JP2007172523A - Information processor, information processing method, and program - Google Patents

Abstract

<P>PROBLEM TO BE SOLVED: To recommend content suitable for context designated by a user. <P>SOLUTION: A mapping part determines the coordinates of each context and each mood in a plane whose horizontal axis is a dimension 1 and whose vertical axis is a dimension 2 by using multi-variable analysis(for example, corresponding analysis) or the like based on the combination of context and mood evaluated for a plurality of contents, and performs mapping of each context and each mood on the same plane. Then, the mood corresponding to the context designated by a user is specified based on the generated context/mood corresponding plane, and the content having a specified mood is recommended to the user. The content such as a musical piece suitable for the context is applied to a system recommended to the user. <P>COPYRIGHT: (C)2007,JPO&INPIT

Description

  The present invention relates to an information processing device, an information processing method, and a program, and more particularly, to an information processing device, an information processing method, and a program that provide content corresponding to a user's context.

  For example, the user ’s context (time, time, season, temperature, humidity, location, event, etc.), like listening to a beach song during a summer drive or listening to a Christmas song at the end of the year. In the past, a person who wants to listen to a song (hereinafter referred to as a user) creates a playlist of songs suitable for the context, and a viewing history ( There is a method for generating a playlist based on the correspondence information between the actually viewed music and its context (see, for example, Patent Document 1).

  In addition, there is an example in which a music critic or the like determines a context suitable for a song and creates a database of the context as attribute information for the song (for example, MoodLogic, AllMusicGuide).

JP 2005-165632 A

  However, it is costly for a user, a music critic, and the like to add a context suitable for music by human power because it requires many people and time. Also, the choice of context by a person tends to be arbitrary.

  The generation of a playlist based on history information cannot cope with all assumed contexts until sufficient history information is collected.

  The present invention has been made in view of such circumstances, and it is possible to recommend content suitable for the context specified by the user by generating in advance information indicating the correlation between the user's context and the feature amount of the content. It is to make.

  An information processing apparatus according to an aspect of the present invention provides a feature amount of content in an information processing apparatus that recommends content suitable for the context of a recommendation condition using a context representing a user's situation when viewing the content as a recommendation condition. Correlation indicating the correlation between the feature quantity and the context by analyzing the combination of the feature quantity and the context input by the input means and the input means for inputting the combination of the context associated with the content A generating unit that generates information; a specifying unit that specifies a context as a recommendation condition; a specifying unit that specifies a feature quantity correlated with the context specified by the specifying unit based on the correlation information; and the specifying Content having the feature amount specified by the means And a recommendation means that the recommendation to.

  The generation unit analyzes a combination of the feature amount and the context input by the input unit, thereby obtaining a plurality of states of the feature amount and the context as correlation information indicating a correlation between the feature amount and the context. It is possible to generate a feature quantity-context correspondence space in which a plurality of states are mapped on the same space.

  The specifying unit may specify a feature amount having a distance close to the context specified by the specifying unit on the feature amount-context correspondence space.

  The same space and the feature quantity-context corresponding space may be two-dimensional planes.

  The input means may input a combination of a feature amount of the content and a context associated with the content obtained from an evaluator who viewed the content.

  The input unit may input a combination of a feature amount of the content obtained from a user's content viewing history and a context associated with the content.

  The input means may input a combination of the feature amount and the context detected by a predetermined sensor when the content is viewed.

  An information processing method according to an aspect of the present invention is an information processing method for an information processing apparatus that recommends content suitable for a context of a recommendation condition using a context representing a user's situation when viewing the content as a recommendation condition. The combination of the feature quantity and the context associated with the content is input, and the combination of the input feature quantity and the context is analyzed to generate correlation information indicating the correlation between the feature quantity and the context And specifying a context as a recommendation condition, specifying a feature quantity correlated with the specified context based on the correlation information, and recommending content having the specified feature quantity to the user.

  A program according to one aspect of the present invention is a program for recommending content suitable for a context of a recommendation condition, using a context representing a user's situation when viewing the content as a recommendation condition, and a feature amount of the content By inputting a combination of the context associated with the content, and analyzing the combination of the input feature quantity and the context, correlation information indicating a correlation between the feature quantity and the context is generated, and a recommendation condition A context is specified, and based on the correlation information, a feature amount correlated with the specified context is specified, and a process including a step of recommending the content having the specified feature amount to the user is executed on the computer Let

  In one aspect of the present invention, by analyzing a combination of a feature amount of a content and a context associated with the content, correlation information indicating a correlation between the feature amount and the context is generated. Then, based on the correlation information, a feature quantity correlated with the context designated as the recommendation condition is specified, and content having the specified feature quantity is recommended to the user.

  As described above, according to one aspect of the present invention, it is possible to recommend content suitable for the context designated by the user by generating in advance information indicating the correlation between the user's context and the feature amount of the content. It becomes.

  Embodiments of the present invention will be described below. Correspondences between constituent elements of the present invention and the embodiments described in the specification or the drawings are exemplified as follows. This description is intended to confirm that the embodiments supporting the present invention are described in the specification or the drawings. Therefore, even if there is an embodiment which is described in the specification or the drawings but is not described here as an embodiment corresponding to the constituent elements of the present invention, that is not the case. It does not mean that the form does not correspond to the constituent requirements. Conversely, even if an embodiment is described here as corresponding to a configuration requirement, that means that the embodiment does not correspond to a configuration requirement other than the configuration requirement. It's not something to do.

  An information processing apparatus according to one aspect of the present invention (for example, the content recommendation apparatus 1 in FIG. 1) recommends a content suitable for the context of a recommendation condition, using a context representing a user's situation when viewing the content as a recommendation condition. Input means (for example, the evaluation input unit 11 and the history information input unit 12 in FIG. 1) for inputting a combination of a feature amount (for example, mood) of the content and a context associated with the content; Generating means (for example, the mapping unit 13 in FIG. 1) that generates correlation information indicating a correlation between the feature quantity and the context by analyzing a combination of the feature quantity and the context input by the input means; , Designation means for designating a context as a recommendation condition (for example, the recommendation condition input unit 15 in FIG. 1) Based on the correlation information, a specifying unit (for example, the distance calculation unit 16 in FIG. 1) that specifies a feature amount correlated with the context specified by the specifying unit, and a feature amount specified by the specifying unit Recommending means (for example, the content extracting unit 17 and the content recommending unit 18 in FIG. 1) for recommending the content having the above to the user.

  An information processing method according to an aspect of the present invention is an information processing method for an information processing apparatus that recommends content suitable for a context of a recommendation condition using a context representing a user's situation when viewing the content as a recommendation condition. The combination of the feature amount and the context associated with the content is input (for example, step S1 in FIG. 5), and the combination of the input feature amount and the context is analyzed, whereby the feature amount and the context are analyzed. (For example, step S2 in FIG. 5), a context is specified as a recommendation condition (for example, step S41 in FIG. 9), and the specified context is correlated with the specified context based on the correlation information. The related feature quantity is identified (for example, step S42 in FIG. 9), and the identified feature quantity is identified. Content having recommended to the user (e.g., step S43 of FIG. 9) comprising the steps.

  A program according to one aspect of the present invention is a program for recommending content suitable for a context of a recommendation condition, using a context representing a user's situation when viewing the content as a recommendation condition, and a feature amount of the content The combination of the context associated with the content is input (for example, step S1 in FIG. 5), and by analyzing the combination of the input feature and the context, the correlation between the feature and the context is obtained. The correlation information is generated (for example, step S2 in FIG. 5), the context is specified as the recommendation condition (for example, step S41 in FIG. 9), and the feature has a correlation with the specified context based on the correlation information The amount is specified (for example, step S42 in FIG. 9), and the content having the specified feature amount is specified. Recommend Ntsu to the user (e.g., step S43 in FIG. 9) to execute a process including the step in the computer.

  Hereinafter, specific embodiments to which the present invention is applied will be described in detail with reference to the drawings.

  FIG. 1 shows a configuration example of a content recommendation device according to an embodiment of the present invention. This content recommendation device 1 is a user context (time, time zone, season, temperature, humidity, location, event, physical condition (body temperature, heart rate, respiratory rate, etc.)) and content (music, moving image, still image, etc.). A context-mood plane is generated by mapping the feature amount (specifically, for example, mood) on the same plane, and content suitable for the context designated by the user is generated based on the generated context-mood plane. It is something to recommend. The generated context-mood plane can be applied to a new content with a context suitable for viewing.

  Note that classifiable information (genre, style, keyword, etc.) included in content metadata may be used instead of the mood used as the feature amount of the content.

  The content recommendation device 1 includes an evaluation input unit 11 for inputting an evaluator's evaluation (context suitable for viewing and a mood in which the content should be classified), content viewing history information (information indicating the viewed content, Based on an input from at least one of the history information input unit 12, the evaluation input unit 11, or the history information input unit 12 that inputs a context when the content is viewed and a mood that the user feels and inputs to the content) A mapping unit 13 that maps the mood on the same plane to generate a context-mood plane, a context-mood plane holding unit 14 that holds the generated context-mood plane, and a recommendation condition for inputting a context as a recommendation condition Input unit 15, context of recommendation conditions A distance calculation unit 16 for identifying a mood having a correlation with the content, a content extraction unit 17 for extracting content having the identified mood as metadata, a content recommendation unit 18 for recommending the extracted content to the user, and a recommendation A content database (DB) 19 in which a plurality of contents to be processed and metadata (including a feature amount (in this case, mood)) are held, and a word-context conversion for converting a word into a context The word-context conversion table holding unit 20 holds a table.

  The evaluation input unit 11 inputs a context and mood evaluation for each content, for example, as shown in FIG. 2 obtained by allowing a plurality of evaluators to view a large number of contents for learning data, to the mapping unit 13. However, the mood for each content may be obtained by analyzing the content or using information included in the metadata instead of adopting the evaluator's evaluation.

  The history information input unit 12 inputs viewing history information of one or more users to the mapping unit 13. The context at the time of viewing included in the history information includes not only the date and time but also the environmental state (temperature, humidity, speed, acceleration, position (latitude, longitude), etc.) detected by a predetermined sensor when viewing the content. , User physical condition (body temperature, heart rate, blood pressure, etc.) can be included.

  The mapping unit 13 uses, for example, FIG. 3 using multivariate analysis (for example, correspondence analysis) based on an input (a plurality of combinations of context and mood) from at least one of the evaluation input unit 11 or the history information input unit 12. As shown in FIG. 4, the coordinates of each context and mood in a plane with dimension 1 on the horizontal axis and dimension 2 on the vertical axis are determined. For example, as shown in FIG. Map on the same plane. Then, the context-mood plane corresponding to the mapping result is output to the context-mood plane holding unit 14.

  Alternatively, the mapping unit 13 does not use the context included in the history information input from the history information input unit 12, but reads the metadata of the viewed content from the content DB 19, such as a review sentence included in the metadata A word may be extracted from text data, and the extracted word may be converted into a context based on a word-context conversion table.

  The mapping by the mapping unit 13 is not limited to a plane, and may be performed on a three-dimensional space or more.

  The recommendation condition input unit 15 inputs contexts such as summer sea, date, and Christmas designated by the user to the distance calculation unit 16 as recommendation conditions. When one context is input as the recommendation condition, the distance calculation unit 16 is closest to the coordinate position of the context as the recommendation condition on the context-mood correspondence plane (or the distance is equal to or less than a predetermined length). ) The mood located at the coordinates is specified as the mood that meets the recommended conditions.

  When a plurality of contexts are input as recommendation conditions, the coordinates are closest to the center of the coordinate positions of the plurality of contexts (or the distance is equal to or less than a predetermined length) on the context-mood plane. The mood to be specified is specified as the mood that meets the recommended conditions. Alternatively, the mood located at the coordinate closest to the coordinate position of each context that is the recommendation condition (or the distance is equal to or less than a predetermined length) is specified as the mood that meets the recommendation condition. Alternatively, the distance to the coordinate position of each context that is the recommendation condition is calculated and summed for each coordinate position of the mood, and the mood with the shortest total distance is specified as the mood that meets the recommendation condition. Note that the method for determining the mood that meets the recommended conditions is not limited to the method described above.

  The content extraction unit 17 refers to the content DB 19 to extract content having the mood specified by the distance calculation unit 16 as metadata, and outputs the content to the content recommendation unit 18. The content recommendation unit 18 generates a playlist having the content input from the content extraction unit 17 as an element and recommends it to the user.

  In the word-context conversion table holding unit 20, for example, the word “early morning, morning, morning, awakening” or the like is converted into the context “morning”, or the word “snow, cold, coat, teacher” or the like is converted into the context “winter”. A word-context conversion table is stored.

  The evaluation input unit 11 to the word-context conversion table holding unit 20 constituting the content recommendation device 1 do not need to be housed in a single case, and a network in which some components are represented by the Internet. It may be connected with other components via. For example, the evaluation input unit 11 to the mapping 13 related to the generation of the context-mood plane may be realized by a server on the Internet.

  Next, the operation of the content recommendation device 1 will be described. First, the content recommendation device 1 generates a context-mood plane as preparation for recommending content. Then, based on the generated context-mood plane, the mood corresponding to the context specified by the user is specified, and the content having the specified mood is recommended to the user.

  Hereinafter, four types of methods for generating a context-mood plane will be described with reference to FIGS.

  FIG. 5 is a flowchart for explaining the first context-mood plane generation process. In step S <b> 1, the evaluation input unit 11 inputs a context and mood evaluation for each content obtained by allowing a plurality of evaluators to view a large number of contents for learning data to the mapping unit 13. In step S <b> 2, the mapping unit 13 determines the coordinates of each context and each mood using multivariate analysis or the like based on the input from the evaluation input unit 11 (a plurality of combinations of the context and the mood). To map. Then, the context-mood plane corresponding to the mapping result is output to the context-mood plane holding unit 14.

  FIG. 6 is a flowchart for explaining the second context-mood plane generation process. In step S <b> 11, the history information input unit 12 inputs a context (such as a viewing date and time) when viewing the content to the mapping unit 13 based on the user viewing history information. In step S <b> 12, the history information input unit 12 inputs the mood that the user feels and inputs to the content (or the mood given in advance to the music) to the mapping unit 13. Note that the processing in steps S11 and S12 is repeatedly executed for the number of contents having viewing history information. In step S13, the mapping unit 13 determines the coordinates of each context and each mood using multivariate analysis or the like based on the input from the history information input unit 12 (a plurality of combinations of context and mood), and the same plane. Map up. Then, the context-mood plane corresponding to the mapping result is output to the context-mood plane holding unit 14.

  FIG. 7 is a flowchart for explaining a third context-mood correspondence plane generation process. In step S <b> 21, the history information input unit 12 determines the context when viewing the content based on the user's viewing history information (environmental state (temperature, humidity, speed, acceleration, position (latitude, longitude) detected by a predetermined sensor). Etc.) and the user's physical condition (body temperature, heart rate, blood pressure, etc.) are input to the mapping unit 13. In step S <b> 22, the history information input unit 12 inputs the mood that the user feels and inputs to the content (or the mood given in advance to the music) to the mapping unit 13. Note that the processes in steps S21 and S22 are repeatedly executed for the number of contents having viewing history information. In step S23, the mapping unit 13 determines the coordinates of each context and each mood using multivariate analysis or the like based on the input from the history information input unit 12 (a plurality of combinations of contexts and moods). Map up. Then, the context-mood plane corresponding to the mapping result is output to the context-mood plane holding unit 14.

  FIG. 8 is a flowchart for explaining a fourth context-mood correspondence plane generation process. In step S31, a word-context conversion table for converting a word into a context is supplied to the word-context conversion table holding unit 20 and held therein. In step S32, the mapping unit 13 acquires the viewing history information of the user from the history information input unit 12, and specifies the content viewed by the user. In addition, the mapping unit 13 acquires metadata of the identified content from the content DB 19 and extracts words from a review sentence or the like included in the metadata. Further, the mapping unit 13 uses the word-context conversion table held in the word-context conversion table holding unit 20 to convert the extracted word into a context.

  In step S <b> 33, the history information input unit 12 inputs the mood that the user feels and inputs to the content (or the mood given in advance to the music) to the mapping unit 13. Note that the processes in steps S32 and S33 are repeatedly executed for the number of contents having viewing history information. In step S34, the mapping unit 13 determines the coordinates of each context and each mood based on a plurality of combinations of contexts and moods using multivariate analysis or the like, and maps them on the same plane. Then, the context-mood plane corresponding to the mapping result is output to the context-mood plane holding unit 14.

  The context-mood correspondence plane generated by any one of the first to fourth context-mood correspondence plane generation processes described above is used for the content recommendation process described below.

  FIG. 9 is a flowchart for explaining the content recommendation process.

  In step S41, the recommendation condition input unit 15 inputs one or more contexts designated by the user to the distance calculation unit 16 as recommendation conditions. In step S42, the distance calculation unit 16 has a mood located at a coordinate closest to the coordinate position of the context input as the recommendation condition (or the distance is equal to or less than a predetermined length) on the context-mood plane. Is identified as a mood that conforms to the recommendation condition, and is notified to the content extraction unit 17. In step S43, the content extraction unit 17 refers to the content DB 19 to extract the content having the mood specified by the distance calculation unit 16 as metadata (or feature amount), and sends it to the content recommendation unit 18. Output. The content recommendation unit 18 generates a playlist having the content input from the content extraction unit 17 as an element and recommends it to the user. Alternatively, a content group including the content input from the content extraction unit 17 as an element is recommended to the user.

  As described above, according to the content recommendation device 1 to which the present invention is applied, the user can acquire a playlist or a content group of content suitable for the context only by inputting the context as the recommendation condition.

  The context-mood correspondence plane generated by any of the first to fourth context-mood correspondence plane generation processes described above can also be applied to the case where a context is added to new content.

  By the way, the series of processes described above can be executed by hardware, but can also be executed by software. When a series of processing is executed by software, a program constituting the software may execute various functions by installing a computer incorporated in dedicated hardware or various programs. For example, it is installed from a recording medium in a general-purpose personal computer configured as shown in FIG.

  The personal computer 100 includes a CPU (Central Processing Unit) 101. An input / output interface 105 is connected to the CPU 101 via the bus 104. A ROM (Read Only Memory) 102 and a RAM (Random Access Memory) 103 are connected to the bus 104.

  The input / output interface 105 includes a keyboard for inputting operation commands by a user, an input unit 106 including an input device such as a mouse, a CRT (Cathode Ray Tube) or LCD (Liquid Crystal) for displaying an operation screen, a recommended playlist, and the like. Output unit 107 composed of a display such as Display), storage unit 108 composed of a hard disk drive for storing programs and various data, etc., and a modem, a LAN (Local Area Network) adapter, etc., via a network represented by the Internet A communication unit 109 that executes communication processing is connected. Also, a magnetic disk (including a flexible disk), an optical disk (including a CD-ROM (Compact Disc-Read Only Memory), a DVD (Digital Versatile Disc)), a magneto-optical disk (including an MD (Mini Disc)), or a semiconductor A drive 110 for reading / writing data from / to a recording medium 111 such as a memory is connected.

  A program for causing the personal computer 100 to execute the above-described series of processing is supplied to the personal computer 100 in a state stored in the recording medium 111, read by the drive 110, and installed in a hard disk drive built in the storage unit 108. Has been. The program installed in the storage unit 108 is loaded from the storage unit 108 to the RAM 103 and executed in response to a command from the CPU 101 corresponding to a command from the user input to the input unit 106.

  In this specification, the steps executed based on the program are executed in parallel or individually even if they are not necessarily processed in time series, as well as processes executed in time series according to the described order. It also includes processing.

  The program may be processed by a single computer, or may be distributedly processed by a plurality of computers. Furthermore, the program may be transferred to a remote computer and executed.

  The embodiment of the present invention is not limited to the above-described embodiment, and various modifications can be made without departing from the gist of the present invention.

It is a block diagram which shows the structural example of the content recommendation apparatus to which this invention is applied. It is a figure explaining the context with respect to a content, and evaluation of a mood. It is a figure which shows an example of the two-dimensional coordinate determined with respect to each context and each mood. It is a figure which shows an example of the context-mood corresponding plane where the context and the mood are mapped. It is a flowchart explaining the 1st context-mood corresponding plane generation processing. It is a flowchart explaining the 2nd context-mood corresponding plane generation process. It is a flowchart explaining the 3rd context-mood corresponding plane generation process. It is a flowchart explaining the 4th context-mood corresponding plane generation process. It is a flowchart explaining a content recommendation process. It is a block diagram which shows the structural example of a general purpose personal computer.

Explanation of symbols

  DESCRIPTION OF SYMBOLS 1 Content recommendation apparatus, 11 Evaluation input part, 12 History information input part, 13 Mapping part, 14 Context-mood corresponding plane holding part, 15 Recommendation condition input part, 16 Distance calculation part, 17 Content extraction part, 18 Content recommendation part, 19 content DB, 20 word-context conversion table holding unit, 100 personal computer, 101 CPU, 111 recording medium

Claims (9)

In an information processing apparatus that recommends content suitable for the context of a recommendation condition, with a context representing a user's situation when viewing content as a recommendation condition,
An input means for inputting a combination of a feature amount of the content and a context associated with the content;
Generating means for generating correlation information indicating a correlation between the feature quantity and the context by analyzing a combination of the feature quantity and the context input by the input means;
A designation means for designating a context as a recommendation condition;
Based on the correlation information, specifying means for specifying a feature quantity correlated with the context specified by the specifying means;
An information processing apparatus comprising: recommendation means for recommending content having the feature amount specified by the specifying means to the user. The generation unit analyzes a combination of the feature amount and the context input by the input unit, thereby obtaining a plurality of states of the feature amount and the context as correlation information indicating a correlation between the feature amount and the context. The information processing apparatus according to claim 1, wherein a feature quantity-context-corresponding space in which the plurality of states are mapped on the same space is generated. The information processing apparatus according to claim 2, wherein the specifying unit specifies a feature amount that is close in distance to the context specified by the specifying unit in the feature amount-context correspondence space. The information processing apparatus according to claim 2, wherein the same space and the feature quantity-context corresponding space are two-dimensional planes. The information processing apparatus according to claim 1, wherein the input unit inputs a combination of a feature amount of the content and a context associated with the content obtained from an evaluator who viewed the content. The information processing apparatus according to claim 1, wherein the input unit inputs a combination of a feature amount of the content and a context associated with the content obtained from a user's content viewing history. The information processing apparatus according to claim 1, wherein the input unit inputs a combination of the feature amount and the context detected by a predetermined sensor when content is viewed. In the information processing method of the information processing apparatus that recommends content suitable for the context of the recommendation condition, using the context representing the user's situation when viewing the content as the recommendation condition,
Enter the combination of the feature amount of the content and the context associated with the content,
By analyzing the combination of the input feature value and the context, correlation information indicating the correlation between the feature value and the context is generated,
Specify the context as a recommendation condition,
Based on the correlation information, identify a feature quantity correlated with the specified context,
An information processing method including a step of recommending content having a specified feature quantity to the user. A program for recommending content suitable for the context of the recommendation condition, with the context representing the user's situation when viewing the content as the recommendation condition,
Enter the combination of the feature amount of the content and the context associated with the content,
By analyzing the combination of the input feature value and the context, correlation information indicating the correlation between the feature value and the context is generated,
Specify the context as a recommendation condition,
Based on the correlation information, identify a feature quantity correlated with the specified context,
A program for causing a computer to execute a process including a step of recommending to the user content having a specified feature amount.

Images