JP2014056426A - Content recommendation system, content recommendation method and program, and terminal device and program - Google Patents

Abstract

PROBLEM TO BE SOLVED: To recommend a content suited for a user's preference and situation.SOLUTION: A context acquisition unit 21 acquires a context indicating a target user's current situation. A similarity calculation unit 25, for each candidate user stored by a history storage unit 23, calculates similarity of the candidate user with the target user with the use of a value based on the number of contents associated both with the candidate user and with the target user. A recommendation degree calculation unit 26, for each content associated with one or a plurality of candidate users whose similarity calculated by the similarity calculation unit 25 is high, calculates a recommendation degree of the content with the use of a value indicating a degree of association between a context associated with the content and the context acquired by the context acquisition unit 21. A recommendation unit 27 recommends one or a plurality of contents whose recommendation degree calculated by the recommendation degree calculation unit 26 is high, to the target user.

Description

  The present invention relates to a content recommendation system for recommending content to a user, a content recommendation method, a program, a terminal device, and a program.

In recent years, the provision of content using a network, such as music data distribution and video data distribution, has become widespread.
In Patent Document 1, contents used in the situation are stored in advance in association with weight values for a plurality of keywords indicating the status of the terminal, and weight values and contents indicating the current terminal status are stored for each keyword. It is described that a product of associated weight values is calculated, and the sum is used as a recommendation level to recommend content with a high recommendation level.

  In Patent Literature 2, the content used in the situation is stored in advance in association with the situation vector indicating the user situation, and the inner product of the situation vector indicating the current user situation and the situation vector associated with the content is stored. It is described that a situation similarity is calculated and content with a high situation similarity is recommended.

JP 2009-129359 A JP 2010-262383 A

  However, when the methods disclosed in Patent Documents 1 and 2 are used, it is possible to recommend content used in a situation similar to the current situation, but the preference of the user receiving the recommendation and the other used in the situation The user's preference is not always similar. For this reason, when the methods disclosed in Patent Documents 1 and 2 are used, there is a risk of recommending content that does not match the user's preference.

  The objective of this invention is providing the content recommendation apparatus, the content recommendation method, and program which solve the subject mentioned above.

  The present invention has been made to solve the above-described problem, and is a content recommendation system that recommends content to a user when the content is used in association with the content used by the user for each user. A storage unit that stores a context indicating the user's situation, a context acquisition unit that acquires a context indicating a current situation of the target user who is a user to be recommended, and the target among the users stored in the storage unit For each candidate user who is a user other than the user, the similarity between the candidate user and the target user is calculated using a value based on the number of contents commonly associated with the candidate user and the target user. The similarity calculation unit is associated with one or a plurality of candidate users having a high similarity calculated by the similarity calculation unit. A recommendation level calculation unit that calculates a recommendation level of the content using a value indicating a high degree of association between the context associated with the content and the context acquired by the context acquisition unit for each content A recommendation unit that recommends one or a plurality of contents having a high recommendation level calculated by the recommendation level calculation unit to the target user.

  In addition, in the present invention, the similarity calculation unit, for each candidate user, for each piece of content that is commonly associated with the candidate user and the target user, in common with the candidate user and the target user A value based on the number of contexts associated with the content is calculated, and a similarity between the candidate user and the target user is calculated using a sum of the values.

  In addition, in the present invention, the similarity calculation unit, for each candidate user, for each piece of content that is commonly associated with the candidate user and the target user, in common with the candidate user and the target user Dividing the number of contexts associated with the content by the number of contexts associated with the target user and the content, and calculating the similarity between the candidate user and the target user using the sum of the values It is characterized by.

  Further, in the present invention, the recommendation degree calculation unit includes, for each content associated with one or a plurality of candidate users having a high degree of similarity calculated by the similarity degree calculation unit, the context associated with the content and the context acquisition unit. A value obtained by dividing the value indicating the degree of association with the acquired context by the number of contexts associated with the target user and the content is calculated as the recommendation level of the content. .

  In the present invention, the storage unit is a flag for a plurality of contexts associated with the content used by the user for each user, and indicates the status of the user when the context uses the content. A status flag indicating whether or not the content flag is stored, and when there is content for which the status flag is not set among the content stored in the storage unit, a user similar to the user associated with the content and the content A context supplementing unit is provided for setting a status flag associated with the content to which the status flag is not set.

  In addition, the present invention is a content recommendation method using a content recommendation system including a storage unit that stores a context indicating a situation of the user when the user uses the content in association with the content used by the user. The context acquisition unit acquires a context indicating the current situation of the target user who is the user to be recommended, and the similarity calculation unit is a user other than the target user among the users stored in the storage unit For each candidate user who is a user, using a value based on the number of contents commonly associated with the candidate user and the target user, calculating a similarity between the candidate user and the target user; The recommendation level calculation unit associates with one or a plurality of candidate users having a high similarity level calculated by the similarity level calculation unit. Calculating a recommendation level of the content using a value indicating a high degree of association between the context associated with the content and the context acquired by the context acquisition unit for each content And a step of recommending, to the target user, one or a plurality of contents having a high recommendation level calculated by the recommendation level calculation unit.

  In addition, the present invention is a storage unit for storing a context indicating a situation of the user when the computer is associated with the content used by the user for each user and the user is used, and a user to be recommended. For each candidate user who is a user other than the target user among the users stored in the storage unit, the context acquisition unit that acquires a context indicating the current status of the target user, and in common for the candidate user and the target user A similarity calculation unit that calculates a similarity between the candidate user and the target user using a value based on the number of associated contents, and one or a plurality of candidate users with a high similarity calculated by the similarity calculation unit For each content associated with the context, the context associated with the content and the context acquisition unit Using a value indicating the degree of association with the acquired context, a recommendation level calculation unit that calculates the recommendation level of the content, and one or a plurality of contents with a high recommendation level calculated by the recommendation level calculation unit. A program for causing a function to serve as a recommendation unit recommended to the target user.

  In addition, the present invention is a terminal device that recommends content to a user, and a context acquisition unit that acquires a context indicating a current situation of the user of the own device, and the same content used in the past by the user of the own device Other users who have used the content recommend content that has been used in the past to the user of the own device with content having a high degree of association with the context acquired by the context acquisition unit. And a presentation unit that presents the content.

  The present invention also provides a context acquisition unit for acquiring a context indicating the current status of a user of the computer, another user who has used the same content as the content previously used by the user of the computer However, in order to function as a presentation unit that presents content that has been used in the past and has a context that is highly relevant to the context acquired by the context acquisition unit as content recommended to the user of the computer. It is a program.

  ADVANTAGE OF THE INVENTION According to this invention, the content suitable for a user preference and a condition can be recommended.

It is the schematic which shows the structure of the content recommendation system by one Embodiment of this invention. It is a schematic block diagram which shows the structure of the terminal device by one Embodiment of this invention. It is a schematic block diagram which shows the structure of the content recommendation apparatus by one Embodiment of this invention. It is a figure which shows the example of the information which a log | history memory | storage part memorize | stores. It is a flowchart which shows the outline of a content recommendation method. It is a flowchart which shows the operation | movement which records the usage history of a content. It is a figure which shows an example of the input form screen which a content recommendation apparatus transmits and is displayed on a user's terminal device. It is a flowchart which shows the operation | movement which recommends a content.

Hereinafter, embodiments of the present invention will be described in detail with reference to the drawings.
FIG. 1 is a schematic diagram showing a configuration of a content recommendation system 1 according to an embodiment of the present invention.
The content recommendation system 1 includes a plurality of terminal devices 10 used by a user and a content recommendation device 20 that performs content recommendation processing.
When the user uses the content, the terminal device 10 transmits the usage status of the content to the content recommendation device 20. Further, when receiving a content recommendation from the content recommendation device 20, the terminal device 10 transmits the current situation to the content recommendation device 20.
The content recommendation device 20 accumulates the usage status of the content received from the terminal device 10. Further, the content recommendation device 20 recommends content suitable for the user of the terminal device 10 based on the current status received from the terminal device 10 and the accumulated usage status.

FIG. 2 is a schematic block diagram showing the configuration of the terminal device 10 according to the embodiment of the present invention.
The terminal device 10 includes a situation input unit 11, a situation transmission unit 12, a recommendation reception unit 13, and a presentation unit 14.
When the user uses the content or recommends the content to the user, the situation input unit 11 receives the sensor information indicated by the sensor of the own device and the text input on the input screen displayed on the display of the own device. And received as information indicating the user's situation. Examples of the user situation include the current time, location, direction, temperature, and mood.
The status transmission unit 12 transmits the information received by the status input unit 11 to the content recommendation device 20.
The recommendation receiving unit 13 receives content information recommended to the user of the own device from the content recommendation device 20.
The presenting unit 14 presents information received by the recommendation receiving unit 13. Examples of the presentation method include a method of displaying on a display and a method of outputting from a speaker as sound.

FIG. 3 is a schematic block diagram showing the configuration of the content recommendation device 20 according to one embodiment of the present invention.
The content recommendation device 20 includes a context acquisition unit 21, a history addition unit 22, a history storage unit 23, a history acquisition unit 24, a similarity calculation unit 25, a recommendation degree calculation unit 26, a recommendation unit 27, and a context supplement unit 28.

  The context acquisition unit 21 acquires a context indicating the user status based on the information indicating the user status transmitted from the terminal device 10. Examples of the context include “morning” and “night” indicating the time, “fun” and “sad” indicating the user's mood, “single” and “family” indicating the content information, and the like.

  The history adding unit 22 records the context acquired by the context acquiring unit 21 in the history storage unit 23 in association with the user and the content when the user uses the content.

FIG. 4 is a diagram illustrating an example of information stored in the history storage unit 23.
The history storage unit 23 associates a user who uses the content with the content and a context indicating a situation where the user has used the content, and stores it as a content usage history. The history storage unit 23 stores context information as a flag indicating whether or not the context indicates a situation where the content is used. In addition, when the flag indicates “0”, it indicates that the context does not indicate a situation where the content is used. When the flag indicates “1”, the context indicates a situation where the content is used. It represents that.

  In addition, according to FIG. 4, it can be seen that the user U001 used the content C001 for singles in the morning in a pleasant mood. In addition, it can be seen that the user U001 used the family-oriented content C002 in a pleasant mood at night. On the other hand, it can be seen that the user U002 used the family-oriented content C002 in a sad mood at night. Thus, the usage status of the content may vary depending on the user's preference.

  The history acquisition unit 24 obtains, from the history storage unit 23, the content usage history of the user (hereinafter referred to as the target user) to be recommended and the content usage history of the user other than the target user (hereinafter referred to as the candidate user). get.

  The similarity calculation unit 25 calculates the similarity between the target user and the candidate user's preference based on the content usage history of the target user acquired by the history acquisition unit 24 and the content usage history of the candidate user.

  The recommendation degree calculation unit 26 indicates a degree of recommendation indicating a degree to be recommended in a situation indicated by the context acquired by the context acquisition unit 21 for each content used in the past by a candidate user having a high similarity calculated by the similarity calculation unit 25. Is calculated.

  The recommendation unit 27 recommends to the user by transmitting a list of contents having a high recommendation level calculated by the recommendation level calculation unit 26 to the user terminal device 10.

  When there is a flag that is not set in the context included in the content usage history acquired by the history acquisition unit 24, the context supplementing unit 28 uses a user usage history similar to the user associated with the usage history. Supplement the flag.

Here, an outline of the operation of the content recommendation device 20 according to the present embodiment will be described.
FIG. 5 is a flowchart showing an outline of the content recommendation method.
First, the context acquisition unit 21 acquires a context indicating the current situation of the target user (step S1). Next, the similarity calculation unit 25 uses, for each candidate user stored in the history storage unit 23, a value based on the number of contents commonly associated with the candidate user and the target user, The degree of similarity with the target user is calculated (step S2).

Next, for each content associated with one or a plurality of candidate users having a high similarity calculated by the similarity calculation unit 25, the recommendation degree calculation unit 26 acquires the context associated with the content and the context acquisition unit 21. The recommendation level of the content is calculated using a value indicating the high degree of association with the context (step S3). Then, the recommendation unit 27 recommends one or more contents with a high recommendation level calculated by the recommendation level calculation unit 26 to the target user (step S4).
Thereby, the content recommendation apparatus 20 can recommend the content according to the user's preference and situation.

Hereinafter, the operation of the content recommendation system 1 according to the present embodiment will be described in detail.
First, the operation of recording content usage history in the content recommendation device 20 will be described.
FIG. 6 is a flowchart showing the operation of recording the content usage history.
When the user starts using the content using the terminal device 10, the status transmission unit 12 of the terminal device 10 transmits a user ID indicating the user and a content ID indicating the content to the content recommendation device 20. The context acquisition unit 21 of the content recommendation device 20 receives the user ID and the content ID from the terminal device 10 (step S101). Next, the context acquisition unit 21 transmits an input form screen for inputting the current state of the user to the user terminal device 10 (step S102).

FIG. 7 is a diagram showing an example of an input form screen transmitted from the content recommendation device 20 and displayed on the user terminal device 10.
The status input unit 11 of the terminal device 10 receives the input form screen from the content recommendation device 20 and displays a screen as shown in FIG. 7 on the display.
On the input form screen generated by the context acquisition unit 21, a button B1, a cancel button B2, and an OK button B3 for accepting the user's situation selection are displayed.
The button B1 for accepting the selection of the situation includes a button for accepting a selection of “fun”, “sad”, or “happy” indicating the current mood of the user, and “house” indicating the location where the user is currently located. , A button for accepting selection of either “company” or “going out”.

  The situation input unit 11 accepts the press of the OK button B3 after accepting the press of the button B1 for accepting the selection of the situation. As a result, the status transmission unit 12 transmits the information associated with the pressed button B1 to the content recommendation device 20. Note that when the cancel button B2 is pressed, the terminal device 10 does not transmit information. At this time, the situation input unit 11 acquires the current time, place, direction, temperature, and the like, and the situation transmission unit 12 transmits the information to the content recommendation device 20 in addition to the information input on the input form screen.

  FIG. 7 shows an example in which a context is selected by pressing a button. However, the present invention is not limited to this. For example, the context acquisition unit 21 generates an input form screen including a text input form and The input may be received and the sentence may be transmitted to the content recommendation device 20.

  When receiving the information from the terminal device 10 (step S103), the context acquisition unit 21 of the content recommendation device 20 acquires the context from the information (step S104). For example, when the “fun” button and the “house” button are pressed as input values for the input form screen shown in FIG. 7, the context “fun” and the context “home” are extracted from the information. Also, for example, when a text input form is included in the input form screen, a sentence input to the input form is subjected to morphological analysis, and from the obtained morphemes, defined as a context in the history storage unit 23 Is extracted as a context. When the content recommendation device 20 includes a thesaurus in advance, it is determined whether or not a word similar to the defined context exists in the morpheme. If there is, the morpheme may be extracted as the context. . The thesaurus is a word database in which synonyms, synonyms, upper concept words, lower concept words, and the like are linked.

  For example, the context may be extracted by applying information such as time, place, direction, and temperature acquired from the terminal device 10 to a predetermined context condition. For example, when the condition “extract the context“ morning ”when“ time ”is from 5 am to 11 am” ”is set in advance, information on the obtained time is set as the condition. By applying, it is determined whether or not to enable the context “morning”.

  Further, the context acquisition unit 21 acquires metadata associated with the content ID acquired in step S101. Examples of the metadata include data that is input when the content is generated and data that is input as a POP (Point Of Purchase) advertisement by the content seller. Such metadata can be acquired from a content providing server or the like. And the context acquisition part 21 acquires contexts, such as a content author, a genre, and a publisher, from these metadata.

  When the context acquisition unit 21 acquires a context in step S104, the history adding unit 22 associates the context acquisition unit 21 with the user ID and content ID received in step S101, and indicates a context flag indicating the context acquired by the context acquisition unit 21 Is recorded in the history storage unit 23 (step S105).

  Accordingly, the content recommendation device 20 accumulates the content usage history in the history storage unit 23 every time the user uses the content.

Next, an operation in which the content recommendation device 20 recommends content will be described.
FIG. 8 is a flowchart showing an operation for recommending content.
When recommending content to a user, first, the context acquisition unit 21 acquires the user ID and context information of a target user who is a user to be recommended by a method similar to steps S101 to S104 described above (step S101). S201). This corresponds to step S1 shown in FIG.

  Next, the history acquisition unit 24 acquires, from the history storage unit 23, the usage history of content associated with the user ID of the target user and the usage history of candidate users who are users other than the target user (step S202). .

  Next, the similarity calculation unit 25 selects one candidate user acquired by the history acquisition unit 24 one by one, and executes the processes of Step S204 to Step S210 shown below for all candidate users (Step S203).

  First, the similarity calculation unit 25 compares the usage history of the target user with the usage history of the candidate user selected in step S203, and the content common to both, that is, both the target user and the candidate user have used in the past. A certain content is specified (step S204). Next, the similarity calculation unit 25 selects the common contents identified in step S204 one by one, and executes the processes of steps S206 to S208 shown below for all the common contents (step S205).

  The similarity calculation unit 25 acquires a context status flag associated with the target user and the selected candidate user for the content selected in step S205 (step S206). Next, the similarity calculation unit 25 calculates the number of status flags indicating “1” for both the target user and the selected candidate user among the status flags acquired in step S206 (step S207). Next, the similarity calculation unit 25 divides the calculated number of common status flags by the number of status flags indicating “1” for the target user, so that the context between the target user and the candidate user is determined. A common ratio is calculated (step S208).

When the similarity calculation unit 25 calculates the ratio calculated in step S208 for all the common contents, the similarity calculation unit 25 calculates the sum of the ratios (step S209). Then, the similarity calculation unit 25 calculates the similarity of the preference between the target user and the candidate user by dividing the sum of the ratios by the number of contents associated with the target user (step S210). The similarity calculated here indicates that the higher the value, the higher the possibility that the target user and the candidate user will use the same content under the same situation, that is, the preference is similar.
Note that the procedure of steps S203 to S210 corresponds to step S2 shown in FIG.

  When the similarity calculation unit 25 calculates the similarity for all candidate users acquired by the history acquisition unit 24, the recommendation degree calculation unit 26 selects some of the top similarities calculated by the similarity calculation unit 25 (for example, the top 10 Among the contents associated with the candidate user, the contents excluding the contents associated with the target user acquired by the history acquisition unit 24 are extracted (step S211).

Next, the recommendation level calculation unit 26 selects the acquired contents one by one, and executes the processes of steps S213 to S215 described below for all the contents (step S212).
First, the recommendation level calculation unit 26 acquires a context status flag associated with the selected candidate user for the content selected in step S212 (step S213). Next, the recommendation level calculation unit 26 calculates the number of contexts that are included in the context information acquired in step S201 and whose status flag indicates “1” for the selected candidate user (step S214). Next, the recommendation level calculation unit 26 calculates the recommendation level of the content by dividing the calculated number of common status flags by the number of contexts acquired in step S201 (step S215).
In addition, the procedure of the said step S211-step S215 is corresponded to step S3 shown in FIG.

  When the recommendation level calculation unit 26 calculates the recommendation level for all the contents selected in step S212, the recommendation unit 27 recommends the content to the user in descending order of recommendation level (step S216). Note that step S216 corresponds to step S4 shown in FIG. The recommendation unit 27 makes a recommendation by generating a content list in which content is arranged in descending order of recommendation and transmitting the content list to the terminal device 10 of the target user. Thereby, the recommendation receiving unit 13 of the terminal device 10 of the target user receives the content list, and the presentation unit 14 displays the content list on the display.

  By the way, depending on the usage form of the content, the content recommendation device 20 may not be able to acquire the context information when the content is used, for example, when the content stored in the medium purchased by the user at the actual store is used. is there. In such a case, since the context acquisition unit 21 of the content recommendation device 20 cannot acquire context information, the context status flag stored in the history storage unit 23 may be NULL.

  In such a case, the context supplementation unit 28 of the content recommendation device 20 supplements the context stored in the history storage unit 23, and the similarity calculation unit 25 and the recommendation degree calculation unit 26 use the supplemented context for content. Make a recommendation.

  If there is a context status flag associated with the candidate user selected by the similarity calculation unit 25 in step S203, the context supplement unit 28 is a quasi-candidate who is the user most similar to the candidate user. Identify users. The similarity calculation unit 25 calculates the similarity between the candidate user and another user through the processing of step S203 to step S208, and specifies the other user with the highest similarity as a quasi-candidate user.

  Next, the context supplementing unit 28 overwrites the candidate user's status flag which is NULL with the status flag of the quasi-candidate user. At this time, the context supplement unit 28 overwrites the status flag on the internal memory and does not overwrite the history storage unit 23. This is because it is assumed that the context whose status flag is NULL will be satisfied later, and that more appropriate estimation can be made by collecting further usage history, so that these are not hindered. It is to do.

Next, the context supplement unit 28 determines whether all candidate user status flags have been supplemented. When it is determined that the status flag indicating NULL remains, the context supplementing unit 28 specifies a user similar to the candidate user as the quasi-candidate user after the user specified as the quasi-candidate user. Repeatedly.
On the other hand, when the context supplementation unit 28 determines that all the situation flags have been supplemented, the content recommendation device 20 executes the processes after step S204.
Thereby, the content recommendation apparatus 20 can calculate the similarity between the target user and the candidate user and the recommendation degree of the content in a situation where the situation flags are set in all contexts.

As described above, the embodiment of the present invention has been described in detail with reference to the drawings. However, the specific configuration is not limited to the above, and various design changes and the like can be made without departing from the scope of the present invention. It is possible to
For example, in the above-described embodiment, the case where the similarity between users is calculated according to the procedure of step S204 to step S210 has been described, but is not limited thereto. For example, the similarity calculation unit 25 divides the number of contents commonly associated with the target user and the candidate user by the number of contents associated with the target user, so that the preference between the target user and the candidate user is obtained. The similarity may be calculated.

  In the embodiment described above, the recommendation level calculation unit 26 calculates the recommendation level by dividing the number of contexts common to the target user and the candidate user by the number of contexts associated with the target user. Although explained, it is not limited to this. For example, instead of “the number of contexts common to the target user and the candidate user”, “the number of contexts related between the target user and the candidate user” may be used as a numerator. The “number of related contexts between the target user and the candidate user” is a value obtained by weighting according to the similarity between contexts. For example, in the case where “bright” is included in the context of the target user and “fun” is not included, the case where “fun” is included in the context of the candidate user and “bright” is not included will be described as an example. At this time, “the number of contexts common to the target user and the candidate user” is zero. On the other hand, since “bright” and “fun” are related contexts, these contexts are “related contexts between the target user and the candidate user”. Therefore, a weight value (for example, 0.8) according to a predetermined degree of association may be used for calculating the recommendation degree.

  As the weight value, for example, a reciprocal of the number of links until reaching a link from one context to the other context in a predetermined thesaurus can be used. Note that the method of calculating the weight value is not limited to this. For example, the weight value may be calculated using a distance between words in a word vector space created in advance, a distance obtained by word clustering, or the like.

  Further, in the above-described embodiment, the case where the context supplement unit 28 supplements the situation flag is described when there is a candidate user context in which the situation flag is not set. However, the present invention is not limited thereto. For example, the similarity calculation unit 25 and the recommendation level calculation unit 26 may perform the processing by replacing the case where the status flag is NULL with “0.5”, or the case where the status flag is NULL. The similarity may be calculated while ignoring.

  Moreover, although embodiment mentioned above demonstrated the case where the context supplement part 28 overwrites a candidate user's condition flag on a semi-candidate user's condition flag in order similar to the said candidate user, it is not restricted to this. For example, the context supplement unit 28 may supplement the situation flag by merging and overwriting the situation flags of a plurality of semi-candidate users whose similarity with the candidate user is equal to or greater than a predetermined threshold.

  In the embodiment described above, the case where the configurations of the terminal device 10 and the content recommendation device 20 are those illustrated in FIGS. 2 and 3 has been described, but the present invention is not limited thereto. For example, the terminal device 10 may have a part of the functions of the content recommendation device 20 or may have all the functions.

  The terminal device 10 and the content recommendation device 20 described above have a computer system inside. The operation of each processing unit described above is stored in a computer-readable recording medium in the form of a program, and the above processing is performed by the computer reading and executing this program. Here, the computer-readable recording medium means a magnetic disk, a magneto-optical disk, a CD-ROM, a DVD-ROM, a semiconductor memory, or the like. Alternatively, the computer program may be distributed to the computer via a communication line, and the computer that has received the distribution may execute the program.

  The program may be for realizing a part of the functions described above. Furthermore, what can implement | achieve the function mentioned above in combination with the program already recorded on the computer system, what is called a difference file (difference program) may be sufficient.

  DESCRIPTION OF SYMBOLS 1 ... Content recommendation system 10 ... Terminal device 11 ... Situation input part 12 ... Situation transmission part 13 ... Recommendation reception part 14 ... Presentation part 20 ... Content recommendation apparatus 21 ... Context acquisition part 22 ... History addition part 23 ... History storage part 24 ... History acquisition unit 25 ... similarity calculation unit 26 ... recommendation degree calculation unit 27 ... recommendation unit 28 ... context supplement unit

Claims (9)

A content recommendation system for recommending content to a user,
For each user, a storage unit that stores a context indicating a situation of the user when the content is used in association with the content used by the user;
A context acquisition unit that acquires a context indicating a current situation of a target user who is a user to be recommended;
For each candidate user who is a user other than the target user among the users stored in the storage unit, the candidate user using a value based on the number of contents commonly associated with the candidate user and the target user A similarity calculation unit that calculates the similarity between the target user and the target user;
For each content associated with one or a plurality of candidate users having a high similarity calculated by the similarity calculation unit, a high degree of association between the context associated with the content and the context acquired by the context acquisition unit A recommendation level calculation unit that calculates a recommendation level of the content using a value indicating
A content recommendation system comprising: a recommendation unit that recommends one or a plurality of contents having a high recommendation level calculated by the recommendation level calculation unit to the target user. For each candidate user, the similarity calculation unit is associated with the content in common with the candidate user and the target user for each content commonly associated with the candidate user and the target user. The content recommendation system according to claim 1, wherein a value based on the number of contexts is calculated, and a similarity between the candidate user and the target user is calculated using a sum of the values. For each candidate user, the similarity calculation unit is associated with the content in common with the candidate user and the target user for each content commonly associated with the candidate user and the target user. The number of contexts is divided by the number of contexts associated with the target user and the content, and the similarity between the candidate user and the target user is calculated using the sum of the values. Item 3. The content recommendation system according to Item 2. For each piece of content associated with one or a plurality of candidate users having a high degree of similarity calculated by the similarity calculation unit, the recommendation level calculation unit includes a context associated with the content and a context acquired by the context acquisition unit. The value obtained by dividing the value indicating the degree of relevance between the content by the number of contexts associated with the target user and the content is calculated as the recommendation level of the content. Item 4. The content recommendation system according to any one of Items3. The storage unit is a flag for a plurality of contexts in association with the content used by the user for each user, and indicates whether the context indicates the status of the user when the content is used. Remember the status flag to indicate,
When there is content for which the status flag is not set among the content stored in the storage unit, the status flag associated with the user similar to the user associated with the content and the content is represented by the status flag. The content recommendation system according to any one of claims 1 to 4, further comprising: a context supplement unit configured to set content that has not been set. A content recommendation method using a content recommendation system including a storage unit that stores a context indicating a situation of the user when the user uses the content in association with the content used by the user for each user,
A step of acquiring a context indicating a current situation of a target user who is a user to be recommended by a context acquisition unit;
For each candidate user who is a user other than the target user among the users stored in the storage unit, the similarity calculation unit calculates a value based on the number of contents commonly associated with the candidate user and the target user. Using to calculate the similarity between the candidate user and the target user;
For each content associated with one or a plurality of candidate users having a high similarity calculated by the similarity calculation unit, the recommendation degree calculation unit calculates the context associated with the content and the context acquired by the context acquisition unit. Calculating a recommendation level of the content using a value indicating a high degree of association between the content,
A content recommendation method comprising: a recommendation unit recommending one or a plurality of contents having a high recommendation level calculated by the recommendation level calculation unit to the target user. Computer
A storage unit that stores, for each user, a context that indicates a situation of the user when the content is used in association with the content used by the user;
A context acquisition unit for acquiring a context indicating a current situation of a target user who is a user to be recommended;
For each candidate user who is a user other than the target user among the users stored in the storage unit, the candidate user using a value based on the number of contents commonly associated with the candidate user and the target user A similarity calculation unit that calculates the similarity between the target user and the target user;
For each content associated with one or a plurality of candidate users having a high similarity calculated by the similarity calculation unit, a high degree of association between the context associated with the content and the context acquired by the context acquisition unit A recommendation degree calculation unit for calculating a recommendation degree of the content using a value indicating
A program for causing one or a plurality of contents having a high recommendation level calculated by the recommendation level calculation unit to function as a recommendation unit recommended to the target user. A terminal device that recommends content to a user,
A context acquisition unit that acquires a context indicating the current status of the user of the device;
There is a context that is used in the past by another user who has used the same content as the content used in the past by the user of the device, and has a high relevance with the context acquired by the context acquisition unit. And a presentation unit that presents the attached content as content recommended to the user of the device. Computer
A context acquisition unit for acquiring a context indicating the current status of the user of the computer;
There is a context that has been used in the past by another user who has used the same content as the content used in the past by the user of the computer and has a high relevance with the context acquired by the context acquisition unit. A program for causing the attached content to function as a presentation unit that presents the content as recommended content to the user of the computer.

Images