another keyword is set to the name of the group located in a higher hierarchy. When a hierarchical group exists in advance,
Claims further comprising priority setting means for setting a higher priority to a keyword belonging to the same category as a category to which a keyword including the name of a group located in a higher hierarchy belongs. Item 8. The content classification device according to Item 7.   The content classification apparatus according to claim 1, further comprising group deletion means for deleting the group designated by the user. A group deleting means for deleting the group instructed by the user;
8. The content classification apparatus according to claim 7, further comprising priority setting means for setting a priority of a keyword including the name of the group deleted by the group deletion means to a lower level.   2. The content classification apparatus according to claim 1, further comprising display means for displaying a result of classification by the content classification means on a predetermined display. A preference learning unit for learning the keyword selected by the user;
The content classification device according to claim 2, further comprising content recommendation means for recommending content to a user based on the keyword learned by the preference learning unit. Content adding means for adding new content to the group based on user instructions;
When the new content is added to the group by the content adding means, the maximum of the distance between the keyword representing the content of the new content and the keyword representing the content of the content included in the group in advance Keyword distance measuring means for calculating the distance of
4. The content classification apparatus according to claim 3, further comprising threshold setting means for setting the maximum distance as a new threshold when the maximum distance is larger than a predetermined threshold. Content deletion means for removing the content instructed by the user from the group;
When the content is removed from the group by the content deletion means, a keyword representing the content of the first content arbitrarily selected from the group and a first content different from the first content arbitrarily selected from the group A keyword distance measuring means for calculating the maximum distance among the distances to other keywords representing the contents of the two contents;
4. The content classification apparatus according to claim 3, further comprising threshold setting means for setting the maximum distance as a new threshold when the maximum distance is smaller than the predetermined threshold. . A content adding means for adding new content designated by the user to the group;
The content classification means
When the new content is added to the group by the content adding means, a character string common to the keyword representing the content of each content included in the group is set as a new name of the group. The content classification device according to claim 3. A content deleting means for removing the content instructed by the user from the group;
The content classification means
When the content is removed from the group by the content deletion means, a character string common to the keyword representing the content of each content newly included in the group is newly set as the name of the group The content classification device according to claim 2. Program receiving means for receiving a predetermined program as the content;
The content classification device according to claim t, further comprising program information acquisition means for acquiring program information representing the contents of the program as the additional information. The additional information includes a predetermined keyword representing the content of the program,
The other additional information includes a predetermined other keyword representing the content of the other program,
The content relevance determining means is
The content classification apparatus according to claim 21, wherein when it is determined that the keyword and the other keyword are at least partially matched, it is determined that the program and the other program are related to each other. . Program information storage means for storing the program information representing the program requested by the user for at least one of recording and viewing on a predetermined recording medium;
The apparatus further comprises priority setting means for setting a higher priority to the keyword including the character string included in the program information stored in the recording medium. 22. The content classification device according to 22. A preference learning unit for learning the keyword including a character string included in the program information stored in the recording medium;
The content classification apparatus according to claim 23, further comprising program recommendation means for recommending a program to a user based on the keyword learned by the preference learning unit. Keyword selection means for accepting selection of the keyword by the user;
A preference learning unit for learning the keyword selected by the user;
The content classification apparatus according to claim 22, further comprising program recommendation means for recommending a program to a user based on the keyword learned by the preference learning unit. It is determined that the content and the other content are related to each other when the additional information representing the content and the other additional information representing the content of other content different from the content are at least partially matched. A content relevance determining step,
A content classification comprising a content classification step of classifying the content and the other content into the same group when it is determined that the content and the other content are related to each other Method.   A content classification program for operating the content classification device according to any one of claims 1 to 25 for causing a computer to function as each of the means. A computer-readable recording medium on which the content classification program according to claim 27 is recorded.

Images