JP2006094018A - PROGRAM RECOMMENDATION DEVICE, PROGRAM RECOMMENDATION METHOD, PROGRAM, AND RECORDING MEDIUM CONTAINING THE PROGRAM - Google Patents

Abstract

PROBLEM TO BE SOLVED: To appropriately recommend a program suitable for a user's preference considering a genre.
A program recommendation device includes a program guide data storage unit 8 for storing program guide data for each receivable program, and a program for a program specified by user information among programs indicated by the program guide data. Based on the extracted information and the user information, the keyword, the genre of the keyword, the keyword The recommended program is determined based on the user data update unit 11 that creates user data including an evaluation value for user preference, the user data storage unit 12, and the evaluation value of each keyword of the user data and the genre of the program to which the keyword belongs. Preference point calculation unit 14, matching score calculation unit 15, and display information generation Part 16 is provided.
[Selection] Figure 1

Description

  The present invention relates to a program recommendation device, a program recommendation method, a program, and a recording medium on which the program is recorded. The program recommendation apparatus recommends a program to a user using, for example, program information (EPG: Electronic Program Guide) of a television program.

  2. Description of the Related Art Conventionally, systems that determine a user's degree of preference for a program based on an operation associated with the user's program viewing and recommend a program to be viewed to the user are known. For example, Japanese Patent Laid-Open No. 2002-142160 discloses a system that recommends a program to a user using EPG text information. Specifically, the EPG text information of the program viewed by the user is decomposed, the keywords obtained thereby are accumulated, and parameters such as keyword type and keyword viewing frequency (viewing frequency for programs including the keyword) are stored. Based on the above, a preference point representing the degree of preference of the user with respect to the keyword is calculated, and a program including the keyword is recommended based on the calculated preference point.

  A storage example of the user's viewing history in this configuration is shown in FIG. In the example of the figure, from the EPG text information of the program viewed by the user, the program title, performer, and genre (for example, drama, sports, etc.) are extracted as keywords, the keyword type is set for each of these keywords, Stores the viewing frequency of each keyword. The keyword and the viewing frequency are updated every time the user views the program. The keyword types may include program titles, performers, program genres, words included in EPG text information, and the like.

The preference point for each keyword is calculated based on the viewing frequency, and it is determined that the higher the preference point, the stronger the degree of preference for the keyword. Therefore, when recommending a program, a program including many keywords with high preference points is extracted and recommended.
JP 2002-142160 (released May 17, 2002)

  However, in the above conventional configuration, preference points for each keyword are determined based on the frequency of only the keywords from the EPG text information of all programs viewed by the user, and the EPG texts of all programs scheduled to be broadcast. A keyword matching a keyword with a high preference point is searched from the information, and a viewing program is recommended. For this reason, for example, a program that does not match the viewing tendency of the user who places importance on the genre of the program may be recommended.

  For example, performers such as actors and singers tend to appear depending on the genre of the program. Actors often appear in dramas, and singers often appear in music programs. Among the user's preference points, the preference point of a predetermined performer A is high, and when the performer A is an actor, there is a high possibility that the drama in which the performer A appears matches the user's preference. However, there is a high possibility that the sports broadcast program in which performer A happens to appear as a guest does not match the user's preference.

  Thus, as a user's preference tendency, a program of a predetermined genre is often viewed when performer A appears, but a program of another genre may not be viewed even when performer A appears. In such a case, the conventional method of accumulating keywords from the entire programs viewed by the user and searching for recommended programs based only on the keywords cannot cope with user preferences that differ depending on the genre.

  Further, when searching for a recommended program, even if a preference point of a genre of a program, which is one of keyword types, is used, a similar adverse effect may occur. For example, when a user often watches both a drama and a sports program, the preference points of keywords whose genre is a keyword type such as drama and sports are high. At the same time, when the user often watches the drama of the actor A, the preference point of the actor A whose keyword type is the performer is also high. Therefore, when searching for a recommended program, if a sports program in which actor A appears as a guest is scheduled to be broadcast, it is considered that both the actor A and the sports have high preference points, and the program is recommended. However, even if the user likes the actor A and the sports program, he likes the actor A who appears in the drama. Therefore, just because actor A appears in a sports program does not necessarily view that program.

  Thus, when the tendency of preference for a keyword differs depending on the genre, the conventional configuration cannot recommend a program that matches the user's preference.

  Therefore, the present invention provides a program recommendation device, a program recommendation method, and a program recommendation method that can appropriately recommend a plurality of items included in the program guide information that have different categories, for example, a program suitable for the user's preference considering keywords and genres. An object is to provide a program and a recording medium on which the program is recorded.

  The program recommendation apparatus according to the present invention includes a plurality of programs having different categories included in the data from the program guide data for the program specified by the user information among the programs indicated by the program guide data for each receivable program. The program data extracting means for extracting the items, and the user's information about the related item group under the relation between the plurality of items based on the information indicated by the item and the user information extracted by the program data extracting means. User data creating means for creating user data including an evaluation value for preference and recommended program determining means for determining a recommended program based on the evaluation value of the user data are provided.

  Further, the program recommendation method of the present invention is based on the program guide data for the program specified by the user information among the programs indicated by the program guide data for each receivable program. A program data extraction step for extracting a plurality of different items, and information related to the items related to each other under the relationship between the plurality of items, based on information and user information indicated by the items extracted by the program data extraction step. It comprises a user data creation step for creating user data including an evaluation value for user preference, and a recommended program determination step for determining a recommended program based on the evaluation value of the user data.

  According to the above configuration, among the programs indicated by the program guide data, the user information (for example, history information about the user's viewing program, recording reservation information by the user, or input information specifying the program by the user) is specified. A plurality of items having different categories included in the data are extracted from the program guide data for the program.

  Next, based on the information indicated by the extracted item and the user information, user data including an evaluation value for the user's preference for the related item group under the relationship between the plurality of items is created. Thereafter, a recommended program is determined based on the evaluation value of the user data.

  As described above, in the present invention, instead of determining an evaluation value for a user's preference in a keyword from all programs viewed by the user, a group of items related to each other under the relationship between a plurality of items having different categories. An evaluation value for the user's preference is determined, and a recommended program is determined based on the evaluation value. Therefore, according to the present invention, it is possible to appropriately recommend a plurality of items included in the program guide information that have different categories, for example, a program that matches the user's preference in consideration of keywords and genres.

  The program recommendation device according to the present invention includes a program guide data storage means for storing program guide data for each receivable program, and the program for a program specified by user information among programs indicated by the program guide data. Program data extraction means for extracting a keyword and a program genre included in the data from the guide data, and based on the information extracted by the program data extraction means and user information, the keyword and the program to which the keyword belongs User data creation means for creating user data including a genre and an evaluation value for user preference in the genre of the program to which the keyword belongs, user data storage means for storing the user data, and the user data The evaluation value and keyword of each keyword It is characterized by and a recommended program determining means for determining a recommended program based on the genre of a program to be.

  The program recommendation method according to the present invention includes a keyword and a program genre included in the program guide data for the program specified by the user information among the programs indicated by the program guide data for each receivable program. A program data extraction step, and based on the information extracted by the program data extraction step and the user information, the keyword, the genre of the program to which the keyword belongs, and the genre of the program to which the keyword for the keyword belongs A user data creating step for creating user data including an evaluation value for the user's preference, and a recommended program determining step for determining a recommended program based on the evaluation value of each keyword of the user data and the genre of the program to which the keyword belongs And features that It is set to.

  According to the above configuration, among the programs indicated by the program guide data, the user information (for example, history information about the user's viewing program, recording reservation information by the user, or input information specifying the program by the user) is specified. A keyword and a program genre included in the data are extracted from the program guide data for the program. Next, based on the extracted information and user information, the user data including the keyword, the genre of the program to which the keyword belongs, and the evaluation value for the preference of the user in the genre of the program to which the keyword belongs. Created. Thereafter, a recommended program is determined based on the evaluation value of each keyword of the user data and the genre of the program to which the keyword belongs.

  The recommended program may be determined based on evaluation values of one or more keywords included in the program guide data of the program. In this case, for example, evaluation values of a plurality of keywords are evaluation values of keywords of the same genre. Further, when recommending a program, a program including many keywords with high evaluation values is extracted and recommended based on the evaluation value of each keyword.

  As described above, in the present invention, the evaluation value for the keyword is determined for each genre of the program to which the keyword belongs, instead of determining the evaluation value for the user's preference in the keyword from all the programs viewed by the user. . The recommended program is determined based on the evaluation value of each keyword and the genre of the program to which the keyword belongs. Therefore, in the present invention, it is possible to appropriately recommend a program suitable for the user's preference in consideration of keywords and genres.

  That is, even if the keywords are the same, the evaluation value for the user's preference differs depending on the genre. Therefore, by determining the evaluation value of the keyword for each genre, it is possible to cope with a situation where the same keyword is a favorite keyword in one genre but not a favorite keyword in another genre.

  Specifically, when the user often watches the drama in which the actor A appears, the evaluation value of the actor A is high in the drama genre. Therefore, when a recommended program is searched, if there is a program in which the actor A appears as a guest and the genre is a sport, a method that does not consider the program genre can be recommended against the user's preference. Increases nature. However, in the present invention, since the keyword evaluation value is determined for each genre, the program is not recommended unless the evaluation value of the actor A is high in sports.

  Here, the genre is information included in program guide data, for example, EPG text information, and is a classification of programs. The keywords are program titles, performers, program genres, words included in EPG text information, and the like.

  The program recommendation device may further include a recommended program display unit that displays the recommended program determined by the recommended program determination unit.

  According to the above configuration, the recommended program can be clearly notified to the user.

  In the program recommendation device, the user information includes information indicating a program viewed by the user, and the evaluation value is an integrated value of the number of occurrences of the keyword in the program guide data of the program viewed by the user. Also good.

  According to said structure, the said evaluation value about each keyword can be set appropriately based on the viewing history information of the program by a user.

  In the program recommendation device, the recommended program determination unit includes an evaluation value calculation unit that calculates a second evaluation value based on the first evaluation value when the evaluation value is the first evaluation value. And determining a recommended program based on a second evaluation value, wherein the evaluation value calculating means weights the first evaluation value based on a keyword notation order in the program guide data, and The evaluation value of 2 may be calculated.

  According to the above configuration, the first evaluation value is weighted and the second evaluation value is calculated based on the notation order of the keywords in the program guide data, so that an appropriate evaluation value is set for each keyword. Is possible. This makes it possible to more appropriately recommend a program that matches the user's preference.

  For example, when the genre of a program is a drama, there is usually a high possibility that the user of a plurality of performers likes the user. In this case, in the notation order of the program guide data (for example, EPG text information), the names of performers are often written in order from the person who starred. It is preferable to lighten the later person.

  In addition, even if the user likes actor A who is the lead of a certain drama X, he / she is watching another drama Y and there is an actor B who appears in both drama X and drama Y. Actor B is higher than actor A in frequency (keyword evaluation value). As a result, the evaluation value of the actor B becomes high, and a drama in which the actor B appears is preferentially recommended. However, such a problem can be avoided in the configuration of the present invention.

  As described above, in the configuration of the present invention, since the keyword evaluation value is determined for each genre, it is possible to apply a weight to the keyword evaluation value for each genre. Thereby, a user's preference can be appropriately reflected according to the feature of a genre, and it is possible to recommend a program that more matches the user's preference.

  In the program recommendation device, the recommended program determination unit includes an evaluation value calculation unit that calculates a second evaluation value based on the first evaluation value when the evaluation value is the first evaluation value. And determining a recommended program based on a second evaluation value, wherein the program guide data includes role information indicating a role in the program for performers of the program, and the evaluation value calculating means includes a program Based on the role information in the guidance data, the first evaluation value of the performer as a keyword may be weighted to calculate the second evaluation value.

  According to the above configuration, when the program guide data (for example, EPG text information) includes a tag for distinguishing whether the performer is a program presenter or a guest, for example, in a music program, the guest is more than the presenter. The user is likely to like Therefore, in this case, it is preferable to reduce the weight of the presenter.

  As described above, the first evaluation value of the performer as the keyword is weighted based on the role information in the program guide data, and the second evaluation value is calculated, so that the evaluation value of each keyword is appropriately set. It can be set. As a result, it is possible to appropriately reflect the user's preference and recommend a program that better matches the user's preference.

  In the program recommendation device, the recommended program determination unit includes an evaluation value calculation unit that calculates a second evaluation value based on the first evaluation value when the evaluation value is the first evaluation value. And determining a recommended program based on a second evaluation value, wherein the evaluation value calculating means weights the first evaluation value to calculate a second evaluation value, and the weighting amount May be changed according to the genre of the program to which the keyword belongs.

  According to the above configuration, the first evaluation value is weighted to calculate the second evaluation value, and the weighting amount is changed according to the genre of the program to which the keyword belongs. Corresponding to the case where the appearance frequency of the keyword in the guidance data is different, the balance of the evaluation value of the keyword can be appropriately set between different genres. As a result, it is possible to appropriately reflect the user's preference and recommend a program that better matches the user's preference.

For example, it is assumed that the number of viewing times and the number of days from the viewing date to the present are set as parameters for determining the keyword evaluation value. At this time, the evaluation value (preference point) of the keyword A in the genre α can be expressed as follows.
f (α, A) = X × g (X, α) + Y × g (Y, α)
X: number of viewing times, Y: number of days from viewing date to present, g (x, y): weight when parameter x is genre y In this case, the number of viewing times and the last viewing date for one keyword And the evaluation value (preference point) is determined by the parameter values. When the drama A is viewed every time and the keyword A1 included in the drama A is stored, if the drama A is viewed every time, the number of times of viewing increases, and the evaluation value (preference point) of the keyword A1 increases. On the other hand, when the broadcast of the drama A is finished and viewing of another new drama B is started, the evaluation value (preference point) of the keyword B1 included in the drama B increases. However, since the number of times of viewing is initially low, it is expected that the evaluation value (preference point) remains low compared to the keyword A1.

  Here, there is a case where the user recommends a preference based on a preference estimated from recent viewing data, rather than the size of the viewing count. In such a case, for example, adjustment is performed by weighting so that the evaluation value (preference point) becomes smaller as the number of days from the viewing date to the present time increases. The degree of weighting at this time may be changed depending on the genre. The reason why weights are set for each genre is that keywords appear differently depending on the genre, and that there are long-lasting preferences and rapidly changing preferences. is there.

  For example, in the case of a continuous drama, there are many cases where the program changes every three months. Although the interest in each keyword is maintained during the drama broadcast, the interest in the keyword may decrease when the drama broadcast ends. In order to deal with such a case, it may be determined that the user has disliked after at least three months have passed since the last viewing date, and the evaluation value (preference point) may be weighted to be zero.

  In the case of a drama of a series such as a 2-hour drama, the next series may be broadcast after 3 months. The weight may be set so as not to become too small.

  Similarly, guest performers of music programs such as singers do not appear frequently, so the number of times of viewing does not increase, but there are cases where users like it. In such a case, it is considered that the preference lasts long, and the evaluation value (preference point) may be set so that the evaluation value (preference point) does not become too small even if the number of days from the last viewing date to the present is large.

  The history itself may be deleted depending on the number of days from the last viewing date to the present. When there is a limit to the number of histories for storing keywords, the number of days from the last viewing date to the present day may be limited, and the limited number of days may be set according to the genre of the program from which the keywords are extracted.

  For example, the maximum storage period of the keyword A extracted from the sports genre program is 50 days, and the maximum storage period of the keyword B extracted from the drama genre program is 30 days.

  In this way, by changing the weight applied to the parameter for calculating the evaluation value (preference point) according to the genre, it becomes possible to deal with the appearance degree of the keyword and the user's interest level which are different for each genre.

  In the program recommendation device, the recommended program determination unit includes an evaluation value calculation unit that calculates a second evaluation value based on the first evaluation value when the evaluation value is the first evaluation value. The recommended program is determined based on the second evaluation value, and the evaluation value calculation means validates only the evaluation value for the keyword having the highest first evaluation value in a predetermined genre, The second evaluation value may be calculated by invalidating the evaluation values of other keywords.

  According to the above configuration, in a predetermined genre, for example, when the genre of the program is sports, only the evaluation value for the keyword having the highest first evaluation value is made valid, and the evaluation values for other keywords are made invalid. Since the second evaluation value is calculated, it is possible to appropriately reflect the user's preference and recommend a program that better matches the user's preference.

  For example, when the genre is baseball, it is assumed that the team name is extracted as a word keyword. If the user likes team A, the first evaluation value (viewing frequency) is expected to increase because the team A watches the game that the team A battles. However, although the other teams (team B, team C, team D) that compete with team A are not as much as team A, it is considered that the first evaluation value (viewing frequency) gradually increases. If this happens, the evaluation values of other teams will also rise, and there is a possibility that a game in which team B will play against team C will be recommended.

  In this case, when the first evaluation value (viewing frequency) of the team A is the highest, the second evaluation value is given only to the team A, and the second evaluation value of the other team is set to 0. It is possible that only the game A is recommended.

  Of the keywords in the genre, the preference points are calculated only for the keywords with the highest first evaluation value (viewing frequency), and the preference points for other keywords are set to zero. If the team name can be identified, such as registering all team names in advance, the second evaluation value is calculated only for the team with the highest viewing frequency among the team names, and the second evaluation for the other team names A configuration in which the value is 0 may be used.

  In this way, by storing keywords for each genre, it is possible to set conditions for calculating the evaluation value of the keyword for each genre.

  In the program recommendation device, the recommended program determination unit includes an evaluation value calculation unit that calculates a second evaluation value based on the first evaluation value when the evaluation value is the first evaluation value. And the recommended program is determined based on the second evaluation value, and the evaluation value calculation means uses the program guide data when calculating the second evaluation value based on the first evaluation value. The second evaluation value may be calculated by weighting the first evaluation value based on the keyword ranking.

  According to the above configuration, when calculating the second evaluation value based on the first evaluation value, the first evaluation value is weighted based on the keyword ranking in the program guide data. Since the evaluation value 2 is calculated, it is possible to appropriately reflect the user's preference and recommend a program that more matches the user's preference.

  In the above program recommendation device, the program data extracting means extracts only performers whose notation rank in the program guide data is up to a predetermined Nth as a keyword when the genre of the program is drama or movie. It is good.

  According to the above configuration, since the Nth and subsequent performers are generally considered to have a low user preference level, it is preferable not to extract them as keywords. As a result, it is possible to avoid a situation in which a program is recommended based on performers who do not like the user very much.

  That is, in the case of a drama or a movie, it is generally considered that the user is likely to like the actor in the main performer among the performers. However, in fact, it is more likely that the supporting actor will appear in more dramas and movies than the main actor. For this reason, when the user views a plurality of dramas, the supporting frequency of the actor acting in the supporting role increases, and it is not possible to recommend a program that matches the user's preference. As described above, the present invention can avoid such a problem.

  The program recommendation device according to the present invention is a program data for extracting, from the program guide data for a program specified by user information, among a program indicated by the program guide data, a plurality of items included in the data that have different categories. A user including an evaluation value for the user's preference for the related item group under the relation between the plurality of items based on the information indicated by the item extracted by the extraction means and the program data extraction means and the user information User data creating means for creating data and recommended program determining means for determining a recommended program based on an evaluation value of the user data.

  The program recommendation device according to the present invention includes a program guide data storage unit, a keyword included in the data from the program guide data for a program specified by user information among the programs indicated by the program guide data, and Program data extracting means for extracting the genre of the program, and based on the information extracted by the program data extracting means and user information, the keyword, the genre of the program to which the keyword belongs, and the keyword for the keyword belong Based on user data creation means for creating user data including an evaluation value for user preference in the genre of the program, user data storage means, the evaluation value of each keyword of the user data, and the genre of the program to which the keyword belongs A recommended program determining means for determining a recommended program; It is a configuration that example.

  The program recommendation method of the present invention is a program data for extracting a plurality of items having different categories included in the data from the program guide data for the program specified by the user information among the programs indicated by the program guide data. A user including an evaluation value for a user's preference for the related item group under the relationship between the plurality of items based on the extraction step and the information and user information indicated by the item extracted by the program data extraction step A user data creating step for creating data, and a recommended program determining step for determining a recommended program based on an evaluation value of the user data.

  Further, the program recommendation method of the present invention is a program for extracting a keyword and a program genre included in the data from the program guide data for the program specified by the user information among the programs indicated by the program guide data. Based on the data extraction step, the information extracted by the program data extraction step and the user information, the keyword, the genre of the program to which the keyword belongs, and the user preference in the genre of the program to which the keyword belongs for the keyword A user data creating step for creating user data including an evaluation value for the user, and a recommended program determining step for determining a recommended program based on the evaluation value of each keyword of the user data and the genre of the program to which the keyword belongs. It is a configuration.

  In the present invention, the evaluation value for the keyword is determined for each genre of the program to which the keyword belongs, instead of determining the evaluation value for the user's preference in the keyword from all the programs viewed by the user. The recommended program is determined based on the evaluation value of each keyword and the genre of the program to which the keyword belongs. Therefore, in the present invention, it is possible to appropriately recommend a program that matches the user's preference considering the genre.

An embodiment of the present invention will be described below with reference to the drawings.
FIG. 1 is a block diagram showing the structure of the program recommendation device of this embodiment. In the present embodiment, the program recommendation device performs, for example, a recommendation operation of a viewing program for a user using program information (electronic program guide: EPG) of a television program.

  As shown in FIG. 1, the program recommendation apparatus according to the present embodiment includes an input unit 4, an input unit 5, an EPG receiving unit 6, and a display unit (recommended program display means) 17. The viewing information 1 is history information of a program viewed by the user, and the input is accepted by the input unit 4. The viewing information 1 is input to the input unit 4 every time the user views a certain program.

  The recommendation request information 2 is a recommendation request for a program to be viewed from the system or the user, and the input is accepted by the input unit 5. EPG data 3 including EPG text information is received by the EPG data receiving unit 6. The display unit 17 is for displaying recommended programs.

  The program recommendation device further includes an EPG data processing unit 7, a program guide data storage unit (program guide data storage unit) 8, a program guide data reference unit (program data extraction unit) 9, and a genre identification unit (program data extraction unit) 10. , User data update unit (user data creation unit) 11, user data storage unit (user data storage unit) 12, user data reference unit 13, preference point calculation unit (recommended program determination unit) 14, match degree calculation unit (recommended program) Determination means) 15 and a display information generation unit (recommended program determination means) 16.

  When EPG data is received by the EPG receiving unit 6, the EPG data processing unit 7 extracts keywords from the EPG data 3. The program guide data storage unit 8 stores the keywords extracted by the EPG data processing unit 7 for each program as program guide data together with the program genre and keyword type as shown in FIG. 3B, for example. The keyword types are title, performer, and word in FIG.

  When the viewing information 1 is input to the input unit 4, the program guide data reference unit 9 watches the program indicated by the viewing information 1 among the program guide data stored in the program guide data storage unit 8, that is, the user has watched. A keyword is extracted from program guide data of a program (hereinafter simply referred to as a user viewing program). The genre identifying unit 10 identifies the genre of the user viewing program.

  As shown in FIG. 4, the user data storage unit 12 stores keywords and their attribute data (such as viewing frequency) as user data for each genre, and for each keyword type within one genre. Therefore, the user data update unit 11 determines the attribute of the keyword extracted from the program guide data reference unit 9 when there is a matching keyword in the same genre and the same keyword type in the user data of the user data storage unit 12. Update the data. On the other hand, for keywords extracted from the program guide data reference unit 9, if there is no matching keyword in the same genre and the same keyword type in the user data of the user data storage unit 12, the keyword is added and used as attribute data. For example, 1 is given to the viewing frequency.

  When the recommendation request information 2 is input to the input unit 5, the user data reference unit 13 refers to the user data in the user data storage unit 12, and the preference point calculation unit 14 calculates a preference point for each keyword of the user data. Information to do this is supplied to the preference point calculation unit 14. The preference point calculation unit 14 calculates a preference point for each keyword of the user data based on the information supplied from the user data reference unit 13. The calculated preference points are stored in the user data storage unit 12. The recommendation request information 2 is input by a user's operation of, for example, a button provided in the program recommendation device or a remote controller (hereinafter simply referred to as a remote controller).

  Based on the preference points calculated by the preference point calculation unit 14, the degree-of-match calculation unit 15 calculates the program guide data stored in the program guide data storage unit 8 (for example, each program guide data scheduled to be broadcast) for the user. The degree of coincidence with the user's preference is calculated. The display information generation unit 16 extracts a program recommended to the user based on the degree of matching calculated by the degree of matching calculation unit 15.

  The display unit 17 displays the program extracted by the display information generation unit 16. A display example of the recommended program on the display unit 17 is shown in FIG. In this example, a list of recommended program dates and times, broadcast stations, and program titles is displayed. In this case, the programs may be displayed in order of programs with a high degree of match, or may be displayed in order of programs with the earliest broadcast start date and time.

  In the program recommendation device, a recommended program is extracted within the range of the EPG data 3 received by the EPG receiving unit 6. Note that the EPG receiving unit 6 receives the EPG data 3 from the current date and time up to N days later (N is an arbitrary value) in one reception.

  The program recommendation device 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. FIG. 2 shows an internal configuration of the computer system 200.

  Referring to FIG. 2, this 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 controller 204.

  In addition to the optical disk drive 201 and the magnetic disk drive 202, the computer 220 includes a remote control interface (I / F) 205 that receives a signal from the remote control 204, a CPU (central processing unit) 206, a remote control interface 205, a CPU 206, and an optical disk. A bus 207 connected to the drive 201 and the magnetic disk drive 202, a read-only memory (ROM) 208 connected to the bus 207 and storing a boot-up program and the like, and also connected to the bus 207, program instructions, system programs, And a random access memory (RAM) 209 for storing work data and the like.

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

  The program for causing the computer system 200 to realize the function of the program recommendation device, the EPG data 3, and the viewing information 1 are all stored in the optical disc 210 or the magnetic disc 211 inserted into the optical disc drive 201 or the magnetic disc drive 202, Further, it is transferred to the hard disk 212. Alternatively, the program, the EPG data 3, and the viewing information 1 may be transmitted to the computer 220 through a network (not shown) and stored in the hard disk 212. The program is loaded into the RAM 209 when executed. 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 program includes a plurality of instructions for causing the computer 220 to realize the function of the program recommendation device 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 program does not necessarily include all the functions necessary for realizing the functions of the program recommendation device according to this embodiment. If this program contains only instructions for controlling the program recommendation device described above by calling an appropriate function or “tool” in a controlled manner so that a desired result can be obtained. Good. The operation of computer system 200 is well known and will not be repeated here.

  A recording medium on which a program for realizing the function of the program recommendation device, EPG data 3, and viewing information 1 is recorded is a CD-ROM (Compact Disc Read Only Memory), MO (Magneto-Optical disc), MD. It is not limited to the optical disk 210 such as (Mini Disc) and DVD (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.

Next, each functional unit shown in FIG. 1 will be described in more detail.
An example of the EPG data 3 received by the EPG receiving unit 6 is shown in FIG. As shown in the figure, EPG data 3 includes, for each program, the start date / time and end date / time of the program broadcast, the name of the broadcasting station that broadcasts the program, the name of the genre to which the program belongs, the program name (title), and the performer name. A program description describing the contents of the program is described as text information. The performers may be tagged with a tag that identifies the moderator, commentator, etc.

  The EPG data processing unit 7 extracts the program name, performers (keywords), and genre from the text information of the input EPG data 3, and stores these information as program guide data for each program in the program guide data storage unit 8. Store. Further, the EPG data processing unit 7 extracts words representing program characteristics from the program name and program description.

  When the program feature keyword representing the feature of the program is already included in the EPG data 3, the word may be stored in the program guide data storage unit 8 as a word as it is. This program feature keyword has a data frame in which keywords are stored in the EPG data 3 in the same manner as the names of performers are stored. It is what has been. Therefore, the keyword extracted from the EPG data 3 by the EPG data processing unit 7 is different. If the program feature keyword is not included, the text data may be subjected to morphological analysis, and only words having nouns from the decomposed words may be stored. Alternatively, a list of words to be stored in advance may be created, and only words that match the list may be stored.

  FIG. 3B shows an example of the program guide data stored in the program guide data storage unit 8 by the above processing in the EPG data processing unit 7. This program guide data stores drama as a genre, title A as a title, actor B and actress C as performers, and word D and word F as words (program feature keywords). In addition, the start date and time, the end date and time, and the broadcast station name are stored together so that the program can be identified. The information for identifying the program is not limited to these, and may be a G code or a program ID.

  In the present embodiment, the above-mentioned title, performer, genre, and word are used as the keyword type (see FIG. 4) in the user data in the user data storage unit 12, but the present invention is not limited to these. . For example, the performers may be further subdivided into moderators, commentators, and screenplays.

  In the present embodiment, the viewing information 1 is history information of a program viewed by the user, but it may be information of a program reserved for recording or information of a program designated as a user's favorite program. The program designation method in this case is not particularly limited, and may be, for example, a method of registering with an input button while the user is viewing or a method of registering with an input button while viewing the EPG data 3. Also, when viewing content in which text data that describes the content content is added to the content data, the text data is handled in the same way as the EPG data 3, so that the data in the text data can be given user preference. It can be stored in the user data storage unit 12 as a keyword to be shown.

  When the viewing information 1 including the viewing history information of the user is input, the program guidance data reference unit 9 refers to the program guidance data in the program guidance data storage unit 8 regarding the user viewing program indicated by the viewing information 1. The keyword of the user viewing program is extracted. Specifically, from the program guide data storage unit 8, based on the viewing information 1, a keyword of a program whose start date / time, end date / time, and broadcast station match the user viewing program, for example, title, performer, genre, and word are extracted. .

  The genre identification unit 10 identifies the genre of the user viewing program based on the information extracted by the program guide data reference unit 9. The user data storage unit 12 stores user data shown in FIG. As this user data, keywords are stored by genre and keyword type, and the number of times of viewing (viewing frequency: first evaluation value) and the last viewing date and time of the keyword (program including the keyword) are stored. Therefore, the user data updating unit 11 uses the keyword in the user data storage unit 12 for each keyword type identified by the genre identification unit 10 and for each keyword extracted from the user viewing program by the program guide data reference unit 9. Update attribute data for. In FIG. 4, keyword attribute data includes viewing frequency (viewing frequency) and last viewing date and time.

  Specifically, the user data update unit 11 searches the user data storage unit 12 for a keyword extracted for each keyword type within a genre that matches a new user viewing program, and is already stored in the user data storage unit 12. If it is, the number of viewing times (viewing frequency) is incremented by 1 and the last viewing date and time is updated. In addition, when the keyword is not stored in the user data storage unit 12, the user data storage unit 12 provides a new table, registers the keyword name, registers the number of times of viewing as one time, and registers the last viewing date and time. To do.

  When the user data storage unit 12 updates the keyword attribute data of the user viewing program in the user data storage unit 12, if there is a limitation of the notation information, the user data update unit 11 updates the attribute data only for the keyword that satisfies the notation information condition. . The restriction of the notation information is considered to be meaningful in the order / form of the notation of the keywords appearing in the text information of the EPG data 3, and whether or not to update the keyword attribute data, Whether or not to change the addition value is predetermined for each genre.

  For example, when the genre is drama, the performer indicated by the EPG data 3 is described first in the lead, and is often described in the order of important performers. An important performer may be a performer with a long appearance in a drama or a popular performer. In addition, since the degree of preference for the performer or the important performer is also high, the user may limit the notation order so that the priority is higher for the performers in the higher order.

  Further, for example, when the genre is drama, the program guide data reference unit 9 extracts only up to Nth performers having the notation order among the performers as keywords, and the user data update unit 11 uses the keywords for the keywords. The user data in the data storage unit 12 may be updated. The Nth and subsequent performers are usually not extracted as keywords because the user's preference is considered to be low. Thus, it is possible to solve the problem that the program is recommended based on performers that the user does not like very much.

  This point will be specifically described. Although the user likes the actor A who is a star of a certain drama X, the user also watches another drama Y. Actor B appears in both Drama X and Drama Y. In this case, the viewing frequency is higher for actor B than for actor A, and the preference point for actor B may be higher. Here, when there is a drama P in which the actor B appears and a drama Q in which the actor A appears, the drama P is higher in the recommendation order in the method of extracting the recommended program based only on the viewing frequency. However, in this program recommendation device, for example, the program guide data reference unit 9 sets the value of N so that it is limited to the main character, so that the preference point of the actor A becomes higher. Can be placed higher in the recommendation order.

  In this way, in the case of dramas and movies, it is considered that the starring person among the performers is more likely to like the user, but there are more dramas and movies in the supporting role than the starring person. Likely to appear in Therefore, in the method of extracting a recommended program based only on the viewing frequency, when the user is watching a plurality of dramas and movies, the supporting role is more frequent than the starring person. Recommendations that match your preferences can no longer be made. The configuration of the present embodiment solves such a problem and makes it easy to recommend a program suitable for the user's preference.

  Note that the above method is not limited to the case where the keyword whose genre is drama is a performer, but can also be applied to other keywords in other genres.

In addition to the above, the configuration for obtaining the above-described function may be a configuration in which a value added as the number of viewing times (hereinafter referred to as a keyword addition value) is weighted as described below. That is,
Keyword addition value = (view count) x (weight)
It is.

In this method, the weights are heavier for keywords whose order is higher in the program guide data (EPG data 3). For example, the first weight in the notation order is 1.0, and the second and subsequent weights are decreased by 0.3. Therefore, the keyword addition value is 1 × 1.0 = 1.0 when the notation order is first.
In the second case, 1 × (1.0−0.3) = 0.7
In the third case, 1 × (1.0−0.3 × 2) = 0.4
It becomes. Also in this method, the attribute data (viewing frequency) is updated with the keyword addition value only for the keyword whose keyword addition value is equal to or greater than the arbitrary value N in the user data of FIG.

  Further, regarding a case where a program having program guide data as shown in FIG. 5 is viewed, a difference in keyword addition value between the case where the weighting is performed and the case where the weighting is not performed will be described.

In the case of the program shown in FIG. 5A, the first actor B in the notation order has a weight of 1, and the actors C and D in the second notation order decrease the weight by 0.3. Therefore, the keyword addition value is
In the case of actor B, 1 x 1.0 = 1.0
In the case of actor C, 1 × (1.0−0.3) = 0.7
In the case of actor D, 1 × (1.0−0.3 × 2) = 0.4
It becomes. Similarly, in the case of the programs in FIGS. 5B and 5C, the first actor A in the notation order has a weight of 1, and the actors E and D in the second notation order have a weight of 0. Decrease by 3. Therefore, the keyword addition value is
In the case of actor A, 1 × 1.0 = 1.0
In the case of actor E, 1 × (1.0−0.3) = 0.7
In the case of actor D, 1 × (1.0−0.3 × 2) = 0.4
It becomes.

  Therefore, when the user views all the programs shown in FIGS. 5A to 5C, the keyword addition value of actor A is 2, the keyword addition value of actor D is 1.2, and the keyword addition value of actor A Is bigger. The larger the keyword addition value, the higher the preference point. Therefore, the program in which the actor A appears is preferentially recommended over the program in which the actor D appears.

  On the other hand, in the above-described method without weighting, the number of times of viewing of the performers is evenly added. Therefore, when all the programs in FIGS. 5A to 5C are viewed, the keyword addition value of the actor A is 2, The keyword addition value of actor D is 3, and the keyword addition value of actor D is larger. Therefore, the program in which actor D appears is preferentially recommended. In this case, since the actor D is a supporting role, there is a high possibility that the degree of preference of the user is low, and even if a program on which the actor D appears is recommended, it often does not match the preference of the user. In this way, if the method does not perform the above weighting, the number of times that the supporting role is viewed will increase, whereas if the method performs the above weighting in consideration of the order of the keywords, such problems will be solved. Is possible.

  By defining the calculation condition of the keyword addition value based on the keyword notation order as described above for each genre, it is possible to construct user data in consideration of the tendency of the keyword notation order by genre.

  Regarding the restriction on the keyword addition value, in addition to the restriction based on the notation order as described above, the restriction based on the additional information of the text information can be adopted. For example, if the EPG data 3 includes a tag that distinguishes between a performer and a guest performer among performers, the user is more likely to prefer a guest than a presenter in a music program. Since it is possible, the weight of the presenter may be reduced. Also, if a music program host and a variety program moderator are present, the weight may vary depending on the genre, for example, when the same moderator may be more likely to prefer the moderator of the variety program. It's okay. In this way, when additional information such as a tag that can identify the presenter and guest performer is added to the EPG data 3, the additional information is used to set a weight for each genre, thereby making it more user-friendly. It is possible to construct user data that matches the degree.

  If there is a keyword to be newly stored in the user data storage unit 12 and the memory capacity for storing the keyword in the user data storage unit 12 is already full, the stored keyword is deleted, Store new keywords. The keyword deletion condition, which keyword is deleted, varies depending on the genre.

  Usually, the keyword with the smallest number of views or the keyword with the oldest last viewing date is deleted. In this case, the maximum number of days for storing a keyword whose current date is deleted from the last viewing date and time is determined in advance for each genre.

  In the case of a continuous drama, since one drama is usually in units of three months, the maximum number of days for storing keywords is 90 days, and keywords whose viewing date and time are not updated for 90 days or more are targeted for deletion. In the case of a series drama that ends in three months, it is likely that interest will often be lost about matters related to the drama when the drama ends. Therefore, by limiting the maximum storage days in this way, it is possible to delete the old history, give priority to the new history, and recommend a program based on the history.

  In this way, setting the maximum number of days to be stored can cope with a problem that such a program is continuously recommended even though the program has lost interest such as a temporary preference. .

  Or, since a drama or a movie that is not a continuous drama such as a 2-hour drama is considered to have a preference that lasts longer than a continuous drama, a longer number of days may be set as the maximum storage day. With these settings, it is possible to accumulate user preferences and make long-term recommendations for serialized dramas and movies that consist entirely or partially of the same title, completely or partially the same performer. become.

  In this way, in contrast to the method in which the maximum storage days are set to be the same for keywords of any genre, in this program recommendation device, the maximum storage days of keywords for each genre should be set so that they remain longer. It is possible to delete a history that is better to leave a good history and not to leave longer like a temporary preference.

  The data storage method in the user data storage unit 12 when the viewing history is input has been described above. Next, a method for calculating the preference points from the viewing frequency in the user data storage unit 12 will be described.

  It can be determined that the preference point given to a keyword is that the user likes the keyword as the point is high, and the point increases as the viewing frequency increases. When calculating preference points for a keyword, in addition to the type of keyword and the frequency of viewing the program containing the keyword, the number of days from the date of viewing to the present, the viewing time of the program, the number of broadcasts of programs containing the same keyword, etc. May be used as a parameter. Further, the preference points are calculated by performing processing such as weighting on each parameter.

  When the number of days from the day when the program is viewed to the present is used as a parameter, the preference point is set to be larger as the number of days until the present is shorter. When the viewing time is used as a parameter, the viewing time is stored in the user data storage unit 12 so that the preference point increases as the viewing time increases. Further, the ratio of the time actually viewed by the user during the original broadcast time of the program may be used as a parameter. In this case, the preference point increases as the ratio increases. When the program broadcast count is used as a parameter, the program broadcast count is also stored in the user data storage unit 12, and the preference point increases as the ratio of how many times the program is viewed is increased.

  Here, a case where only the viewing frequency is used as a parameter will be described as an example of a preference point calculation method. The preference point has the same value as the viewing frequency. Referring to the user data in FIG. 4, the preference point f (drama, title A) for which the genre is drama and the keyword is title A is 3, the preference point f for the genre is drama and the keyword is word D (drama, word D). Is 80, the genre is sports, and the preference point f (sports, word D) is 1 for the keyword D. Although word D appears in both genres of drama and sports, preference points differ depending on the genre.

  For a keyword having a preference point of a certain keyword having a preference point equal to or greater than a predetermined threshold in each of a plurality of genres, a value obtained by adding the preference points may be used as a preference point of each genre.

  For example, if the preference point f (1, A) of the keyword A of genre 1 is 30, the preference point f (2, A) of the keyword A of genre 2 is 40, and the threshold is 30, the preference point f (1, A) And the preference point f (2, A), 70, which is a value obtained by adding together the preference points f (2, A), is a preference point for each of the keywords A and B. That is, the value f (1, A) = f (2, A) = 70. As a result, the preference points of the keyword A in the genres 1 and 2 are increased, and a program including the keyword A in the EPG data 3 is easily recommended.

  In this method, preference points are increased by collecting point values of keywords with high preference points, and programs including the keywords are easily recommended. This method is effective when the keywords of genres that add preference points are similar.

  For example, in the genre of movie and drama, both performers are often actors, and the user is likely to watch even if the performers who often see in the movie appear in the drama. Therefore, it is more effective in recommending a program suitable for the user's preference if the preference points of performers with high preference points in both movies and dramas are added together.

  A keyword having a preference point equal to or greater than a predetermined threshold in a certain genre may use the value of the preference point in another genre. For example, when the preference point of keyword A in genre 1 exceeds a predetermined threshold and the preference point of keyword A in genre 2 is a lower value, the preference point of keyword A in genre 2 is selected as the genre. Replace with the value of 1. As a result, a program including the keyword A is easily recommended in the genre 2 as well.

  In this way, the rules for adding preference points of the same keyword between a plurality of genres or replacing the preference points of the same keyword between a plurality of genres between those different genres are only between predetermined specific genres. May be done in

  Further, it is possible to set different conditions for calculating preference points for each genre. For example, in a certain genre, a preference point may be set by setting only a keyword with the highest viewing frequency as valid, while a preference point is set to 0 for other keywords.

  Regarding this technique, for example, a case where the genre is baseball and the team name is extracted as a keyword of words will be described. When the user likes team A, the game frequency of team A is viewed, so the viewing frequency is high. However, although other teams (Team B, Team C, Team D) that compete with Team A are not as much as Team A, the viewing frequency will gradually increase. If this happens, the preference points of other teams will also rise, and there is a possibility that a game in which team B plays against team C will be recommended.

  In this case, it is possible to give preference points only to the team A with the highest viewing frequency and set the preference points of other teams to 0 so that only the game of the team A is recommended.

  In the above method, the preference points of only the keyword with the highest viewing frequency among the keyword types having the same genre are calculated, and the preference points of other keywords are set to zero. On the other hand, if the team name can be identified, such as registering all team names (for example, professional baseball team names) in advance, only the team with the highest viewing frequency among the team names, not the same keyword type Preference points may be calculated, and the preference points of other team names may be set to zero. In this way, by storing keywords for each genre, it is possible to set conditions for calculating preference points for each genre.

In addition, when calculating the preference points of each keyword, it is possible to set the weighting coefficient applied to the parameter value used for calculating the preference points for each genre. For example, the case where the number of viewing times and the number of days from the viewing date to the present time are set as parameters will be described. In this case, the preference point f of the keyword A in the genre α is
f (α, A) = X × g (X, α) + Y × g (Y, α)
However, it can be expressed as X: number of times of viewing, Y: number of days from the date of viewing to the present, g (x, y): weight when parameter x is genre y.

  In this case, the number of views and the last viewing date are recorded as a history for one keyword, and the preference points are determined by the parameter values. An example of calculating preference points using this method will be described below.

  For example, when the drama A is viewed every time and the keyword A1 included in the drama A is stored, if the drama A is viewed every time, the number of times of viewing increases and the preference point of the keyword A1 increases.

  On the other hand, when the broadcast of the drama A ends and the user starts viewing another new drama B, the preference point of the keyword B1 included in the drama B increases. However, the preference point of the keyword B1 is expected to remain lower than the preference point of the keyword A1, since the number of times of watching the drama B is low at first.

  In such a case, it is considered that selecting the recommended program for the user matches the user's preference based on the preference estimated from recent viewing data rather than the number of viewing times. Therefore, the preference point is adjusted by applying a weight so that the preference point becomes smaller as the number of days from the viewing date to the present time increases. The degree of weighting at this time may be changed depending on the genre. Thus, when the weight is set for each genre, the program can be selected with high accuracy in recommending the program that matches the user's preference. The reason for this is that, as described below, the degree of appearance of a keyword varies depending on the genre, and there are long-lasting preferences and rapidly changing preferences.

  For example, in the case of a continuous drama, broadcast programs are often changed every three months. While the interest in the drama A keyword is maintained during the drama A broadcast period, the interest in the keyword tends to decrease when the drama broadcast period ends. In order to cope with such a rapidly changing preference, it may be determined that the user has lost interest in the drama A when at least three months have passed since the last viewing date by the user, and weight may be applied so that the preference point becomes zero. .

  In addition, in the case of dramas and movies that have been made into series such as 2-hour dramas, the next series may be broadcast after 3 months. Therefore, this may be regarded as a long-lasting preference, and may be set so that the preference point is not so small even if the number of days from the last viewing date to the present is large.

  Similarly, for example, guest performers of a music program such as a singer do not frequently appear and thus the number of times of viewing does not increase, but the user's preference may be strong. In such a case, the preference may be set so that the preference point is not so small even if the number of days from the last viewing date to the present is large, considering the preference as lasting.

  The history itself may be deleted depending on the number of days elapsed from the last viewing date to the present. For example, when there is a limit to the number of histories for storing keywords, the number of days that keywords can be stored is limited based on the number of days from the last viewing date to the present date. The number of days that can be stored may be set according to the genre of the program from which the keyword is extracted.

  With regard to the above method, for example, the maximum storage period of the keyword A extracted from the sports genre program is 50 days, and the maximum storage period of the keyword B extracted from the drama genre program is 30 days.

  As described above, when the weight applied to the parameter for calculating the preference point of the keyword is changed according to the genre, in recommending the program to the user, the appearance degree of the keyword different for each genre, and the user's It becomes possible to respond to the degree of interest.

  Next, a method for calculating the degree of coincidence with the user's preference for each program using the keyword preference points obtained as described above will be described. A program with a higher degree of match is considered to be more likely to match the user's preference.

  The degree of coincidence includes the number of keywords stored in the user data storage unit 12 in the EPG data 3 of each program that can be viewed after the current date and time, and the number of preference points for these keywords. It is calculated by.

  The program for which the degree of match is calculated is a program that can be viewed after the current date and time, but the user or system designates the range of the program for which the degree of match is to be calculated, and all programs within that range are specified. The degree of match may be calculated for and a program with a high degree of match may be recommended.

  In the program guide data storage unit 8, keywords included in the program as program guide data are stored for each program. Further, as shown in FIG. 4, the user data storage unit 12 stores preference points for each keyword for each genre. The degree-of-match calculation unit 15 refers to the preference points for each keyword stored in the user data storage unit 12, and calculates the degree of match with the user's preference for the program stored in the program guide data storage unit 8. . In this case, the preference point of the keyword referred from the user data storage unit 12 is a genre that matches the genre of the program guide data stored in the program guide data storage unit 8. Note that the preference point is set to 0 for keywords that are not stored in the user data storage unit 12.

  As a method for calculating the degree of match, simply, the sum of preference points for each keyword included in the EPG data 3 of each program may be used as the degree of match. Alternatively, the average value of preference points may be calculated for each keyword type, and the total may be used as the degree of match. Here, when there is program guide data as shown in FIG. 3B and user data as shown in FIG. 4, a method for calculating the degree of match by calculating the average value of preference points for each keyword type will be described. To do.

  Since the genre of the program guide data shown in FIG. 3B is drama, the coincidence degree calculation unit 15 refers to the drama data of the genre from the user data storage unit 12. In the user data of FIG. 4, when the genre is drama, there are a title, a performer, and a word as keyword types. Therefore, an average value of preference points for each keyword type is obtained. Regarding the keyword title A in the program guide data, the preference point of the keyword A in the user data is 6, so the average value of the preference points of the keyword type title is 6. Further, regarding the actor B and the actress C of the keyword in the program guide data, the preference point of the actor B in the user data is 7 and the preference point of the actress C is 3, so the average value of the preference points of the performers with the keyword type is 5

  Further, regarding the word D and the word F of the keyword in the program guide data, since the preference points of the word D and the word F in the user data are both 80, the average value of the preference points of the word of the keyword type is 80.

Therefore, regarding the program guide data in FIG. 3B, the degree of match obtained from the total of the average values of the keyword types is
6 + 5 + 80 = 91
It becomes.

  In the above processing, a preference point may be set for available preference points to limit the preference points to be used. For example, when the threshold is set to 5, a keyword having a preference point of 5 or less is not a valid keyword. In this case, the average value of the preference points of the performer whose keyword type is the result is 7 as a result of being calculated only from the preference points of the actor B.

  In this way, by setting a threshold for available preference points, keywords that have small preference points and are difficult to determine as keywords preferred by the user are excluded. As a result, only the keywords that are likely to be preferred by the user are used. Can be calculated. The threshold value may be set to a different value for each genre or for each keyword type.

Similarly, when there is program guide data shown in FIG. 3C, the degree of coincidence is calculated. Since the genre of the program guide data in FIG. 3C is sports, the user data storage unit 12 refers to user data whose genre is sports. Since the preference point of the keyword title G is 6, the average value of the preference points of the keyword type title is 6. Since there are no performers, the average value of preference points for performers with keyword type is 0. Since the preference point for the word D of the keyword is 1 and the preference point for the word E is 95, the average value of the preference points for the keyword type is 48. Therefore, the degree of match of the program guide data shown in FIG.
6 + 0 + 48 = 54
It becomes. Since the degree of match is lower than that of the program guide data in FIG. 3B, the program corresponding to the program guide data in FIG. 3B is preferentially recommended.

  Here, as a comparative example, the calculation result of the degree of matching based on the preference point of the keyword not considering the genre is shown as a comparative example with respect to the calculation result of the degree of matching based on the keyword preference point considering the genre in the program recommendation device. As for the comparative example, as in the case described above, the user data of FIG. 4 is used, and the degree of match is obtained by obtaining an average value for each keyword type for the program guide data of FIGS. 3B and 3C. Is calculated. Here, the viewing frequency in the user data in FIG. 4 is used as a preference point as it is.

In the program guide data of FIG. 3B, the preference point for the keyword type is “6” (genre: drama) for title A only. The preference point of the performer with the keyword type is 7 (genre: drama) for actor B and 3 (genre: drama) for actress C, so these are 5 on average. As for the preference point of the word type, the preference point of the word D is 80 (genre: drama) and 1 (genre: sport), and the preference point of the word F is 80. 80.5. Therefore, the degree of match of the program corresponding to the program guide data in FIG.
6 + 5 + 80.5 = 91.5
It becomes.

Similarly, in the program guide data of FIG. 3 (c), the preference point of the title of keyword type is 6 (genre: sport) of title G only. The preference point of the performer with the keyword type is 0. Since the preference point of the keyword type is the word, the preference point of the word D is 81 (genre: drama) and 1 (genre: sport), and the preference point of the word E is 95 (genre: sport). These are averaged to be 88. Therefore, the degree of match of the program corresponding to the program guide data in FIG. 3C is 6 + 0 + 88 = 94.
It becomes.

  The degree of matching in this comparative example is higher in the program guide data (genre: sport) in FIG. 3C than in the program guide data (genre: drama) in FIG. 3B. However, the preference points of the word D included as keywords in both the program guide data are actually obtained by watching the drama, and the preference points for the genre sports are low. Yes. Therefore, contrary to the calculation result of the degree of match in the comparative example, the program guide data (genre: drama) in FIG. 3 (b) is better than the program corresponding to the program guide data (genre: sport) in FIG. 3 (c). It can be said that the recommendation of the program corresponding to is in accordance with the user's preference. As is clear from this description, selecting a recommended program based on a keyword preference point for each genre can recommend a program that matches the user's preference.

  In calculating the degree of match, an appropriate weight may be applied to each keyword type. For example, when the weight G is given to the title, 3 × G = 3G is the average value of the title. When the title is weighted, the degree of matching of the programs that the user has viewed in the past increases. On the other hand, if a weight is given to the performer, it is possible to recommend the performer with emphasis.

  In this case, the weighting method may be changed for each genre. For example, in the case of a continuous drama, there is a high possibility that the program with the same title will be repeatedly viewed, so that the title may be weighted. In the case of a movie, since the titles are various, it is possible to recommend a movie in which an actor that a user often sees or a content that is often viewed is recommended by giving weight to performers and words.

  Further, even if a keyword having a preference point equal to or greater than a predetermined threshold is in another genre, the preference point of the keyword may be used when calculating the degree of match. For example, when the program guide data having the genre P includes the keyword A, the preference point of the keyword A in the genre P of the user data storage unit 12 is low, and the preference point of the keyword A in the genre Q is greater than or equal to a predetermined threshold. In some cases, the degree of match may be calculated using the preference point in genre Q as the preference point for keyword A. This makes it easier to recommend programs that include keywords with high preference points.

  When the degree of coincidence calculation unit 15 calculates the degree of coincidence of each program as described above, the degree of coincidence data is output to the display information generation unit 16. The display information generation unit 16 creates display data based on the input match degree data and outputs the display data to the display unit 17. The display unit 17 displays the input display data on the screen. In this case, the programs may be displayed in descending order of match, or the match may be displayed on the program guide.

  Next, the flow of user data creation processing (user data update processing) and program recommendation processing in the program recommendation device according to the present embodiment will be described with reference to the flowcharts of FIGS. As a premise of these processes, in the program recommendation device, when the EPG receiving unit 6 receives the EPG data 3, the EPG data processing unit 7 extracts keywords from the EPG data 3, and the program guide data storage unit 8 The keywords extracted by the data processor 7 are stored as program guide data for each program genre and for each keyword type. The program guide data stored in the program guide data storage unit 8 by this processing is, for example, as shown in FIG. Further, the number of times of viewing (viewing frequency) is used as a parameter for calculating the preference points.

  As shown in FIG. 6, when viewing information 1 indicating history information of a program viewed by the user or recommendation request information 2 from a system including the program recommendation device is input (S101), the program recommendation device is S102 or later. Enter the process.

  In S102, when the input information is viewing information 1, user creation processing after S103 is performed, and when the input information is recommendation request information 2, program recommendation processing is performed (S121).

  In S103, the program guide data reference unit 9 refers to the program guide data stored in the program guide data storage unit 8 corresponding to the program (user viewing program) viewed by the user indicated by the viewing information 1, Extract keywords.

  Next, the genre identifying unit 10 identifies the genre of the user viewing program with reference to the program guide data (S104).

  Next, the user data update unit 11 updates the keyword attribute data stored in the user data storage unit 12 for the keywords extracted by the program guide data reference unit 9. In this case, the user data update unit 11 determines whether or not notation information restrictions are set (S105).

  The restriction of the notation information is considered to be significant in the order / format of the notation of keywords appearing in the EPG data 3, and the number of times the keyword is viewed in the user data stored in the user data storage unit 12 (viewing frequency). Whether to update the attribute data and whether to change the value when adding the number of times of viewing (viewing frequency) to the keyword is predetermined for each genre of the program.

  When the restriction of the notation information is set, the user data update unit 11 updates the user data in the user data storage unit 12 only for keywords that do not correspond to the restriction of the notation information (satisfy the notation information condition) (S106 to S106). S114).

  Here, as a limitation of the notation information, only the performers (keywords) whose notation order in the EPG data 3 is the third from the front are subject to update with respect to the keywords whose genre is drama and whose keyword type is performer. Is set.

  In S106, the user data update unit 11 extracts keywords that satisfy the notation information condition when the restriction of the notation information as described above is set, and if the restriction is not set, Extract keywords. Next, the user data update unit 11 determines whether or not the genre of the user viewing program is a drama, and in the case of a drama, checks whether or not a performer is stored, and if the number of performers is three or less All performers (keywords) are extracted. On the other hand, if the number of performers exceeds 3, only the third performer (keyword) from the front is extracted.

  In such a process, the fourth and subsequent performers are supporting roles and are not the user's preference, and are not reflected in the user data, so that only the performers who are likely to like the user are stored as user data. To do. All keywords other than performers are extracted.

  Next, the user data update unit 11 determines whether or not there is an unprocessed keyword among the extracted keywords (S107), and if there is no unprocessed keyword, the process ends. On the other hand, if there is an unprocessed keyword, one of the keywords is extracted together with the keyword type information (S108), and it is determined whether or not the keyword is already stored in the user data storage unit 12 (S109). ). When referring to the user data in the user data storage unit 12, the storage location of the same genre as the genre of the user viewing program is searched, and then the storage location with the same keyword type is searched, and the search matches the same Search for keywords.

  In the above determination, if the keyword has already been stored in the user data storage unit 12, the number of viewing times (viewing frequency) is incremented by 1 (S110), the last viewing date is updated to the viewing date and time (S111), and S107. Return to.

  In S109, when the keyword extracted in S108 is not stored in the user data storage unit 12, it is determined whether or not the keyword can be stored in the user data storage unit 12 (S112). This determination is made based on, for example, an allowable amount of memory that stores the user data storage unit 12.

  As a result of the above determination, if the keyword can be stored in the user data storage unit 12, the keyword is stored in association with the same genre and the same keyword (S114), the number of viewing times (viewing frequency) is set to 1, and the last viewing date Is the viewing date and time.

  On the other hand, in S112, if a keyword cannot be stored due to the limit of the memory capacity, one stored keyword is deleted (S113). The keyword to be deleted is extracted based on the number of elapsed days from the last viewing date to the current date. In this case, the maximum number of storage days from the last viewing date, which determines how many days the elapsed days from the last viewing date to the current date are allowed to be deleted, is determined in advance for each genre. Therefore, in the process of S113, for each keyword stored in the user data storage unit 12, a keyword exceeding the maximum storage days from the last viewing date is searched, and if there is a corresponding keyword, the keyword is deleted. If there is no corresponding item, the keyword that has the number of times of viewing only once and is closest to the maximum number of storage days is deleted.

  Next, the flow of the program recommendation process when the input information is the recommendation request information 2 in S102 will be described based on the flowchart of FIG.

  In the program recommendation device, the preference points are calculated from the viewing frequency of each keyword in the user data storage unit 12 in S201 to S208.

  In this case, the preference point calculation unit 14 determines whether or not there is an unprocessed keyword in the user data in the user data storage unit 12 (S201), and if not, proceeds to the processing after S208. On the other hand, if there is an unprocessed keyword, one of them is extracted (S202).

  Next, when calculating the preference points for each keyword, the preference point calculation unit 14 sets the weighting coefficient to be applied to the parameter value to be used for each genre, with respect to the keyword genre extracted in S202. Reference is made to a preset weight (S203). Here, since only the number of times of viewing (viewing frequency) is used as a parameter, the preference point is (number of times of viewing) × (weight of the corresponding genre).

  Next, the preference point calculation unit 14 refers to conditions for calculating preference points set for each genre (S204), and calculates preference points for keywords (S205).

  Here, for example, it is assumed that a preference point for only the keyword with the highest viewing frequency in the baseball genre is calculated, and a condition for setting the preference point of other keywords to 0 is set. In this case, if the team name can be identified, such as registering all the team names in advance, the preference points of only the team with the highest viewing frequency are calculated, not the same keyword type, Other team name preference points may be set to zero.

  Next, the preference point calculation part 14 sets the threshold value to a preference point, and when the restriction | limiting which restrict | limits the preference point used at the time of coincidence calculation is set, the preference point calculated in S205 exceeds a threshold value Whether or not the preference point is not exceeded (S207).

  When the preference points are calculated for all keywords in the user data in the user data storage unit 12 by the preference point calculation unit 14 as described above, the process proceeds to S208 and subsequent steps.

  S208 is a process when a keyword having a preference point equal to or higher than a threshold value in a certain genre is set for another genre using a condition for calculating the degree of match using the value of the preference point. In this case, in S208, first, all the keywords in the user data in the user data storage unit 12 are searched to check whether there is a keyword having a preference point A exceeding the threshold. Next, when there is a keyword exceeding the threshold, the presence / absence of the keyword in another genre is searched. When the same keyword exists in another genre and the preference point B of the keyword is smaller than the preference point A, the preference point B is replaced with the value of the preference point A.

  In S209 and subsequent steps, a process for calculating the degree of match with the user's preference for each program and displaying on the display unit 17 in order from the program with the highest degree of match will be described.

  The degree-of-match calculation unit 15 checks whether there is unprocessed program guide data for the program guide data stored in the program guide data storage unit 8 (S209), and if there is, extracts one of them (S209). S210).

  Next, the matching degree calculation unit 15 refers to the program guide data storage unit 8 to identify the genre of the extracted program guide data (S211), and further extracts keywords included in the program guide data (S212). ).

  Next, the degree-of-match calculation unit 15 searches the user data storage unit 12 for a keyword of the same genre as the genre of the program guide data extracted in S211 and refers to the preference point of the keyword extracted thereby to determine the degree of match. Calculate (S213). In this case, keywords are stored for each genre in the user data in the user data storage unit 12, and the degree-of-match calculation unit 15 obtains genre data (preference points) that matches the genre of the program extracted in S211 from the user data. Refer to and calculate the degree of match. Therefore, it is possible to calculate the degree of match with respect to the user preference for each genre for the program guide data stored in the program guide data storage unit 8.

  Note that preference points for keywords that do not exist in the user data storage unit 12 are calculated as zero. In addition, when applying weight to a keyword type, it is determined in advance which keyword type and how much weight is applied to which keyword type, and the degree of matching is calculated accordingly.

  The above processing is performed on each program guide data in the program guide data storage unit 8, and when there is no unprocessed program guide data, the display information generating unit 16 extracts program guide data having a high degree of matching (S214). . The display unit 17 displays the program guide data extracted by the display information generation unit 16 as a recommended program. In this case, the degree of match may be given to a program guide (for example, EPG) and displayed.

  As described above, in the present program recommendation device, when the viewing frequency of the keyword extracted from the EPG data 3 of the viewed program is stored in the user data storage unit 12, it is stored for each genre, so that each program is recommended for each genre at the time of program recommendation. It is possible to make a recommendation in consideration of the viewing tendency, whereby a program that matches the user's preference can be recommended with high accuracy.

  In addition to the frequency for each genre, the keyword may be frequency for each item such as a day of the week or a time zone. In this case, information on the day of the week or time zone is extracted from the program guide data, and the frequency is taken for each day of the week or each time zone. The frequency may be further set for each genre in each day of the week or time zone. The day of the week is estimated from the date information, and the time zone is 06: 00-12: 00 in the morning, 12: 00-18: 00 at noon, 18: 00-23: 00 at night, 23: 00- The next day 06:00 should be set in advance, such as midnight.

  It is also possible to collect a plurality of genres into one group and take the frequency for each group. In this case, grouping may be independently performed using information in the EPG text information, such as group 1 when a word is included in the EPG text information, and group 2 when another word is included.

  The present invention is not limited to the above-described embodiments, and various modifications are possible within the scope shown in the claims, and embodiments obtained by appropriately combining technical means disclosed in different embodiments. Is also included in the technical scope of the present invention.

  The program recommendation device of the present invention can be applied to devices such as a recording device, a television, and a personal computer that store user viewing data using EPG. In these devices, the user recommendation is stored from the viewing data. A program is recommended using the accumulated preference data. In this case, since the viewing data is accumulated by genre, it is possible to make a recommendation in consideration of the viewing tendency for each genre, and to recommend a program that more matches the user's preference.

It is a block diagram which shows the functional structure of the program recommendation apparatus in one Embodiment of this invention. It is a block diagram which shows the structure of the hardware of the program recommendation apparatus shown in FIG. 3A is an explanatory diagram of the EPG data received by the program recommendation device shown in FIG. 1, and FIG. 3B is the program guide data extracted from the EPG data in FIG. 3A and shown in FIG. FIG. 3C is an explanatory diagram illustrating an example of program guide data stored in the storage unit, and FIG. 3C is an explanatory diagram illustrating an example of other program guide data. It is explanatory drawing which shows an example of the user data stored in the user data storage part shown in FIG. 5A is an explanatory diagram showing still another example of the program guide data extracted from the EPG data and stored in the program guide data storage unit shown in FIG. 1, and FIG. 3B is a diagram showing the program guide data. FIG. 3C is an explanatory diagram showing still another example, and FIG. 3C is an explanatory diagram showing still another example of program guide data. It is a flowchart which shows the flow of the user data creation process (user data update process) in the program recommendation apparatus shown in FIG. It is a flowchart which shows the flow of the program recommendation process in the program recommendation apparatus shown in FIG. It is explanatory drawing which shows the example of a display in a display part about the recommended program extracted in the display information generation part shown in FIG. It is explanatory drawing which shows an example of the user data stored in the conventional program recommendation apparatus.

Explanation of symbols

DESCRIPTION OF SYMBOLS 1 Viewing information 2 Recommendation request information 3 EPG data 7 EPG data processing part 8 Program guide data storage part (program guide data storage means)
9 Program guide data reference section (program data extraction means)
10 Genre identification part (program data extraction means)
11 User data update unit (user data creation means)
12 User data storage (user data storage means)
13 User data reference part 14 Preference point calculation part (recommended program determination means)
15 Matching degree calculation section (recommended program determining means)
16 Display information generating unit (recommended program determining means)
17 Display section (recommended program display means)

Claims (13)

Program data extraction for extracting, from the program guide data for the program specified by the user information, among the programs indicated by the program guide data for each receivable program, a plurality of items having different categories included in this data Means,
Based on the information indicated by the item extracted by the program data extracting means and user information, user data including an evaluation value for the user's preference for the related item group under the relationship between the plurality of items is created. User data creation means;
A program recommendation device comprising recommended program determination means for determining a recommended program based on an evaluation value of the user data. Program guide data storage means for storing program guide data for each receivable program;
Program data extracting means for extracting a keyword and a program genre included in the data from the program guide data for the program specified by the user information among the programs indicated by the program guide data;
Based on the information extracted by the program data extracting means and user information, the keyword, the genre of the program to which the keyword belongs, and the evaluation value for the preference of the user in the genre of the program to which the keyword belongs for the keyword User data creation means for creating user data;
User data storage means for storing the user data;
A program recommendation device comprising: a recommended program determining unit that determines a recommended program based on an evaluation value of each keyword of the user data and a genre of a program to which the keyword belongs.   The program recommendation device according to claim 1, further comprising a recommended program display unit that displays the recommended program determined by the recommended program determination unit.   The user information includes information indicating a program viewed by the user, and the evaluation value is an integrated value of the number of appearances of the keyword in the program guide data of the program viewed by the user. The program recommendation device described. The recommended program determining means includes evaluation value calculating means for calculating a second evaluation value based on the first evaluation value when the evaluation value is the first evaluation value, and the second evaluation value Based on the recommended program,
3. The program according to claim 2, wherein the evaluation value calculation unit weights the first evaluation value based on a notation order of keywords in the program guide data and calculates a second evaluation value. 4. Recommendation device. The recommended program determining means includes evaluation value calculating means for calculating a second evaluation value based on the first evaluation value when the evaluation value is the first evaluation value, and the second evaluation value Based on the recommended program,
The program guide data includes role information indicating a role in the program for performers of the program,
3. The evaluation value calculating unit calculates a second evaluation value by weighting a first evaluation value of a performer as a keyword based on the role information in program guide data. The program recommendation device described in 1. The recommended program determining means includes evaluation value calculating means for calculating a second evaluation value based on the first evaluation value when the evaluation value is the first evaluation value, and the second evaluation value To determine the recommended program based on
The evaluation value calculating means calculates the second evaluation value by weighting the first evaluation value, and changes the weighting amount according to the genre of the program to which the keyword belongs. 2. The program recommendation device according to 2. The recommended program determining means includes evaluation value calculating means for calculating a second evaluation value based on the first evaluation value when the evaluation value is the first evaluation value, and the second evaluation value To determine the recommended program based on
The evaluation value calculating means calculates the second evaluation value by making only the evaluation value for the keyword having the highest first evaluation value valid within the predetermined genre and invalidating the evaluation value of other keywords. The program recommendation device according to claim 2.   3. The program data extracting means extracts only performers up to a predetermined Nth ranking in the program guide data as keywords when the genre of the program is a drama or a movie. The program recommendation device described in 1. Program data extraction for extracting, from the program guide data for the program specified by the user information, among the programs indicated by the program guide data for each receivable program, a plurality of items having different categories included in this data Steps,
Based on the information indicated by the item extracted by the program data extraction step and the user information, user data including an evaluation value for the user's preference for the related item group under the relation between the plurality of items is created. User data creation step;
And a recommended program determining step of determining a recommended program based on the evaluation value of the user data. A program data extracting step of extracting a keyword and a program genre included in the data from the program guide data for the program specified by the user information among the programs indicated by the program guide data for each receivable program. When,
Based on the information extracted by the program data extraction step and the user information, the keyword, the genre of the program to which the keyword belongs, and the evaluation value for the preference of the user in the genre of the program to which the keyword belongs for the keyword A user data creation step for creating user data;
A program recommendation method comprising: a recommended program determination step for determining a recommended program based on an evaluation value of each keyword of the user data and a genre of a program to which the keyword belongs.   A program for operating the program recommendation device according to any one of claims 1 to 9, wherein the program causes a computer to function as each of the above means.   A computer-readable recording medium on which the program according to claim 12 is recorded.

Images