JP6185379B2 - RECOMMENDATION DEVICE AND RECOMMENDATION METHOD - Google Patents

Description

  The present invention relates to a recommendation device and a recommendation method for making recommendations.

  There is a Content Based Filtering method as a technique for rearranging content based on past browsing content of the user. In this Content Based Filtering method, content having the same characteristics as content (satisfied content) browsed by the user in the past is output to the top. Therefore, in the Content Based Filtering method, in order to specify the same characteristics as the content browsed by the user in the past, a process for estimating the user characteristics is required.

  As such a technique, there is a technique described in Patent Document 1. In the technique described in Patent Document 1, the similarity between the currently viewed content and the history content is calculated, and the similarity of each history content is multiplied by the user evaluation value for each history content to obtain the history content. A corrected evaluation value is calculated. Then, the history is rearranged in descending order of the correction evaluation value, the feature vector of the user is calculated so as to satisfy the arrangement order, and the content similar to the user feature vector is recommended.

JP 2010-262401 A

  However, the technique described in Patent Literature 1 has a problem that the amount of calculation increases because it is necessary to calculate the inner product of the content of the recommendation candidate and the history content.

  Therefore, in order to solve the above-described problems, an object of the present invention is to provide a recommendation device and a recommendation method that make recommendations according to user characteristics while reducing the amount of calculation.

  In order to solve the above-described problem, the recommendation device according to the present invention includes a transposed index storage unit that stores a transposed index table in which one or a plurality of contents are associated with each feature element constituting a content feature vector of content. History information storage means for storing history information of contents accessed by each user terminal, and when a recommendation request is made from the user terminal, the history information stored in the history information storage means corresponds to the user terminal that requested the recommendation Feature element specifying means for specifying the feature elements of the history information content, recommendation candidate search means for searching the content corresponding to the feature elements specified by the feature element specifying means using the transposed index table as recommendation candidates, and recommendation Record from the content searched by candidate search means It includes a recommendation determination means for determining a command target, and recommendation output means for outputting the content of the recommendation subject recommendation determination means has determined a.

  Also, the recommendation method of the present invention is a recommendation method in the recommendation device, in which a transposition index table in which one or a plurality of contents are associated with each feature element constituting a content feature vector of the content is stored in the transposition index storage means. An index storage step, a history information storage step for storing history information of contents accessed by each user terminal in the history information storage means, and a history information stored in the history information storage means when a recommendation request is made from the user terminal Among them, the feature element specifying step for specifying the feature element of the history information content corresponding to the user terminal that requested the recommendation, and the transposition index table using the content corresponding to the feature element specified in the feature element specifying step as a recommendation candidate The Includes a recommendation candidate search step of search, the recommendation determining step of determining a recommendation target from the content retrieved in recommendation candidate search step, the recommendation output step of outputting the content of the recommendation target determined in recommendation determination step.

  According to the present invention, the transposed index table in which one or a plurality of contents are associated is stored for each feature element constituting the content feature vector of the content. When the recommendation request is received from the user terminal, the feature element of the history information content corresponding to the user terminal that requested the recommendation is specified in the history information, and the content corresponding to the specified feature element is transposed as a recommendation candidate. A search is performed using an index table, a recommendation target is determined from the searched content, and the recommendation target content is output. Thereby, a transposed index table in which one or a plurality of contents are associated with each other is stored in advance, and using the transposed index table, the content corresponding to the feature element of the content of the history information is set as a recommendation candidate, and from the recommendation candidate Since the content to be recommended is determined, the amount of calculation can be reduced as compared with the case where the content of the history information and each content are compared and recommended. Thus, by storing the transposed index table in advance, it is possible to make a recommendation by real-time processing.

  Further, in the recommendation information generating apparatus of the present invention, an inverted index table that generates an inverted index table in which one or a plurality of contents are associated with each feature element constituting the content feature vector based on the content feature vector of the content The transposing index storage unit may further include a generating unit, and the transposed index storage unit may store the transposed index table generated by the transposed index table generating unit.

  According to the present invention, a transposed index table in which one or a plurality of contents are associated is generated for each feature element of the content feature vector, and the transposed index table is used to correspond to the feature elements of the history information content. Since the content is recommended, the amount of calculation can be reduced.

  In the recommendation device of the present invention, the feature element specifying means determines the priority among the feature elements of the content of the history information corresponding to the user terminal that requested the recommendation, and the recommendation determining means is determined by the feature element specifying means. You may make it determine a recommendation object based on the priority between feature elements.

  According to the present invention, the priority between the feature elements of the content of the history information is determined, and the recommendation target is determined based on the determined priority between the feature elements, so that the user characteristics derived from the history information are reflected more. You can make recommendations.

  In the recommendation device according to the present invention, the feature element specifying unit may determine the priority between the feature elements based on a plurality of feature elements corresponding to the content of the history information.

  Further, in the recommendation device of the present invention, weighting is performed for each feature element, and the feature element specifying means is based on the weight of the feature element corresponding to the content of the history information and the duplication number of the feature elements. Priorities may be defined.

  In the recommendation device according to the present invention, the recommendation determination unit may determine a recommendation target based on the overlapping number of contents searched by the recommendation candidate search unit.

  According to the present invention, priority is given to the content including many characteristic elements of the content of the history information, so that it is possible to make a recommendation that reflects the user characteristics derived from the history information.

  According to the present invention, it is possible to make recommendations according to user characteristics while reducing the amount of calculation.

1 is a system configuration diagram of a system 1. FIG. 2 is a block diagram showing functional configurations of a terminal device 20 and an information distribution server 30. FIG. 2 is a hardware configuration diagram of an information distribution server 30. FIG. It is a figure which shows the example of data which the content memory | storage part 310 hold | maintains. It is an example of the data which the content feature vector memory | storage part 311 hold | maintains. It is a figure which shows the specific example of an inverted index table. It is a figure which shows the example of data which the log | history memory | storage part 313 hold | maintains. It is a figure which shows the content of log | history information, and the extraction result of the characteristic. It is a figure explaining the appearance frequency of feature ID. It is a figure which shows the priority based on the appearance frequency of feature ID. It is explanatory drawing which shows the content corresponding to feature ID. It is a figure which shows the total result of the duplication of the content corresponding to feature ID. It is the figure which calculated the duplication of the content corresponding to feature ID, and excluded the duplication location. It is a figure which shows recommendation order. It is a flowchart which shows the process which carries out recommendation output. It is a flowchart which shows the process which produces | generates a transposed index.

  Embodiments of the present invention will be described with reference to the accompanying drawings. Where possible, the same parts are denoted by the same reference numerals, and redundant description is omitted.

  FIG. 1 is a system configuration diagram of a system 1 using an information distribution server 30 which is a recommendation device in the present embodiment. As shown in FIG. 1, the information distribution server 30 can communicate with a mobile terminal 20 that is a user terminal via a network and transmit content in accordance with user preferences of the mobile terminal 20. The mobile terminal 20 is a mobile phone or a smartphone, but is not limited to this, and may be a personal computer or a tablet terminal.

  FIG. 2 is a block diagram illustrating functional configurations of the mobile terminal 20 and the information distribution server 30 according to the present embodiment. The portable terminal 20 includes a history transmission unit 201, a recommendation acquisition request transmission / reception unit 202, and a distribution information display unit 203. The information distribution server 30 includes a content collection unit 301, a content feature vector generation unit 302, a transposed index generation unit 303 (a transposition index table generation unit), a history collection unit 304, a recommendation acquisition request reception unit 305, and a user feature specification unit 306. (Feature element specifying means), recommendation candidate search section 307 (recommendation candidate search means), recommendation determination section 308 (recommendation determination means), recommendation output section 309 (recommendation output means), content storage section 310, content feature vector storage section 311 A transposed index storage unit 312 (transposed index storage unit) and a history storage unit 313 (history information storage unit) are included.

  FIG. 3 is a hardware configuration diagram of the information distribution server 30. As shown in FIG. 3, the information distribution server 30 shown in FIG. 2 physically includes one or a plurality of CPUs 11, a RAM 12 and a ROM 13 as main storage devices, and an input device 14 such as a keyboard and a mouse as input devices. The computer system includes an output device 15 such as a display, a communication module 16 that is a data transmission / reception device such as a network card, and an auxiliary storage device 17 such as a hard disk or a semiconductor memory. Each function of the information distribution server 30 in FIG. 2 is configured such that predetermined computer software is read on the hardware such as the CPU 11 and the RAM 12 shown in FIG. 3, so that the input device 14, the output device 15, This is realized by operating the communication module 16 and reading and writing data in the RAM 12 and the auxiliary storage device 17. Hereinafter, each functional block will be described based on the functional blocks shown in FIG.

  First, functional blocks of the mobile terminal 20 will be described. The history transmission unit 201 transmits history information, which is information related to content browsed in the past, to the information distribution server 30 via the network.

  The recommendation acquisition request transmission / reception unit 202 makes a recommendation request to the information distribution server 30, and acquires recommendation information (distribution information) from the information distribution server 30.

  The distribution information display unit 203 outputs recommendation information that is information indicating recommended content to the mobile terminal 20. Specifically, it is a display.

  Next, the information distribution server 30 will be described. The content collection unit 301 collects content from a server (not shown) that manages content. The content collection unit 301 stores the collected content (content metadata or the like) in the content storage unit 310.

  The content feature vector generation unit 302 is a part that generates a content feature vector based on each content stored in the content storage unit 310. The generated content feature vector is stored in the content feature vector storage unit 311. The content feature vector generation unit 302 generates a content feature vector when content registration is performed by a content provider, that is, when content and its metadata are received from the content provider.

  Here, generation of content feature vectors will be described with reference to the drawings. An example of data held by the content storage unit 310 is shown in FIG. The content storage unit 310 stores content information in a table format, and includes content ID, category, title, and text as content information. For example, the content with the content ID “C01” has a category of “CD”, a title of “XX Complete Best”, and a main text of “Cost of attention coming to Japan for the first time in 10 years in May. ○○ .... "

  The content feature vector generation unit 302 acquires content table information from the content storage unit 310 and obtains a content feature vector for each content. Specifically, the content feature vector generation unit 302 performs morphological analysis on the content information, and whether or not it includes a predetermined keyword group that is determined in advance (if the keyword group is included, the number of times may be considered). ) To obtain the content feature vector of each content.

  Note that the keyword group may be defined in a program that functions as the content feature vector generation unit 302, or the information distribution server 30 may hold information in which the keyword group is databased. .

  After obtaining the content feature vector of each content, the content feature vector generation unit 302 stores content feature vector information, which is information in which the content is associated with the content feature vector, in the content feature vector storage unit 311. An example of the content feature vector information stored in the content feature vector storage unit 311 is shown in FIG. As shown in FIG. 5, the content feature vector storage unit 311 stores information in which a content ID and a content feature vector are associated with each other. The content feature vector has a plurality of feature IDs that are identifiers of feature elements (for example, “finance”).

  The content feature vector information is information indicating what feature elements each content has. If a flag is set in the content feature ID, it indicates that the content has a feature element of the feature ID. Specifically, “1” is set to the feature ID of the feature element if it has a feature element, and “0” is set to the feature ID of the feature element if it has no feature element. The In the present embodiment, the case where the feature elements corresponding to the feature ID are “finance”, “travel”, “TV”, “book”, “movie”, “music”, “soccer”, “art” will be described as an example. To do. For example, when the content feature vector of the content with the content ID “C01” is obtained, the content feature vector generation unit 302 has the “category”, “title”, and “text” of the content information with the content ID “C01” illustrated in FIG. As a result of the morphological analysis, “CD” is included in the category, and “CM song” and “music” are included in the text. Therefore, it is determined that the music has a feature ID, and a flag is set for the feature ID “T6”.

  The transposed index generation unit 303 generates a transposed index using the content feature vector information held by the content feature vector storage unit 311. Specifically, the transposed index generation unit 303 uses the content feature vector information to obtain one or a plurality of content IDs having the feature ID (feature ID value is 1) for each feature ID. Identify. In this way, the transposed index generation unit 303 generates a transposed index in which the feature ID and the content ID are associated with each other.

  FIG. 6 shows an example of an inverted index. As shown in FIG. 6, the feature ID and the content ID are associated with each other. For example, the feature ID “T02” indicates that it corresponds to the content ID “C03”. As described above, by storing the transposed index in which the feature ID and the content ID are associated in advance, the content corresponding to the feature ID can be easily searched using the feature ID as a search key.

  The history collection unit 304 collects history information from the mobile terminal 20 at a predetermined timing (for example, when the mobile terminal 20 browses content). The history collection unit 304 stores the collected history information in the history storage unit 313. An example of data held in the history storage unit 313 is shown in FIG. As illustrated in FIG. 7, the history information includes a user ID, a content ID, an operation type, and an operation date / time.

  For example, the user U001 has browsed the content with the content ID “C02” at 12:00 on September 11, 2013.

  The recommendation acquisition request receiving unit 305 receives a recommendation acquisition request from the mobile terminal 20.

  The user feature specifying unit 306 is a part that specifies the user's feature. Specifically, the user feature specifying unit 306 acquires history information corresponding to the mobile terminal 20 from the history storage unit 313, and selects a feature element having a high user preference based on the feature element of the content of the acquired history information. By specifying, the characteristics of the user are specified. The user feature specifying unit 306 also determines the priority order between the feature elements when specifying the user's feature.

  Here, a method in which the user feature specifying unit 306 specifies a user feature when a recommendation acquisition request is made from the mobile terminal 20 corresponding to the user ID “U001” will be described with reference to FIGS.

  First, the user feature specifying unit 306 refers to the content storage unit 310 and extracts three content items (for example, the latest three items) that the user ID “U001” referred to in the past. Note that the number of cases to be extracted is predetermined by the information distribution server 30.

  When the history information illustrated in FIG. 7 is included, the user feature specifying unit 306 extracts contents with content IDs “C02”, “C04”, and “C01”. Then, the user feature specifying unit 306 refers to the content feature vector storage unit 311 and specifies a feature ID corresponding to the extracted content ID.

  FIG. 8 shows the content ID extracted by the user feature specifying unit 306 and the feature ID corresponding to the content ID. As illustrated in FIG. 8, the history information content of the user ID “U001” is the content ID “C02”, “C04”, and “C01”, and the content ID “C02” content is the feature ID “C02”. The content ID “C04” corresponds to the feature IDs “T4” and “T8”, and the content ID “C01” corresponds to the feature ID “T6”.

  The user feature specifying unit 306 calculates feature IDs based on the information shown in FIG. Specifically, the user feature specifying unit 306 calculates the number of appearances for each feature ID. Since the feature ID “T5” corresponds only to the content ID “C02”, the number of appearances is one. Since the feature ID “T8” corresponds to the content IDs “C02” and “C04”, the number of appearances is two. Since the feature ID “T4” corresponds to the content ID “C04”, the number of appearances is one. The calculation result is shown in FIG.

  As illustrated in FIG. 9, the feature ID “T5” has one appearance, the feature ID “T8” has two appearances, and the feature ID “T4” has one occurrence. Thus, the feature ID “T6” appears once. Note that the feature ID “T8” described in the fourth row is collected into the feature ID “T8” in the second row.

  The user feature specifying unit 306 determines the priority based on the number of appearances shown in FIG. As illustrated in FIG. 10, the user feature specifying unit 306 assigns the feature ID “T5”, “T4”, and “T6” of the appearance count one time to the feature ID “T8” of the appearance count two times as the highest priority. Prioritize in order. As described above, the user feature specifying unit 306 specifies the user's favorite tendency based on the feature of the content of the history information. In addition, when a weighting coefficient is set for each feature ID, the user feature specifying unit 306 gives priority based on a result of multiplying the weighting coefficient by the number of appearances or a result of adding the number of appearances to the weighting coefficient. The degree may be determined. That is, the priority may be determined based on the weighting coefficient and the number of appearances.

  The recommendation candidate search unit 307 searches for the content corresponding to the feature ID specified by the user feature specifying unit 306 using the transposed index held by the transposed index storage unit 312.

  FIG. 11 shows the result of searching for the content corresponding to each feature ID shown in FIG. As illustrated in FIG. 11, the content corresponding to the feature ID “T8” is content with content IDs “C02”, “C04”, and “C05”. The content corresponding to the feature ID “T5” is the content with the content ID “C02”. The content corresponding to the feature ID “T4” is the content with the content ID “C04”. The content corresponding to the feature ID “T6” is the content with the content IDs “C01” and “C03”. Since the transposed index storage unit 312 holds the transposed index in advance, the recommendation candidate search unit 307 can search the content having the user's favorite features based on the feature transposed index.

  The recommendation determination unit 308 is a part that determines a recommendation target from the content searched by the recommendation candidate search unit 307. Specifically, the recommendation determination unit 308 determines a recommendation target (including a priority order when making a recommendation) based on the priority of the feature ID and the overlapping number of contents.

  An example of determining a recommendation target will be described with reference to FIGS. First, the recommendation determining unit 308 calculates the overlapping number of contents searched by the recommendation candidate searching unit 307. FIG. 12 is a table showing the result of calculating the duplicate number of contents searched by the recommendation candidate search unit 307. As shown in FIG. 12, since the content ID “C02” corresponds to the features T8 and T5, the overlapping number of the content ID “C02” is 2. Further, since the content ID “C04” corresponds to the feature ID “T8” and “T4”, the overlapping number of the content ID “C04” is 2.

  In the table shown in FIG. 12, when the column of feature IDs with low priority is deleted for the overlapping contents C02 and C04, the result is as shown in FIG. That is, the feature ID of priority 1 is “T8”, and the content IDs of the contents corresponding to the feature ID “T8” are “C02”, “C04”, and “C05”. The feature ID of priority 4 is “T6”, and the content IDs of the contents corresponding to the feature ID “T6” are “C01” and “C03”.

  Based on the table shown in FIG. 13, the recommendation determining unit 308 determines the content to be recommended based on the generated list in the order of the overlap of the content ID and the priority of the feature ID.

Specifically, the recommendation determining unit 308 makes selections as in the following (1) and (2).
(1) A content ID having the highest number of times of duplication is selected from among the contents in the feature ID having the highest priority (if the number of times of duplication is the same, the content of the latest history is given priority).
(2) (2-1) The following process (2-2) is repeated until the number of contents to be recommended to the user (the number of recommendation targets (5 in the present embodiment)) is reached.
(2-2) A content ID that has a large number of times of duplication and is not selected is selected from the feature IDs of the next priority. If there is no next priority, the process moves to the feature ID with the highest priority, and the content ID with the highest number of duplications is selected from among the contents in the priority ID with the highest priority. .

  As described above, the recommendation determining unit 308 determines content having different features as recommendation targets by selecting one from each feature ID. In this case, the information distribution server 30 can give variations to the features of the recommended content. Note that the recommendation determination unit 308 may recommend content corresponding to a feature ID having a high priority.

  FIG. 14 shows a table in which the results selected by the above method are added to the table of FIG. As illustrated in FIG. 14, the recommendation determining unit 308 first selects a content having a plurality of overlaps among the content corresponding to the feature ID “T8” having the highest priority of the feature ID. Here, the content ID “C02” with the overlap number of 2 is selected.

  Subsequently, the recommendation determining unit 308 selects a content having a plurality of overlapping contents from the contents of the next priority. Here, the content ID “C01” with the overlap number of 1 is selected.

  Subsequently, when the selection of content is repeated until the number of items to be recommended is reached, the recommendation determination unit 308 finally performs the order of content IDs “C02”, “C01”, “C04”, “C03”, “C05”. The priority is determined by and the recommendation target is determined.

  The recommendation output unit 309 outputs the content targeted for recommendation determined by the recommendation determination unit 308 to the mobile terminal 20.

  Next, processing of the information distribution server 30 and the mobile terminal 20 configured as described above will be described. FIG. 15 is a flowchart showing a recommendation process by the information distribution server 30.

  First, before receiving a recommendation request from the mobile terminal 20, based on the content, the content collection unit 301, the content feature vector generation unit 302, and the transposed index generation unit 303 generate a transposed index, and the transposed index is converted into the transposed index. It memorize | stores in the memory | storage part 312 (step S1). Details of this processing will be described later.

  Further, the history collection unit 304 acquires content history information from the portable terminal 20 before receiving the recommendation request, and stores the history information in the history storage unit 313 (step S2). When the recommendation acquisition request receiving unit 305 receives a recommendation acquisition request from the predetermined mobile terminal 20 (step S3), the user feature specifying unit 306 acquires the history information of the user who requested the recommendation (step S4). When acquiring the history information, the user feature specifying unit 306 acquires the feature ID of the content of the history information by referring to the content feature vector storage unit 311 (step S5). When the required number of history information has not been acquired (step S6; NO), the process returns to step S5.

  When the required number of history information has been acquired (step S6; YES), the user feature specifying unit 306 prioritizes the feature ID (step S7), and the recommended candidate search unit 307 is held by the transposed index storage unit 312. The content corresponding to the feature ID is extracted using the predetermined feature ID specified by the user feature specifying unit 306 as a search key (step S8). If the recommendation candidate search unit 307 has not searched for a recommendation candidate using all feature IDs as search keys (step S9; NO), the process returns to step S8, and searches for recommendation candidates using all feature IDs as search keys. When the determination is made (step S9; YES), the recommendation determination unit 308 performs the processing of steps S10 to S12 to determine the content to be recommended.

  Specifically, the recommendation determining unit 308 determines whether or not there is duplication for each content. If there is duplication, the recommendation determination unit 308 deletes the content (step S10) and sorts the duplication in descending order (step S11). Based on the sorted result, the content to be recommended is determined (step S12).

  Then, the recommendation output unit 309 outputs the content targeted for recommendation determined by the recommendation determination unit 308 (step S13).

  Next, processing for generating an inverted index will be described with reference to the flowchart shown in FIG. In the information distribution server 100, when the content collection unit 301 receives content from a content provider, the content collection unit 301 stores the content in the content storage unit 310 (step S21). Then, the content feature vector generation unit 302 estimates the content features based on the content metadata keywords stored in the content storage unit 310 (step S22). The content feature vector generation unit 302 generates a content feature vector based on the estimation result of the content feature, and stores content feature vector information based on the content feature vector in the content feature vector storage unit 311. Then, the transposed index generation unit 303 generates a transposed index based on the content feature vector information stored in the content feature vector storage unit 311 and stores the transposed index in the transposed index storage unit 312 (step S23). .

  Below, the effect of the information delivery server 30 of this embodiment is demonstrated. According to the information distribution server 30 of the present embodiment, the transposed index storage unit 312 stores a transposed index table in which one or a plurality of contents are associated with each feature element constituting the content feature vector of the content, and history storage. When the unit 313 stores the history information of the content accessed by each user terminal and the user feature specifying unit 306 makes a recommendation request from the portable terminal 20, the recommendation request is included in the history information stored in the history storage unit 313. The feature element of the history information corresponding to the mobile terminal 20 is identified, and the recommendation candidate search unit 307 uses the transposed index table with the content corresponding to the feature element identified by the user feature identification unit 306 as the recommendation candidate. The recommendation determining unit 308 searches for the recommendation Search section 307 determines a recommendation target from the searched content, the recommendation output unit 309 outputs the content of the recommendation subject recommendation determination unit 308 has determined.

  Thereby, the information distribution server 30 stores a transposed index table in which one or a plurality of contents are associated in advance, and uses the transposed index table to select content corresponding to the feature element of the history information content as a recommendation candidate. Since the content to be recommended is determined from the recommendation candidates, the amount of calculation can be reduced as compared with the case where the content of the history information and each content are compared and recommended. Thus, by storing the transposed index table in advance, it is possible to make a recommendation by real-time processing.

  Further, in the information distribution server 30 of the present embodiment, the transposed index generation unit 303 transposes one or a plurality of contents associated with each feature element constituting the content feature vector based on the content feature vector of the content. An index table may be generated, and the transposed index storage unit 312 may store the transposed index table generated by the transposed index generating unit 303.

  As a result, a transposed index table in which one or a plurality of contents are associated with each feature element of the content feature vector is generated, and the content corresponding to the feature element of the history information content is recommended using the transposed index table. Therefore, the amount of calculation can be reduced.

  Further, in the information distribution server 30 of the present embodiment, the user feature specifying unit 306 determines the priority among the feature elements of the content of the history information corresponding to the mobile terminal 20 that requested the recommendation, and the recommendation determining unit 308 The recommendation target is determined based on the priority between the feature elements determined by the feature specifying unit 306.

  As a result, the priority between the feature elements of the content of the history information is determined, and the recommendation target is determined based on the determined priority between the feature elements. Therefore, the recommendation that more reflects the user characteristics derived from the history information is performed. be able to.

  In the information distribution server 30 according to the present embodiment, the user feature specifying unit 306 may determine the priority between feature elements based on the duplication number of the feature elements corresponding to the history information content.

  Further, in the information distribution server 30 of the present embodiment, weighting is performed for each feature element, and the user feature specifying unit 306 determines the weighting of the feature element corresponding to the content of the history information, the number of appearances of the feature element (duplicate), and the like. The priority between the feature elements may be determined based on the above.

  Further, in the information distribution server 30 of the present embodiment, the recommendation determination unit 308 may determine a recommendation target based on the overlapping number of contents searched by the recommendation candidate search unit 307.

  As a result, priority is given to the content including many characteristic elements of the content of the history information, so that it is possible to make a recommendation reflecting the user characteristics derived from the history information.

  In the above-described embodiment, the recommendation determination unit 308 determines the priority order of the recommendation candidates based on the overlapping number of contents, and the case where the recommendation candidates are all recommended candidates based on the priority order has been described. A part of the recommendation candidates may be determined as a recommendation target based on the overlapping number. As a specific example, the recommendation deciding unit 308 selects a content whose content overlap is a predetermined value or more from among the recommendation candidates.

DESCRIPTION OF SYMBOLS 1 ... System, 20 ... Portable terminal, 30 ... Information delivery server, 201 ... History transmission part, 202 ... Recommendation acquisition request transmission / reception part, 203 ... Delivery information display part, 301 ... Content collection part, 302 ... Content feature vector generation part, 303: transposed index generation unit, 304 ... history collection unit, 305 ... recommendation acquisition request reception unit, 306 ... user feature identification unit, 307 ... recommendation candidate search unit, 308 ... recommendation determination unit, 309 ... recommendation output unit, 310 ... content Storage unit, 311 ... content feature vector storage unit, 312 ... transposed index storage unit, 313 ... history storage unit.

Claims (4)

An inverted index storage means for storing an inverted index table in which one or a plurality of contents are associated with each feature element constituting the content feature vector of the content;
History information storage means for storing history information of content accessed by each user terminal;
When a recommendation request is made from the user terminal, a feature element specifying unit that specifies a feature element of the content of the history information corresponding to the user terminal that requested the recommendation among the history information stored in the history information storage unit;
Recommendation candidate search means for searching using a transposed index table, with the content corresponding to the feature element specified by the feature element specifying means as a recommendation candidate,
Recommendation determination means for determining a recommendation target from the content searched by the recommendation candidate search means;
Recommendation output means for outputting the content to be recommended determined by the recommendation determination means;
Equipped with a,
The feature element specifying means determines the priority of each feature element based on the number of appearances of each specified feature element,
The recommendation determining means is:
Calculating the number of duplicate contents searched by the recommendation candidate search means;
Among the contents related to the feature element with the highest priority, the content with the highest overlap is determined as a recommendation target as the first priority,
The recommendation apparatus which determines the content with the highest duplication number as the second priority among the contents related to the feature element with the second highest priority as a recommendation target . Based on the content feature vector of the content, further comprising a transposed index table generating means for generating a transposed index table in which one or a plurality of contents are associated with each feature element constituting the content feature vector,
The recommendation apparatus according to claim 1, wherein the transposed index storage unit stores the transposed index table generated by the transposed index table generating unit. Each feature element is weighted,
The feature element specifying means includes a weighting of feature elements that correspond to the content of the history information, based on said number of occurrences of times of feature elements, defining the priority between feature elements,
The recommendation apparatus of Claim 1 or 2 . In the recommendation method in the recommendation device,
A transposed index storage step of storing a transposed index table in which one or a plurality of contents are associated with each of the feature elements constituting the content feature vector of the content in the transposed index storage unit;
A history information storage step for storing history information of content accessed by each user terminal in the history information storage means;
When a recommendation request is made from the user terminal, a feature element specifying step for specifying a feature element of the content of the history information corresponding to the user terminal that requested the recommendation among the history information stored in the history information storage unit;
A recommendation candidate search step of searching using a transposed index table as a candidate for content corresponding to the feature element specified in the feature element specifying step;
A recommendation determination step for determining a recommendation target from the content searched in the recommendation candidate search step;
A recommendation output step for outputting the content to be recommended determined in the recommendation determination step;
Equipped with a,
In the feature element specifying step, the priority of each feature element is determined based on the number of appearances of each specified feature element,
In the recommendation determining step,
Calculate the duplicate of the content searched in the recommendation candidate search step,
Among the contents related to the feature element with the highest priority, the content with the highest overlap is determined as a recommendation target as the first priority,
A recommendation method in which, among the contents relating to the feature element having the second highest priority, the content having the highest overlapping number is determined as a recommendation target as the second priority .

Images