JP2010049498A - Information retrieval device, information retrieval method, and information retrieval program - Google Patents

Abstract

<P>PROBLEM TO BE SOLVED: To provide an information retrieval device that enables information retrieval reflecting a user's intention. <P>SOLUTION: The information retrieval device includes: a preference evaluation part 31 for calculating a preference evaluation score for each piece of information to be retrieved; a topicality evaluation part 32 for calculating a topicality evaluation score; a novelty evaluation part 33 for calculating a novelty evaluation score; a recommendability calculation part 34 for calculating recommendability of information by the formula of preference evaluation score×priority α+topicality evaluation score×priority β+novelty evaluation score×priority γ; a recommendation result display part 17 for displaying a recommendation result on a screen; an attention evaluation detection part 19 that when a user operation for the information of the recommendation result is detected, extracts each evaluation score related to the information, detects an attention evaluation item to which a user's attention is paid based on the result of the comparison of the extracted evaluation scores, and counts the number of times of attention evaluations and the accumulated number of times of operations; and a priority setting part 23 for calculating a priority for each evaluation item based on the number of times of attention evaluations and the accumulated number of times of operations, and updating the priorities in a priority control table 15 with the calculated priorities. <P>COPYRIGHT: (C)2010,JPO&INPIT

Description

  The present invention relates to an information search apparatus, an information search method, and an information search program for searching information reflecting a user's intention from a plurality of pieces of search target information.

  In recent years, TV programs (or content distributed over the Internet) such as PCs (Personal Computers) equipped with TV functions and PVRs (Recorders with HDDs and DVDs: Personal Video Recorders) are recorded and played back. Information recording devices are widely used.

  This information recording device has various functions to improve user convenience, for example, using an electronic program guide (metadata) distributed from EPG (Electric Program Guide), It has a function of automatically searching and presenting a user-preferred program and a function of automatically recording the searched program.

  Therefore, in recent years, an information search apparatus that recommends information such as news, programs, products, and the like that the user likes has been proposed.

  In this information search device, user preference search conditions and operation history (viewing, playback, recording, purchase, etc.) for user information are analyzed and extracted, user preference information is managed, and user preference and search target information Are compared and evaluated, and user preference information is searched and recommended from a plurality of pieces of information.

  However, in the information search apparatus, there are various evaluation items as well as user preferences from the viewpoint of searching or recommending information.

  For example, when the information to be searched is news, there are many cases in which information is searched by various evaluation items so that the news to be searched is searched from the viewpoint of topicality or novelty.

  Therefore, in order to cope with such a demand, a technique has been proposed in which information to be searched is comprehensively evaluated with various evaluation items and searched and recommended from a plurality of pieces of information.

  For example, an information search apparatus that comprehensively evaluates news using evaluation items such as user preference and topicality, and recommends news with user preference and topicality from a huge amount of news by the comprehensive evaluation. is there.

  In such an information search device, pre-profiled user preferences and topicality set in advance for individual news are set, and preference evaluation scores and topicality evaluation scores for the news to be searched are calculated. The recommendation level of news is comprehensively evaluated using the priority indicating the priority ratio of the evaluation score for the evaluation score for preference and the evaluation score for topicality.

  For example, when the priority of the preference evaluation score is α1 and the priority of the topical evaluation score is β1, the information search apparatus weights the priority to each evaluation score, and the preference evaluation score × priority. The recommendation degree of news is calculated by the formula of α1 + topic evaluation score × priority β1.

  Therefore, in the information retrieval apparatus, the recommendation degree of each news is sequentially calculated by the formula of preference evaluation score × priority α1 + topic evaluation score × priority β1. Based on this, it is possible to search for news with preference and topicality from a plurality of news.

JP 2003-157278 A

  However, in the above-described conventional information retrieval apparatus, the priorities α1 and β1 are set in the preference and topicality evaluation items, and the information on the basis of the preference score, the topical evaluation score, and the priorities α1 and β1 However, since the priority set for the evaluation items such as preference and topicality is fixed, the preference varies depending on the content of the information. There are individual differences in whether or not to prioritize items, and it is difficult for individual users to set an appropriate priority for each evaluation item.

  Therefore, the present invention has been made in view of the above points, and the purpose of the present invention is to appropriately set the priority for each evaluation item when searching for information, and reflect the user's intention. An object of the present invention is to provide an information search apparatus, an information search method, and an information search program that enable information search.

  The disclosure device includes an information acquisition unit that acquires metadata related to information to be searched, a preference management unit that stores a search condition for searching for the information to be searched, and the meta data for each information to be searched. A preference evaluation unit that calculates a preference evaluation score as a preference evaluation item based on a similarity between data and the search condition stored in the preference management unit, and evaluation of the preference for each piece of information to be searched Another type evaluation unit that calculates an evaluation score as an evaluation item other than the item, a priority management unit that stores a priority indicating a priority ratio of the evaluation score for each evaluation item of each evaluation score, and the search target The evaluation score for each evaluation item related to information is weighted according to the priority for each evaluation item stored in the priority management unit, and based on this weighted evaluation score, the evaluation result related to the information to be searched Each evaluation score is extracted from information on the evaluation result related to the user operation detected by the operation detection unit, an operation detection unit that detects a user operation on the evaluation result information to be output, and these Comparing the extracted evaluation scores, based on the comparison results, based on the attention evaluation item detected by the attention evaluation detection unit that detects the attention evaluation item corresponding to the user attention evaluation item, and the attention evaluation item detected by the attention evaluation detection unit, It is a requirement to include a priority setting unit that calculates the priority for each evaluation item and updates the calculated priority for each evaluation item to the priority management unit.

  In addition, the disclosure method includes an information acquisition step for acquiring metadata relating to information to be searched, and a preference for each piece of information to be searched based on a similarity between the metadata and a search condition stored in the preference management unit. A preference evaluation step for calculating a preference evaluation score as a sex evaluation item, another type evaluation step for calculating an evaluation score as an evaluation item other than the preference evaluation item for each piece of information to be searched, and the search target The evaluation score for each evaluation item related to the information is weighted according to the priority for each evaluation item stored in the priority management unit that stores the priority indicating the priority ratio of the evaluation score, and the weighted evaluation score An evaluation output step for outputting an evaluation result related to the information to be searched, an operation detection step for detecting a user operation for the information on the evaluation result, Each evaluation score is extracted from the information of the evaluation result related to the user operation detected in the operation detection step, the extracted evaluation scores are compared, and based on this comparison result, the attention evaluation corresponding to the evaluation item of user attention Attention evaluation detection step for detecting an item, and based on the attention evaluation item detected at the attention evaluation detection step, the priority for each evaluation item is calculated, and the priority for each evaluation item calculated is the priority And a priority setting step to be updated in the management unit.

  Further, the disclosure program causes the computer apparatus to acquire an information acquisition procedure for acquiring metadata relating to information to be searched, and the similarity between the metadata and the search condition stored in the preference management unit for each information to be searched. A preference evaluation procedure for calculating a preference evaluation score as a preference evaluation item based on the degree, and another type evaluation procedure for calculating an evaluation score as an evaluation item other than the preference evaluation item for each piece of information to be searched The evaluation score for each evaluation item related to the information to be searched is weighted according to the priority for each evaluation item stored in the priority management unit that stores the priority indicating the priority ratio of the evaluation score. An evaluation output procedure for outputting an evaluation result related to the information to be searched based on a weighted evaluation score, and an operation detection procedure for detecting a user operation on the information of the evaluation result Each evaluation score is extracted from the information of the evaluation result related to the user operation detected in the operation detection procedure, the extracted evaluation scores are compared, and the attention corresponding to the evaluation item of user attention is based on the comparison result Based on the attention evaluation detection procedure for detecting the evaluation item and the attention evaluation item detected in the attention evaluation detection procedure, the priority for each evaluation item is calculated, and the priority for each evaluation item calculated is the priority. And a priority setting procedure to be updated by the degree management unit.

  In the disclosure device, the disclosure method, and the disclosure program, the evaluation score for each evaluation item related to the search target information is weighted according to the priority for each evaluation item stored in the priority management unit, and the weighted evaluation score is used. And outputting an evaluation result related to the search target information, extracting each evaluation score from the evaluation result information related to the user operation, comparing the extracted evaluation scores, and based on the comparison result, the user An attention evaluation item corresponding to the attention evaluation item is detected.

  As a result, the disclosure device, the disclosure method, and the disclosure program calculate the priority for each evaluation item based on the detected attention evaluation item, and update the calculated priority for each evaluation item to the priority management unit. As a result, when searching for information, the priority for each evaluation item is appropriately set for each user according to the user's operational tendency, and the information search reflecting the user's intention can be achieved.

  Embodiments relating to an information search apparatus, an information search method, and an information search program of the present invention will be described below in detail with reference to the drawings.

  First, the outline of the present embodiment will be described. The outline relates to an information search apparatus that searches a plurality of information such as programs and news, and each evaluation score related to information to be searched is prioritized for each evaluation item. A recommendation result including a recommendation degree of information is output based on the weighted evaluation score.

  Then, in the information search device, when detecting a user operation on the recommended result information, each evaluation score related to the user operation information is compared, and based on the comparison result, the attention evaluation item to which the user is interested is determined. Based on the number of attention evaluation items (number of attention evaluations), the priority for each evaluation item is calculated, and the calculated priority is dynamically updated.

  As a result, in the information search device, the priority for each evaluation item is appropriately set for each user according to the user's operation tendency, and the information search reflecting the user's intention can be performed.

  FIG. 1 is a block diagram illustrating a schematic configuration inside the information search apparatus according to the first embodiment.

  An information search apparatus 1 shown in FIG. 1 includes an information acquisition unit 11 that acquires information such as news and programs and metadata of the information, and a search condition database 12 that stores search conditions for searching for information to be searched. And a preference management unit 13 that manages the search condition database 12.

  The program metadata (EPG) includes data such as the broadcast start date and time, broadcast length, broadcast station, genre, title, description, and delivery deadline for the program. Further, the news metadata includes data such as a delivery start date and time, a delivery source, a genre, a title, a description, a delivery deadline, and the like regarding the news.

  In addition, the information acquisition unit 11 acquires, for example, from incoming broadcast radio waves or a network in the case of program metadata, and acquires from, for example, a net video or news distribution source in the case of news metadata. .

  In addition, the preference management unit 13 explicitly acquires a keyword or the like according to an input operation on the user side, and automatically acquires a keyword or the like from a user's operation history (viewing or recording of a program, browsing of news, etc.) Then, the acquired keyword is stored in the search condition database 12 as a search condition. Note that when the keyword is automatically acquired from the operation history, the preference management unit 13 analyzes the metadata of the operation target information in the operation history, and acquires search conditions such as important keywords, genre, and broadcast length. Is.

  Further, the information search device 1 comprehensively evaluates the evaluation items of the information to be searched for each piece of metadata of the information to be searched acquired by the information acquisition unit 11, and recommends the degree of recommendation to recommend information to the user side. An information recommendation unit 14 to be provided, a priority management table 15 storing priority used for weighting each evaluation score for each evaluation item used by the information recommendation unit 14, and a priority management for managing the priority management table 15 Part 16.

  The information recommendation unit 14 evaluates the objective topicality of public opinion regarding the information on the search target and the preference evaluation unit 31 that calculates the preference evaluation score for evaluating the user's subjective preference regarding the information on the search target. It has a topicality evaluation unit 32 that calculates a topicality evaluation score, and a novelty evaluation unit 33 that calculates a novelty evaluation score that evaluates the novelty of information regarding information to be searched.

  The preference evaluation unit 31 uses, for example, a vector space method to analyze the metadata of information to be searched and the search conditions of the preference management unit 13, and search information index (vector) and search condition index (Vector) is created. In the metadata analysis, texts such as titles, descriptions, and genres in the metadata are decomposed into keywords by morphological analysis or n-gram analysis.

  The preference evaluation unit 31 is divided into keywords and weighted by using statistics such as the appearance frequency of each keyword to create an index (vector) of information to be searched and an index (vector) of search conditions. is there.

  The statistics are calculated by TF / DF using, for example, TF (Term Frequency) or DF (Document Frequency). TF represents the frequency of occurrence of a specific keyword in the text in the information metadata, and DF represents the number of information including the specific keyword for all information such as search targets and search conditions. It represents.

  In the vector space method, the decomposed and extracted keywords are used as vector bases, and the statistic of each keyword is used as the element value of the vector.

  FIG. 2 is an explanatory diagram conceptually showing a method for calculating a preference evaluation score, and FIG. 3 is an explanatory diagram conceptually showing a method for calculating a novelty evaluation score.

  As shown in FIG. 2, the preference evaluation unit 31 calculates a preference evaluation score based on the similarity between an index (vector) of information to be searched and an index (vector) of a preference search condition. Note that the similarity between vectors is expressed by the inner product of both vectors, the angle formed, or the correlation, and the similarity is higher as the information meets the user's preference.

  Similarly to the preference evaluation, the topicality evaluation unit 32 creates an index (vector) of each piece of information to be searched using the vector space method, calculates the similarity between these created vectors, When the degree is equal to or greater than a specified value, these pieces of information are determined as the same topic, and the degree of similarity and the number of cases are stored.

  Furthermore, the topicality evaluation unit 32 calculates a topicality evaluation score by multiplying the average value of the similarities during storage and the number of cases adjusted and evaluated by a LOG function or the like.

  Further, as shown in FIG. 3, the novelty evaluation unit 33 determines whether the broadcast date / time or distribution date / time is the evaluation date / time based on the difference between the broadcast date / time and distribution date / time of the information to be searched and the evaluation date / time when the evaluation is performed. The closer to, the higher the novelty is evaluated, and the evaluation date / time is used as a reference (f (x | x = evaluation date / time) = 1.0). An evaluation function that approaches (x | x = evaluation date and time) = 1.0 is determined, and f (x) is used as an evaluation score for novelty.

  For example, in the case of news, the novelty evaluation unit 33 evaluates that the news is new and the novelty is higher as the distribution date of the news is closer to the evaluation date. Conversely, the farther the news delivery date is from the evaluation date, the older the news and the lower the novelty.

  In the case of a program, the novelty evaluation unit 33 evaluates that the program is broadcast soon as the broadcast date and time of the program is closer to the evaluation date and is highly novel. Conversely, the farther the broadcast date / time is from the evaluation date / time, the program is broadcast after a while and is evaluated as having low novelty. Note that the broadcast date / time of the program is not the broadcast date / time after the end of the broadcast, but the broadcast start date / time.

  The information recommendation unit 14 includes a preference evaluation score calculated by the preference evaluation unit 31, a topicality evaluation score calculated by the topicality evaluation unit 32, a novelty evaluation score calculated by the novelty evaluation unit 33, and a priority level. The management table 15 includes a recommendation level calculation unit 34 that calculates a recommendation level of information based on the priority for each evaluation score stored in the management table 15.

  The recommendation level calculation unit 34 weights priorities for each evaluation score, and recommends information using a formula of preference evaluation score × priority α + topic evaluation score × priority β + newness evaluation score × priority γ. The degree is calculated. Note that the priority for each evaluation score is not fixed but dynamically changed, as will be described later.

  In addition, the information search apparatus 1 illustrated in FIG. 1 detects a recommended result display unit 17 that displays the recommended result of the information recommendation unit 14 on the screen, and a user operation on the recommended result information displayed on the recommended result display unit 17. The operation detection unit 18, the attention evaluation detection unit 19 that detects the evaluation item to be noticed on the user side, that is, the attention evaluation item, regarding the information on the recommended result related to the user operation detected by the operation detection unit 18, and the attention evaluation detection unit And an attention evaluation management table 20 for storing the detection results detected at 19.

  FIG. 4 is an explanatory diagram briefly showing a recommended result during screen display on the recommended result display unit 17.

  As shown in FIG. 4, the recommended result display unit 17 displays a recommended result 40 related to each information to be searched on the screen, and the recommended result 40 includes a recommendation level, a preference evaluation score, a topicality evaluation score, a new It has a ranking based on sex evaluation score, news ID (program ID), delivery date, genre, news title (program title), and recommendation level.

  The recommended result display unit 17 displays a list of recommended results 40 related to information to be searched in descending order of recommendation. The recommended result display unit 17 may sequentially display the recommended results 40 in units of information without rearranging the recommended results 40 in the order of the recommended level or displaying a list in accordance with the setting. It goes without saying that it is good.

  In addition, when the operation detection unit 18 detects user operations such as viewing, storing, program viewing reservation, viewing reservation recording, copying, moving, and reference for the information of the recommended result 40 displayed on the recommended result display unit 17, For the information related to the recommended result 40, a processing operation corresponding to the operation content is executed.

  Further, when the operation detection unit 18 detects a user operation on the recommended result information displayed on the screen on the recommended result display unit 17, the operation detection unit 18 notifies the attention evaluation detection unit 19 of the operation history of the information. The operation history includes an operation content indicating the operation content and an operation date and time.

  When the attention evaluation detection unit 19 acquires the operation history from the operation detection unit 18, the recommendation level, the preference evaluation score, the topicality evaluation score, and the novelty evaluation of the information related to the operation being displayed on the screen from the recommendation result display unit 17 are obtained. An evaluation score such as a score is acquired.

  When the attention evaluation detection unit 19 acquires the operation history from the operation detection unit 18 and the recommendation level and the evaluation score related to the operation target information from the recommendation result display unit 17, the operation history, the recommendation level, and the evaluation score are obtained. The operation history information that is included is stored in the attention evaluation management table 20.

  The attention evaluation management table 20 includes an operation history information table 20A that stores operation history information, and an operation count table 20B that stores the number of operations described later.

  FIG. 5 is an explanatory diagram briefly showing the table contents of the operation history information table 20A in the attention evaluation management table 20.

  The operation history information table 20A shown in FIG. 5 stores operation history information related to information on each operation target. The operation history information includes a history number for identifying the operation history and news related to the operation history information. ID (program ID), operation content indicating the content of the operation, operation date / time, recommendation degree related to information of the operation history, preference evaluation score, topicality evaluation score, and novelty evaluation score of the information Etc. and an evaluation score.

  FIG. 6 is an explanatory diagram simply showing the contents of the operation count table 20B in the attention evaluation management table 20.

  The operation count table 20B shown in FIG. 6 stores operation count information. The operation count information includes a cumulative operation count M, a preference attention evaluation count P indicating the degree of attention of preference evaluation described later, and a later description. It has a topical attention evaluation count T indicating the attention level of the topicality evaluation, and a novelty attention evaluation count N indicating the attention level of the novelty evaluation described later.

  When acquiring the operation history from the operation detecting unit 18, the attention evaluation detecting unit 19 increments (counts) the total number of operations stored in the operation number table 20B.

  FIG. 7 is an explanatory diagram briefly showing the operation concept of the attention evaluation detecting unit 19.

  When the attention evaluation detection unit 19 acquires the recommendation level and the evaluation score related to the user operation information from the recommendation result display unit 17, as shown in FIG. 7, the attention evaluation detection unit 19 converts the preference evaluation score, topicality evaluation score, and novelty evaluation score. Based on the comparison result, the evaluation item of the maximum evaluation score is determined as the attention evaluation item that the user is paying attention to, and the operation count table 20B corresponding to the attention evaluation item is determined. Is incremented (counted) by +1.

  That is, as shown in FIGS. 7A and 7B, the attention evaluation detection unit 19 selects the evaluation item of the novelty evaluation score as the attention evaluation item when the novelty evaluation score is the maximum value among the evaluation scores. Is determined, and the novelty attention evaluation count in the operation count table 20B is incremented by +1.

  Further, the attention evaluation detection unit 19 compares the evaluation scores, and based on the comparison result, as shown in FIG. 7C, when there is no maximum evaluation score, the attention evaluation focused by the user The item is judged as none.

  The attention evaluation detection unit 19 manages and controls the attention evaluation management table 20, and the storage date and time of operation history information for the operation history information table 20A as a storage management condition in the operation history information table 20A of the attention evaluation management table 20 When the predetermined period has passed, the operation history information that has passed the predetermined period is deleted from the operation history information table 20A, and the accumulated operation count M stored in the operation count table 20B according to the deletion of the operation history information The number of attention evaluations P, T, and N for each evaluation item is recalculated.

  In addition, the recalculation of the cumulative operation count and the attention evaluation count is, for example, when the operation history information determined as the topical attention evaluation item is deleted, from the cumulative operation count M to −1, from the topical attention evaluation count T. Minus one. Further, for example, when the operation history information determined to have no attention evaluation item is deleted, the recalculation is to reduce the cumulative operation number M by −1 and maintain the attention evaluation number P, T, N. .

  In addition, the information search apparatus 1 illustrated in FIG. 1 includes a priority setting unit 23 that updates the priority for each evaluation score stored in the priority management table 15.

  The priority setting unit 23, through the attention evaluation detection unit 19, from the operation number table 20B in the attention evaluation management table 20, the cumulative operation number M, the preference attention evaluation number P, the topical attention evaluation number T, and the novelty attention evaluation number N. Is obtained, the priority α, β, γ of each evaluation score for each evaluation item is respectively determined based on the cumulative operation count M, the preference attention evaluation count P, the topical attention evaluation count T, and the novelty attention evaluation count N. Is to be calculated.

  The priority represents the probability for each evaluation item that is predicted to be operated when the user performs the next operation.

  The priority setting unit 23 applies Bernoulli trial Bayesian estimation, and evaluates the number of evaluation items X, the cumulative number of operations M, the number of attention evaluations E (i) of the evaluation item i (for example, preference attention evaluation number P, topical attention). Evaluation number T, novelty attention evaluation number N), total number of attention evaluation times S (= P + T + N) of all evaluation items, and number of times (MS) that attention evaluation detection unit 19 determines that there is no attention evaluation item , The priority of the evaluation item i = (1 + E (i) + (MS) / X) / (X + M).

  It should be noted that the number of times (M−S) determined that there is no attention evaluation item is considered to be that each evaluation item is equally noticed, and that number is divided by the number of evaluation items X (M−S) / X. This is added to the number of attention evaluations E (i) of the target evaluation item i.

  For example, the priority setting unit 23 sets the preference priority α to a formula of (1 + P + (MS) / 3) / (3 + M), and the topicality priority β to (1 + T + (MS) / 3). / (3 + M) and the novelty priority γ are respectively calculated by the formula (1 + N + (MS) / 3) / (3 + M).

  FIG. 8 is an explanatory diagram briefly showing the relationship between the cumulative operation count M and the attention evaluation count E (i) and the priority according to the first embodiment.

  In the initial state of the information retrieval apparatus 1, the evaluation item number X is “3”, the cumulative operation number M is “0”, the preference attention evaluation number P is “0”, the topical attention evaluation number T is “0”, When the novelty attention evaluation count N is set to “0”, the preference priority α is (1 + 0 + (0−0) / 3) / (3 + 0) ≈0.33, and similarly the topical priority β Is (1 + 0 + (0-0) / 3) / (3 + 0) ≈0.33. Similarly, the priority of novelty γ is (1 + 0 + (0-0) / 3) / (3 + 0) ≈0.33. It becomes.

  Further, when the cumulative operation count M is “10”, the preference attention evaluation count P is “0”, the topical attention evaluation count T is “0”, and the novelty attention evaluation count N is “8”, The priority α is (1 + 0 + (10−8) / 3) / (3 + 10) ≈0.13. Similarly, the topical priority β is (1 + 0 + (10−8) / 3) / (3 + 10) ≈ Further, the novelty priority γ is (1 + 8 + (10−8) / 3) / (3 + 10) ≈0.74, and attention is paid to novelty.

  Further, when the cumulative operation count M is “10”, the preference attention evaluation count P is “4”, the topical attention evaluation count T is “1”, and the novelty attention evaluation count N is “3”, The priority α is (1 + 4 + (10−8) / 3) / (3 + 10) ≈0.44, and the topical priority β is (1 + 1 + (10−8) / 3) / (3 + 10) ≈0.21. The priority γ of novelty is (1 + 3 + (10−8) / 3) / (3 + 10) ≈0.36, and attention is paid to preference and novelty.

  Further, when the cumulative operation count M is “10”, the preference attention evaluation count P is “3”, the topical attention evaluation count T is “3”, and the novelty attention evaluation count N is “4”, Priority α is (1 + 3 + (10−10) / 3) / (3 + 10) ≈0.31, and topical priority β is (1 + 3 + (10−10) / 3) / (3 + 10) ≈0.31, new The gender priority γ is (1 + 4 + (10−10) / 3) / (3 + 10) ≈0.38, and attention is paid equally.

  When the priority setting unit 23 calculates the preference priority α, the topical priority β, and the novelty priority γ, the priority management unit 16 uses the priority management unit 16 to manage the priority for each evaluation item. The table 15 is updated.

  Next, the operation of the information search device 1 according to the first embodiment will be described. FIG. 9 is a flowchart illustrating the processing operation of the information recommendation unit 14 related to the recommendation level providing process according to the first embodiment.

  In the recommendation level providing process illustrated in FIG. 9, the user's preference, topicality, and novelty are comprehensively evaluated to sequentially calculate the recommendation level of information to be searched, and the recommended result 40 of each information is provided to the user. It is processing.

  In FIG. 9, the information recommendation unit 14 acquires information to be searched and metadata of the information through the information acquisition unit 11 (step S11). The information recommendation unit 14 temporarily stores the acquired search target information and metadata.

  The information recommendation unit 14 acquires a search condition for evaluating the preference through the preference management unit 13 (step S12).

  The information recommendation unit 14 creates an index of information to be searched and search conditions (step S13), and the priority for each evaluation item stored in the priority management table 15 through the priority management unit 16, for example, preference. The priority α, the topical priority β, and the novelty priority γ are acquired (step S14).

  The information recommendation unit 14 determines whether or not the calculation of the degree of recommendation for all information to be searched is completed (step S15).

  The preference evaluation unit 31 in the information recommendation unit 14 determines the similarity between the search target information vector and the search condition vector when calculation of the recommendation level for all the search target information has not been completed (No in step S15). Based on the above, a preference evaluation score is calculated (step S16). The preference evaluation unit 31 calculates the similarity between vectors based on the inner product (the angle formed or the correlation coefficient) between the vector of information to be searched and the vector of the search condition, and the preference is calculated based on the calculated similarity. An evaluation score is calculated.

  Further, the topicality evaluation unit 32 inside the information recommendation unit 14 creates an index (vector) for each piece of information to be searched, and calculates a topicality evaluation score based on the similarity between the vectors of each piece of information to be searched. (Step S17). The topicality evaluation unit 32 calculates the similarity between vectors based on the inner product (the angle formed or the correlation coefficient) between the vectors of each information to be searched, and calculates the topicality evaluation score based on the calculated similarity. To do.

  Further, the novelty evaluation unit 33 in the information recommendation unit 14 calculates a novelty evaluation score that increases the evaluation score of the broadcast start date and time and distribution date and time closest to the reference evaluation date and time (step S18).

  Further, the recommendation level calculation unit 34 in the information recommendation unit 14 includes a preference evaluation score, a topicality evaluation score and a novelty evaluation score, a preference priority α, a topicality priority β and a novelty priority γ. Based on the above, the recommendation degree of information is calculated by a formula of preference evaluation score × priority α + topicality evaluation score × priority β + newness evaluation score × priority γ (step S19), and the recommendation degree for the information of all search targets In order to determine whether or not the calculation of has been completed, the process proceeds to step S15.

  If the calculation of the recommendation level for all the search target information is completed in step S15 (Yes in step S15), the information recommendation unit 14 recommends the total search target information, the evaluation score for each evaluation item, and the news ID (program). ID), distribution date and time, genre, and news title are displayed on the recommended result display unit 17 as recommended results (step S20), and the processing operation shown in FIG. As a result, the user can recognize the recommendation level of each information based on the recommendation result 40 related to all the search target information displayed on the screen in the recommendation result display unit 17, and the user of the information for the recommendation result 40 It enables operation.

  In the recommendation level providing process shown in FIG. 9, an evaluation score for each evaluation item is calculated for each information to be searched, a priority corresponding to the evaluation score for each evaluation item is acquired, and an evaluation score for each evaluation item is obtained. And based on the priority, the preference level score + priority level α + topicality evaluation score × priority level β + novelty level evaluation score × priority level γ, so that the recommendation level related to each information to be searched is calculated. It is possible to calculate the degree of recommendation of information to be searched corresponding to user preference, topicality, and novelty.

  Furthermore, in the recommendation level providing process, the recommendation result 40 including the recommendation level related to all information to be searched is displayed on the recommended result display unit 17, so that the user is displaying the recommended result display unit 17 on the screen. Based on the recommendation result 40 related to all information to be searched, the recommendation level of each information reflected in the user intention can be recognized.

  Next, the operation of the attention evaluation detection unit 19 inside the information search apparatus 1 will be described. FIG. 10 is a flowchart illustrating the processing operation of the attention evaluation detection unit 19 related to the attention evaluation detection process of the first embodiment.

  In the attention evaluation detection process shown in FIG. 10, when a user operation on the recommended result 40 information displayed on the screen is detected on the recommended result display unit 17, an evaluation score for each evaluation item is acquired from the recommended result 40 of the information. This is a process for detecting an attention evaluation item corresponding to a user operation tendency based on a comparison result of evaluation scores.

  In FIG. 10, when the recommended result display unit 17 detects the user operation of information related to the recommended result 40 being displayed on the screen, the attention evaluation detection unit 19 acquires each evaluation score for each evaluation item from the information (step S31). ). Note that the attention evaluation detection unit 19 acquires a preference evaluation score, a topic evaluation score, and a novelty evaluation score related to user operation information.

  The attention evaluation detecting unit 19 obtains each evaluation score for each evaluation item, compares the evaluation scores, and determines whether there is a difference between the evaluation scores (step S32).

  When there is a difference between the evaluation scores (Yes at Step S32), the attention evaluation detecting unit 19 extracts the maximum evaluation score from all the evaluation scores (Step S33). For example, in the case of FIG. 7A, the attention evaluation detecting unit 19 extracts the novelty evaluation score as the maximum value.

  When the maximum evaluation score is extracted, the attention evaluation detecting unit 19 determines that the evaluation item of the maximum evaluation score is the attention evaluation item (step S34), and increments the number of attention evaluations corresponding to the attention evaluation item by +1 ( Step S35), the processing operation shown in FIG. In the case of FIG. 7A, the novelty attention evaluation count corresponding to the novelty evaluation score evaluation item is incremented by +1.

  Further, when there is no difference between the evaluation scores in Step S32 (No in Step S32), the attention evaluation detecting unit 19 does not have the maximum evaluation score and there is no attention evaluation item as shown in FIG. The processing operation shown in FIG. 10 is terminated.

  In the attention evaluation detection process illustrated in FIG. 10, when a user operation on the information of the recommended result 40 being displayed on the screen is detected in the recommended result display unit 17, each evaluation score is acquired from the recommended result 40 of the information, and the acquired evaluation Of the scores, the evaluation item with the maximum evaluation score is determined as the attention evaluation item, and the number of attention evaluations of this evaluation item is incremented by +1. Can do.

  Next, the operation of the priority setting unit 23 in the information search apparatus 1 will be described. FIG. 11 is a flowchart illustrating the processing operation of the priority setting unit 23 related to the priority update process according to the first embodiment.

  The priority update process shown in FIG. 11 is based on the attention evaluation counts P, T, N and the cumulative operation count M for each evaluation item detected by the attention evaluation detection unit 19, and the priority α, β, γ for each evaluation item. And the calculated priority is updated in the priority management unit 16.

  The priority setting unit 23 acquires the cumulative operation count M on the user side stored in the operation count table 20B in the attention evaluation management table 20 through the attention evaluation detection unit 19 (step S41).

  The priority setting unit 23 receives the number of attention evaluations for each evaluation item stored in the operation number table 20B in the attention evaluation management table 20 through the attention evaluation detection unit 19, for example, the preference attention evaluation number P, the topical attention evaluation number. T and the novelty attention evaluation count N are acquired (step S42).

  When the priority setting unit 23 obtains the attention evaluation count for each evaluation item, the priority setting section 23 subtracts the total attention evaluation count S for all the evaluation items from the cumulative operation count M, and the difference result is determined to be no attention evaluation item. The number of times is calculated (step S43). The total attention evaluation count S is calculated by P + T + N.

  The priority setting unit 23 calculates the priority α corresponding to the evaluation item of the preference with the mathematical formula of (1 + P + (MS−3) / 3) / (3 + M) (step S44).

  Further, the priority setting unit 23 calculates the priority β corresponding to the topical evaluation item using the equation (1 + T + (MS) / 3) / (3 + M) (step S45).

  Furthermore, the priority setting unit 23 calculates the priority γ corresponding to the novelty evaluation item using the formula (1 + N + (MS) / 3) / (3 + M) (step S46).

  The priority setting unit 23 calculates the priority α corresponding to the preference evaluation item calculated in step S44, the priority β corresponding to the topic evaluation item calculated in step S45, and calculated in step S46. The priority γ corresponding to the novel evaluation item is updated in the priority management table 15 through the priority management unit 16 (step S47), and the processing operation shown in FIG.

  In the priority update process illustrated in FIG. 11, the preference evaluation item, the topical evaluation item, and the novelty are based on the cumulative operation count M on the user side for the recommended result and the attention evaluation counts P, T, and N for each evaluation item. Priority α, β, and γ corresponding to the evaluation items are calculated, and the priority management unit 16 manages the calculated priorities α, β, and γ corresponding to the preference, topicality, and novelty evaluation items. Since the priority management table 15 is updated, the priority corresponding to the evaluation item that is noticed by the user can be dynamically changed from the user operation tendency.

  In the priority update process, when calculating the priority corresponding to the evaluation score, not only the attention evaluation counts P, T, and N but also the number of times (MS) determined that there is no attention evaluation item is taken into consideration. The change in the priority with respect to the change in the number of attention evaluations P, T, and N does not become steep, and the priority corresponding to each evaluation item can be gradually reflected according to the user operation tendency.

  FIG. 12 is an explanatory diagram briefly showing the relationship between the detection of the attention evaluation item and the change in priority according to the first embodiment.

  12, when the attention evaluation detecting unit 19 detects a novel attention evaluation item (step S51), the priority setting unit 23 increases the novelty priority γ, and preference and topicality priority α. β decreases, and these priorities are updated in the priority management table 15.

  Next, the priority setting unit 23 detects the novel attention evaluation item in the attention evaluation detection unit 19 (step S52), the novelty priority γ increases, the preference and topicality priority α, γ decreases, and these priorities are updated in the priority management table 15.

  Next, the priority setting unit 23 detects a preference attention evaluation item in the attention evaluation detection unit 19 (step S53), the preference priority α is increased, the novelty and topicality priority β, γ decreases, and these priorities are updated in the priority management table 15.

  Next, the priority setting unit 23 detects a preference attention evaluation item by the attention evaluation detection unit 19 (step S54), the preference priority α increases, the novelty and topicality priority β, γ decreases, and these priorities are updated in the priority management table 15.

  Next, when the attention evaluation detection unit 19 detects that there is no attention evaluation item (step S55), the priority setting unit 23 does not increase or decrease the priorities α, β, and γ of preference, novelty, and topicality. Keep it as it is.

  Furthermore, when the attention evaluation detection unit 19 detects a topical attention evaluation item (step S56), the priority setting unit 23 increases the topical priority β, and preference and novelty priorities α and γ. Decreases, and these priorities are updated in the priority management table 15.

  That is, the priority setting unit 23 dynamically changes the priority according to the attention evaluation item (number of attention evaluations) for each evaluation item of the attention evaluation detection unit 19.

  As a result, the information recommendation unit 14 calculates a total evaluation score (preference evaluation score, topicality evaluation score, and novelty evaluation score) for the information to be searched, and for each evaluation item stored in the priority management unit 16. Priority α, β, γ is obtained, and the recommendation degree of each information to be searched is calculated by a formula of preference evaluation score × priority α + topic evaluation score × priority β + newness evaluation score × priority γ. The recommended result of each information is displayed on the recommended result display unit 17 on the screen.

  In the first embodiment, when a user operation is performed on the information of the recommended result 40 being displayed on the recommended result display unit 17, each evaluation score is acquired from the recommended result 40 of the information, and the maximum of the acquired evaluation scores is obtained. The evaluation item of the value evaluation score is determined as the attention evaluation item, and the number of attention evaluations of this evaluation item is incremented by +1, and the priority for each evaluation item is calculated based on the cumulative number of operations and the number of attention evaluations for each evaluation item. Therefore, when searching for information, the priority for each evaluation item is appropriately set for each user according to the user's operation tendency, and information search reflecting the user's intention can be performed.

  In Example 1, for each piece of information to be searched, an evaluation score for each evaluation item is calculated, a priority corresponding to the evaluation score for each evaluation item is acquired, and based on the evaluation score and priority for each evaluation item. , The recommendation degree related to each piece of information to be searched is calculated by the formula of preference evaluation score × priority α + topic evaluation score × priority β + newness evaluation score × priority γ. In addition, it is possible to calculate the degree of recommendation of information to be searched corresponding to topicality and novelty.

  In the first embodiment, since the recommended result 40 including the recommendation level related to all the information to be searched is displayed on the recommended result display unit 17, the user can perform a search while the screen is displayed on the recommended result display unit 17. It is possible to recognize the recommendation level of each piece of information based on the recommendation result 40 related to all the target information.

  Further, in the first embodiment, when a user operation for the information of the recommended result 40 displayed on the recommended result display unit 17 is detected, each evaluation score is acquired from the recommended result 40 of the information, and among the acquired evaluation scores, Since the evaluation item of the maximum evaluation score is determined as the attention evaluation item, and the number of attention evaluations of this evaluation item is incremented by +1, the user attention evaluation item can be automatically aggregated from the user operation tendency.

  Further, in the first embodiment, the preference evaluation item, the topic evaluation item, and the novelty evaluation item are based on the cumulative operation count M on the user side with respect to the recommended result and the attention evaluation counts P, T, and N for each evaluation item. Priority α, β, and γ corresponding to the priority, and the priority management unit 16 manages the priorities α, β, and γ corresponding to the calculated evaluation items for preference, topicality, and novelty, respectively. Since the management table 15 is updated, the priority corresponding to the evaluation item that is noticed by the user can be dynamically changed from the user operation tendency without requiring the priority setting operation by the user.

  Further, in Example 1, when calculating the priority corresponding to the evaluation score, not only the attention evaluation counts P, T, and N but also the number of times (MS) determined that there is no attention evaluation item is considered. The change in the priority with respect to the change in the number of attention evaluations P, T, and N does not become steep, and the priority corresponding to each evaluation item can be gradually reflected according to the user operation tendency.

  In the first embodiment, when a predetermined period has elapsed from the storage date and time of operation history information for the operation history information table 20A, the operation history information after the predetermined period has been deleted from the operation history information table 20A and the operation history information Since the total number of operations M stored in the number-of-operations table 20B and the number of attention evaluations P, T, and N for each evaluation item are recalculated in accordance with the deletion, the past user operation tendency after a predetermined period Can be reflected in the latest user operation tendency.

  In the first embodiment, the Bayesian estimation of the Bernoulli trial is applied, and the priority of the evaluation item i is calculated by the equation (1 + E (i) + (MS) / X) / (X + M). An appropriate priority can be calculated based on statistical information.

  In the attention evaluation detection process of the first embodiment, when a user operation is detected with respect to the information of the recommended result 40, all evaluation scores related to the information of the recommended result are acquired, and the maximum value among these evaluation scores is evaluated. The score is extracted, and the evaluation item of the extracted maximum evaluation score is determined as the attention evaluation item. If the difference from all other evaluation scores exceeds a predetermined threshold for each evaluation score, the predetermined value is determined. The evaluation item of the larger evaluation score that exceeds the threshold value may be determined as the attention evaluation item, and an example in this case will be described below as Example 2.

  FIG. 13 is a block diagram illustrating a schematic configuration inside the information search apparatus 1A according to the second embodiment. Note that the same components as those of the information search apparatus 1 of the first embodiment are denoted by the same reference numerals, and the description of the overlapping configuration and operation is omitted.

  The difference between the information search device 1A shown in FIG. 13 and the information search device 1 of the first embodiment is that when a user operation on information related to the recommended result 40 displayed on the screen is detected on the recommended result display unit 17, the user operation is detected. When the evaluation score for each evaluation item of the recommended result 40 related to information is acquired and the difference from all other evaluation scores exceeds a predetermined threshold for each evaluation score, the larger evaluation score of the larger evaluation score exceeding the predetermined threshold There is an attention evaluation detection unit 19A that determines an evaluation item as an attention evaluation item.

  FIG. 14 is an explanatory diagram briefly showing the operation concept of the attention evaluation detecting unit 19A of the second embodiment.

  As shown in FIG. 14A, the attention evaluation detection unit 19 </ b> A determines the novelty evaluation score when the novelty evaluation score is larger than the predetermined threshold D compared to both the preference evaluation score and the topicality evaluation score. The novelty attention evaluation count N corresponding to the evaluation item is incremented by +1.

  Further, the attention evaluation detection unit 19A has no difference exceeding the predetermined threshold D among the three preference evaluation scores, the topic evaluation score, and the novelty evaluation score (see FIG. 14B), or three Even if there is a difference exceeding the predetermined threshold D among the evaluation scores, if one evaluation score is extremely small (see FIG. 14C), a specific evaluation item is not incremented as the number of attention evaluations.

  Next, the operation of the information search apparatus 1A according to the second embodiment will be described. FIG. 15 is a flowchart illustrating the processing operation of the attention evaluation detection unit 19A related to the attention evaluation detection process of the second embodiment.

  In the attention evaluation detection process illustrated in FIG. 15, when a user operation is detected with respect to information of the recommended result 40 displayed on the recommended result display unit 17, an evaluation score for each evaluation item is acquired from the recommended result 40 of the information. This is a process for detecting an attention evaluation item corresponding to a user operation tendency based on the comparison result of the obtained evaluation scores.

  When the attention evaluation detection unit 19A illustrated in FIG. 15 detects a user operation on the information of the recommendation result 40 displayed on the screen of the recommendation result display unit 17, the attention evaluation detection unit 19A acquires each evaluation score for each evaluation item from the recommendation result 40 of the information. (Step S61).

  When the attention evaluation detection unit 19A obtains each evaluation score for each evaluation item from the recommended result 40, the evaluation score for comparison is designated from among the evaluation scores (step S62), and the evaluation score for comparison and all other evaluations are specified. It is determined whether or not the difference from the score exceeds a predetermined threshold D (step S63). For example, when the evaluation score to be compared is “preference evaluation score”, all other evaluation scores are the topicality evaluation score and the novelty evaluation score.

  When the difference between the comparison target evaluation score and all the other evaluation scores exceeds the predetermined threshold D (Yes in step S63), the attention evaluation detection unit 19A extracts a larger evaluation score that exceeds the predetermined threshold D ( Step S64). Note that the attention evaluation detection unit 19A, for example, if the difference between the preference evaluation score and the topicality evaluation score and the novelty evaluation score exceeds a predetermined threshold D, the attention evaluation detection unit 19A generates a larger preference evaluation score exceeding the predetermined threshold D. To extract.

  When the attention evaluation detection unit 19A extracts a larger evaluation score that exceeds the predetermined threshold D, the attention evaluation detection unit 19A determines that the evaluation item of the evaluation score is the attention evaluation item (step S65), and calculates the number of attention evaluations corresponding to the attention evaluation item. It is incremented by +1 (step S66), and it is determined whether there is an unspecified comparison target evaluation score (step S67).

  Note that, for example, when the evaluation item of the preference evaluation score is determined as the attention evaluation item, the attention evaluation detection unit 19A increments the preference attention evaluation count P by +1 and updates the operation count table 20B.

  When there is an unspecified comparison target evaluation score (Yes in step S67), the attention evaluation detection unit 19A proceeds to step S62 to specify an unspecified comparison target evaluation score.

  When there is no unspecified comparison target evaluation score (No in step S67), the attention evaluation detection unit 19A determines that all evaluation scores have been compared, and ends the processing operation illustrated in FIG.

  In addition, when the difference between the evaluation score of the comparison target and all other evaluation scores does not exceed the predetermined threshold value D (No in step S63), the attention evaluation detection unit 19A determines the evaluation score of the unspecified comparison target In order to determine whether or not there is, the process proceeds to step S67.

  In the attention evaluation detection process shown in FIG. 15, when a user operation is detected with respect to information of the recommended result 40 displayed on the recommended result display unit 17, an evaluation score is acquired from the recommended result 40 of the information, and these acquired Among the evaluation scores, the evaluation score of the comparison target is specified, the specified evaluation score of the comparison target is compared with all the other evaluation scores, and the evaluation item having a large evaluation score whose difference exceeds the predetermined threshold D is evaluated with attention Since the item is determined as an item and the number of attention evaluations of this evaluation item is incremented by +1, the evaluation item of user attention can be automatically added from the normal user operation tendency.

  As a result, the priority setting unit 23 uses the priority corresponding to the evaluation score for each evaluation item based on the attention evaluation counts P, T, N for each evaluation item and the cumulative operation count M detected by the attention evaluation detection unit 19A. Since α, β, and γ are sequentially calculated, the priority corresponding to the evaluation item that is noticed by the user can be dynamically changed from the user operation tendency.

  In Example 2, when a user operation is detected with respect to the information of the recommended result 40 being displayed on the recommended result display unit 17, an evaluation score is acquired from the recommended result 40 of the information, and among these acquired evaluation scores, An evaluation score to be compared is designated, the designated evaluation score to be compared is compared with all other evaluation scores, and an evaluation item having a large evaluation score whose difference exceeds a predetermined threshold D is determined as an attention evaluation item.

  In other words, in the case where the difference between the evaluation scores is large, the information is relatively low in the recommendation level. When the information is arranged in descending order by the recommendation level, the information is displayed behind the list of the recommended results 40. It will be. As a result, it is considered that the user has intentionally manipulated information behind the scene by determining that he wants to process. Therefore, the evaluation score is considered to be an attention evaluation item because it is highly likely that the evaluation score matches the intention of the user (which evaluation item was focused on).

  On the other hand, in the case where the difference between the evaluation scores is small, the information is relatively high in the recommendation level, and when the information is arranged in descending order according to the recommendation level, the information is forward of the list of the recommended results 40. Will be displayed. As a result, since the user can easily select and operate these pieces of information, it is considered difficult to capture the user's intention, and therefore it is determined that there is no attention evaluation item.

  In the priority setting unit 23 in the information search apparatus 1 of the first embodiment, the priority of the evaluation item i is calculated using the formula (1 + E (i) + (MS) / X) / (X + M). However, another formula may be used, and an example in this case will be described as Example 3.

  FIG. 16 is a block diagram illustrating a schematic configuration inside the information search apparatus 1B according to the third embodiment. Note that the same components as those of the information search apparatus 1 of the first embodiment are denoted by the same reference numerals, and the description of the overlapping configuration and operation is omitted.

  The difference between the information search device 1B shown in FIG. 16 and the information search device 1 of the first embodiment is in the mathematical formula for calculating the priority used in the priority setting unit 23B.

  The priority setting unit 23B calculates the priority of the evaluation item i using the formula of (1 + E (i)) / (X ′ + M) + (1+ (MS)) / (X ′ + M) / X. To do.

  It should be noted that the difference between the formula used in the priority setting unit 23B and the formula used in the priority setting unit 23 is considered as one of the evaluation items when it is determined that there is no attention evaluation item, and the number of evaluation items X ′ = 1 + X, after obtaining a probability (1+ (M−S)) / (X ′ + M) of not paying attention to any evaluation item, the probability is equally divided by the number of evaluation items X and added to the probability of each evaluation item Is.

  The priority setting unit 23B sets, for example, a preference priority α as a formula of (1 + P) / (4 + M) + (1+ (MS)) / (4 + M) / 3, and a topicality priority β as (1 + T). ) / (4 + M) + (1+ (MS)) / (4 + M) / 3, the novelty priority γ is (1 + N) / (4 + M) + (1+ (MS)) / (4 + M) / 3 respectively.

  FIG. 17 is an explanatory diagram briefly showing the relationship between the cumulative operation count M and the attention evaluation count and priority in the third embodiment.

  The priority setting unit 23B sets the cumulative operation count M to “10”, the preference attention determination count P to “0”, the topical attention determination count T to “0”, and the novelty attention determination count N to “8”. In this case, the preference priority α is (1 + 0) / (4 + 10) + (1+ (10−8)) / (4 + 10) /3≈0.14, and the topicality priority β is (1 + 0) / ( 4 + 10) + (1+ (10−8)) / (4 + 10) /3≈0.14, the novelty priority γ is (1 + 8) / (4 + 10) + (1+ (10−8)) / (4 + 10) We will pay attention to novelty by setting /3≈0.71.

  Further, when the cumulative operation count M is “10”, the preference attention determination count P is “4”, the topical attention determination count T is “1”, and the novelty attention determination count N is “3”, The priority α is (1 + 4) / (4 + 10) + (1+ (10−8)) / (4 + 10) /3≈0.43, and the topical priority β is (1 + 1) / (4 + 10) + (1+ (10−8)) / (4 + 10) /3≈0.21, and the novelty priority γ is (1 + 3) / (4 + 10) + (1+ (10−8)) / (4 + 10) / 3≈0. 36 and pay attention to palatability and novelty.

  Further, when the cumulative operation count M is “10”, the preference attention determination count P is “3”, the topical attention determination count T is “3”, and the novelty attention determination count N is “4”, the preference The priority α is (1 + 3) / (4 + 10) + (1+ (10−10)) / (4 + 10) /3≈0.31, and the topical priority β is (1 + 3) / (4 + 10) + (1+ (10−10)) / (4 + 10) /3≈0.31, and the novelty priority γ is (1 + 4) / (4 + 10) + (1+ (10−10)) / (4 + 10) / 3≈0. 38, and pay close attention.

  When the priority setting unit 23B calculates the preference priority α, the topical priority β, and the novelty priority γ, the priority management unit 23B calculates the priorities α, β, and γ for each evaluation item. 16 in the priority management table 15.

  In Example 3, the case where it is determined that there is no attention evaluation item is considered as one of the evaluation items, and (1 + E (i)) / (X ′ + M) + (1+ (MS)) / (X ′ + M) / Since the priority E (i) for each evaluation item i is calculated using the formula of X, an appropriate priority can be calculated based on statistical information.

  In the first to third embodiments, when the operation history information stored in the attention evaluation management table 20 has passed a predetermined period from the storage date and time, the operation history information after the predetermined period has been deleted from the operation history information table 20A. However, when the number of operation histories managed in the attention evaluation management table 20 reaches a predetermined number, not the predetermined period, the oldest operation history is deleted from the stored operation histories, and the operation It goes without saying that the cumulative operation count M stored in the operation count table 20B and the attention evaluation counts P, T, and N for each evaluation item may be recalculated in accordance with the deletion of the history information 41.

  In the first to third embodiments, the degree of importance of the search condition stored in the search condition database 12 of the preference management unit 13 is also extracted, and the preference evaluation unit 31 uses the vector space method to change the preference. Similarly, it may be expressed by a vector, the keyword may be a vector base, the importance of each keyword may be the element value of the vector, and the importance of the keyword may be determined using statistics.

  Moreover, in the said Examples 1-3, although the evaluation item of preference, the evaluation item of topicality, and the evaluation item of novelty were mentioned as an example, even if it is another evaluation item, the same effect is obtained. It goes without saying that it is obtained.

  Moreover, in the said Examples 1-3, although the preference evaluation item and the topic evaluation item were calculated using the vector space method, the preference evaluation score and the topic evaluation score were calculated. It goes without saying that the same effect can be obtained even if the preference evaluation score and the topic evaluation score are calculated by the method.

  In the first to third embodiments, when a user operation on the recommended result information displayed on the screen is detected on the recommended result display unit 17, the recommendation level and the recommended degree are included in the recommended result information from the recommended result display unit 17. Although each evaluation score is notified to the attention evaluation detecting unit 19 (19A), it is needless to say that only the evaluation score may be notified.

  Moreover, in the said Examples 1-3, although the recommendation result containing an evaluation score was displayed on the screen on the recommendation result display part 17, you may make it display only a recommendation level and information on a screen. Needless to say, when a user operation on the information is detected, the evaluation score may be notified to the attention evaluation detection unit 19 (19A) in the recommended result information.

  In the first to third embodiments, when a user operation on the recommended result information displayed on the screen is detected on the recommended result display unit 17, the recommendation level and the recommended degree are included in the recommended result information from the recommended result display unit 17. Each evaluation score is notified to the attention evaluation detection unit 19 (19A). However, not the recommendation result display unit 17 but the information recommendation unit 14 detects each evaluation score from the recommendation result information. Needless to say, the unit 19 (19A) may be notified.

  In the first to third embodiments, the information search device for searching for information such as news and programs has been described as an example. For example, a product search device for searching for a user-intended product from a plurality of products. Needless to say, the same effect can be obtained.

  As mentioned above, although the Example of this invention was described, the range of the technical idea of this invention is not limited by this Example, Unless it deviates from the range of the technical idea described in the claim, it is various. It goes without saying that various embodiments can be implemented. Moreover, the effect described in the present Example is not limited to this.

  Of course, all or some of the processes described as being automatically performed among the various processes described in this embodiment can be performed manually, and vice versa. It goes without saying that all or part of the processing described as being possible can be performed automatically. Further, it goes without saying that the processing procedure, control procedure, specific name, information including various data and parameters described in the present embodiment can be appropriately changed unless otherwise specified.

  In addition, each component of each device illustrated is functionally described, and is not necessarily physically configured as illustrated, and a specific aspect of each device is illustrated. Needless to say, it cannot be limited.

  Further, various processing functions performed in each device are performed on a CPU (Central Processing Unit) (or a micro computer such as an MPU (Micro Processing Unit) or MCU (Micro Controller Unit)) or on the same CPU (or MPU, MCU, etc.). It goes without saying that all or any part of the program may be executed on a program that is analyzed and executed by a microcomputer) or on hardware based on wired logic.

  As described above, the following supplementary notes are further disclosed regarding the embodiment including the present example.

(Appendix 1) An information acquisition unit that acquires metadata related to information to be searched;
A preference management unit for storing a search condition for searching the information to be searched;
A preference evaluation unit that calculates a preference evaluation score as a preference evaluation item based on the similarity between the metadata and the search condition stored in the preference management unit for each search target information;
For each information to be searched, another kind of evaluation unit that calculates an evaluation score as an evaluation item other than the preference evaluation item,
For each evaluation item of each evaluation score, a priority management unit that stores a priority indicating a priority ratio of the evaluation score;
The evaluation score for each evaluation item related to the information to be searched is weighted according to the priority for each evaluation item stored in the priority management unit, and the information related to the information to be searched is based on the weighted evaluation score. An evaluation output unit for outputting an evaluation result; and
An operation detection unit for detecting a user operation for the information of the evaluation result;
Each evaluation score is extracted from the information on the evaluation result related to the user operation detected by the operation detection unit, the extracted evaluation scores are compared, and the attention corresponding to the evaluation item of user attention is based on the comparison result. Attention evaluation detection unit for detecting evaluation items;
A priority setting unit that calculates the priority for each evaluation item based on the attention evaluation item detected by the attention evaluation detection unit, and updates the calculated priority for each evaluation item to the priority management unit; ,
An information retrieval apparatus comprising:

(Supplementary Note 2) The attention evaluation detecting unit
The information search apparatus according to appendix 1, wherein an evaluation item having the highest evaluation score among the plurality of evaluation scores is detected as the attention evaluation item based on a comparison result of the evaluation scores.

(Supplementary Note 3) The attention evaluation detecting unit
When one of the plurality of evaluation scores is designated based on the comparison result of the evaluation scores, and the difference between the designated evaluation score and all the evaluation scores other than the designated evaluation score exceeds a predetermined threshold The information search device according to appendix 1, wherein an evaluation item having a large evaluation score exceeding the predetermined threshold is detected as the attention evaluation item.

(Supplementary Note 4) The attention evaluation detecting unit
When a user operation on the information of the evaluation result is detected by the operation detection unit, the operation history is managed, and the total number of operations is counted. When the attention evaluation item is detected, an evaluation corresponding to the attention evaluation item is performed. Having an attention evaluation management table that counts the number of attention evaluations for each item, and manages the total number of operations and the number of attention evaluations for each evaluation item;
The priority setting unit
The priority for each evaluation item is calculated based on the cumulative number of operations managed in the attention evaluation management table and the number of attention evaluations for each evaluation item, and the priority for each evaluation item calculated is the priority management The information search device according to any one of appendices 1 to 3, wherein the information search device is updated to a unit.

(Supplementary Note 5) The priority setting unit
When the number of evaluation items is X, the cumulative number of operations is M, the evaluation item is i, the number of attention evaluations of the evaluation item i is E (i), and the total number of attention evaluations of all evaluation items is S, The information search device according to appendix 4, wherein the priority for each of the evaluation items is calculated by a mathematical formula of 1 + E (i) + (MS) / X) / (X + M).

(Supplementary Note 6) The priority setting unit
When the number of evaluation items is X, the cumulative number of operations is M, the evaluation item is i, the number of attention evaluations of the evaluation item i is E (i), and the total number of attention evaluations of all evaluation items is S, The information search device according to appendix 4, wherein a priority for each evaluation item is calculated by a mathematical formula of 1 + E (i)) / (1 + X + M) + (1+ (MS)) / (1 + X + M) / X .

(Supplementary Note 7) The attention evaluation detecting unit
When the operation history deletion condition in the attention evaluation management table is satisfied, the operation history satisfying the deletion condition is deleted, and the cumulative number being managed in the attention evaluation management table according to the deletion of the operation history The information search device according to any one of appendices 4 to 6, wherein the number of operations and the number of attention evaluations are recalculated.

(Supplementary Note 8) The attention evaluation detecting unit
When the operation history has passed a predetermined period from the start of management, the operation history that has passed the predetermined period is deleted, and the cumulative operation being managed in the attention evaluation management table according to the deletion of the operation history The information search device according to appendix 7, wherein the number of times and the number of attention evaluations are recalculated.

(Supplementary Note 9) The attention evaluation detecting unit
When the number of operations managed in the operation history exceeds a predetermined number, the oldest operation history being managed is deleted, and the cumulative number of operations managed in the attention evaluation management table according to the deletion of the operation history The information search device according to appendix 7, wherein the number of attention evaluations is recalculated.

(Additional remark 10) The said other kind evaluation part is
For each search target information, a topicality evaluation unit that calculates a topicality evaluation score as a topicality evaluation item based on the similarity between the metadata related to the search target information and metadata related to other information,
For each information to be searched, a novelty evaluation unit that calculates a novelty evaluation score using a difference between a provision start date and time of the information to be searched and a reference date as a novelty evaluation item,
The information search device according to any one of appendices 1 to 9, characterized by comprising:

(Supplementary Note 11) The operation detection unit includes:
11. The information search apparatus according to any one of appendices 1 to 10, wherein a user operation of reference, storage reservation, storage, copy, or movement with respect to the information of the evaluation result is detected.

(Supplementary Note 12) An information acquisition step of acquiring metadata related to information to be searched;
A preference evaluation step for calculating a preference evaluation score as a preference evaluation item based on the similarity between the metadata and the search condition stored in the preference management unit for each search target information;
For each information to be searched, another type evaluation step for calculating an evaluation score as an evaluation item other than the preference evaluation item,
The evaluation score for each evaluation item related to the information to be searched is weighted according to the priority for each evaluation item stored in the priority management unit that stores the priority indicating the priority ratio of the evaluation score. An evaluation output step for outputting an evaluation result related to the information to be searched based on the evaluated score;
An operation detecting step for detecting a user operation for the information of the evaluation result;
Each evaluation score is extracted from the information of the evaluation result related to the user operation detected in the operation detection step, the extracted evaluation scores are compared, and the attention corresponding to the evaluation item of user attention based on the comparison result Attention evaluation detection step for detecting evaluation items;
A priority setting step of calculating the priority for each evaluation item based on the attention evaluation item detected in the attention evaluation detecting step, and updating the priority for the calculated evaluation item in the priority management unit; ,
An information search method characterized by including:

(Supplementary note 13)
An information acquisition procedure for acquiring metadata related to information to be searched;
A preference evaluation procedure for calculating a preference evaluation score as a preference evaluation item based on the similarity between the metadata and the search condition stored in the preference management unit for each search target information;
For each information to be searched, another kind of evaluation procedure for calculating an evaluation score as an evaluation item other than the preference evaluation item,
The evaluation score for each evaluation item related to the information to be searched is weighted according to the priority for each evaluation item stored in the priority management unit that stores the priority indicating the priority ratio of the evaluation score. An evaluation output procedure for outputting an evaluation result related to the information to be searched based on the evaluated score;
An operation detection procedure for detecting a user operation on the information of the evaluation result;
Each evaluation score is extracted from the information of the evaluation result related to the user operation detected in the operation detection procedure, the extracted evaluation scores are compared, and the attention corresponding to the evaluation item of user attention is based on the comparison result Attention evaluation detection procedure for detecting evaluation items;
A priority setting procedure for calculating the priority for each evaluation item based on the attention evaluation item detected in the attention evaluation detection procedure, and updating the priority for the calculated evaluation item in the priority management unit; ,
An information retrieval program for executing a process including:

1 is a block diagram illustrating a schematic configuration inside an information search apparatus according to a first embodiment. It is explanatory drawing which shows notionally the calculation method of a preference evaluation score. It is explanatory drawing which shows notionally the calculation method of a novelty evaluation score. It is explanatory drawing which shows the recommendation result during a screen display on the recommendation result display part. It is explanatory drawing which shows briefly the table content of the operation history information table in an attention evaluation management table. It is explanatory drawing which shows the table content of the operation frequency table in an attention evaluation management table simply. It is explanatory drawing which shows the operation | movement concept of the attention evaluation detection part of Example 1 simply. It is explanatory drawing which shows the relationship between the total operation frequency of Example 1, the attention evaluation frequency, and a priority. 6 is a flowchart illustrating a processing operation of an information recommendation unit related to a recommendation level providing process according to the first embodiment. 6 is a flowchart illustrating a processing operation of an attention evaluation detection unit related to an attention evaluation detection process according to the first embodiment. 6 is a flowchart illustrating a processing operation of a priority setting unit related to a priority update process according to the first embodiment. It is explanatory drawing which shows briefly the relationship of the detection of the attention evaluation item of Example 1, and the change of a priority. It is a block diagram which shows schematic structure inside the information search device of Example 2. It is explanatory drawing which shows the operation | movement concept of the attention evaluation detection part of Example 2 simply. 10 is a flowchart illustrating a processing operation of an attention evaluation detection unit related to attention evaluation detection processing according to the second embodiment. FIG. 10 is a block diagram illustrating a schematic configuration inside an information search apparatus according to a third embodiment. It is explanatory drawing which shows directly the relationship between the frequency | count of total operation of Example 3, the frequency | count of attention evaluation, and a priority.

Explanation of symbols

DESCRIPTION OF SYMBOLS 1 Information retrieval apparatus 1A Information retrieval apparatus 1B Information retrieval apparatus 11 Information acquisition part 12 Search condition database 13 Preference management part 14 Information recommendation part 15 Priority management table 16 Priority management part 17 Recommendation result display part 18 Operation detection part 19 Attention evaluation Detection unit 19A Attention evaluation detection unit 20 Attention evaluation management table 20A Operation history information table 20B Operation frequency table 23 Priority setting unit 23B Priority setting unit 31 Preference evaluation unit 32 Topicity evaluation unit 33 Novelty evaluation unit 33 Recommendation calculation Part

Claims (7)

An information acquisition unit for acquiring metadata related to information to be searched;
A preference management unit for storing a search condition for searching the information to be searched;
A preference evaluation unit that calculates a preference evaluation score as a preference evaluation item based on the similarity between the metadata and the search condition stored in the preference management unit for each search target information;
For each information to be searched, another kind of evaluation unit that calculates an evaluation score as an evaluation item other than the preference evaluation item,
For each evaluation item of each evaluation score, a priority management unit that stores a priority indicating a priority ratio of the evaluation score;
The evaluation score for each evaluation item related to the information to be searched is weighted according to the priority for each evaluation item stored in the priority management unit, and the information related to the information to be searched is based on the weighted evaluation score. An evaluation output unit for outputting an evaluation result; and
An operation detection unit for detecting a user operation for the information of the evaluation result;
Each evaluation score is extracted from the information on the evaluation result related to the user operation detected by the operation detection unit, the extracted evaluation scores are compared, and the attention corresponding to the evaluation item of user attention is based on the comparison result. Attention evaluation detection unit for detecting evaluation items;
A priority setting unit that calculates the priority for each evaluation item based on the attention evaluation item detected by the attention evaluation detection unit, and updates the calculated priority for each evaluation item to the priority management unit; ,
An information retrieval apparatus comprising: The attention evaluation detecting unit
The information search device according to claim 1, wherein an evaluation item having the highest evaluation score among the plurality of evaluation scores is detected as the attention evaluation item based on the comparison result of the evaluation scores. The attention evaluation detecting unit
When one of the plurality of evaluation scores is designated based on the comparison result of the evaluation scores, and the difference between the designated evaluation score and all the evaluation scores other than the designated evaluation score exceeds a predetermined threshold The information retrieval apparatus according to claim 1, wherein an evaluation item having a large evaluation score exceeding the predetermined threshold is detected as the attention evaluation item. The attention evaluation detecting unit
When a user operation on the information of the evaluation result is detected by the operation detection unit, the operation history is managed, and the total number of operations is counted. When the attention evaluation item is detected, an evaluation corresponding to the attention evaluation item is performed. Having an attention evaluation management table that counts the number of attention evaluations for each item, and manages the total number of operations and the number of attention evaluations for each evaluation item;
The priority setting unit
The priority for each evaluation item is calculated based on the cumulative number of operations managed in the attention evaluation management table and the number of attention evaluations for each evaluation item, and the priority for each evaluation item calculated is the priority management The information search device according to claim 1, wherein the information search device is updated to a unit. The other-type evaluation unit is
For each search target information, a topicality evaluation unit that calculates a topicality evaluation score as a topicality evaluation item based on the similarity between the metadata related to the search target information and metadata related to other information,
For each information to be searched, a novelty evaluation unit that calculates a novelty evaluation score using a difference between a provision start date and time of the information to be searched and a reference date as a novelty evaluation item,
The information search device according to claim 1, wherein the information search device includes: An information acquisition step for acquiring metadata relating to information to be searched;
A preference evaluation step for calculating a preference evaluation score as a preference evaluation item based on the similarity between the metadata and the search condition stored in the preference management unit for each search target information;
For each information to be searched, another type evaluation step for calculating an evaluation score as an evaluation item other than the preference evaluation item,
The evaluation score for each evaluation item related to the information to be searched is weighted according to the priority for each evaluation item stored in the priority management unit that stores the priority indicating the priority ratio of the evaluation score. An evaluation output step for outputting an evaluation result related to the information to be searched based on the evaluation score, an operation detection step for detecting a user operation for the information on the evaluation result,
Each evaluation score is extracted from the information of the evaluation result related to the user operation detected in the operation detection step, the extracted evaluation scores are compared, and the attention corresponding to the evaluation item of user attention based on the comparison result Attention evaluation detection step for detecting evaluation items;
A priority setting step of calculating the priority for each evaluation item based on the attention evaluation item detected in the attention evaluation detecting step, and updating the priority for the calculated evaluation item in the priority management unit; ,
An information search method characterized by including: Computer equipment,
An information acquisition procedure for acquiring metadata related to information to be searched;
A preference evaluation procedure for calculating a preference evaluation score as a preference evaluation item based on the similarity between the metadata and the search condition stored in the preference management unit for each search target information;
For each information to be searched, another kind of evaluation procedure for calculating an evaluation score as an evaluation item other than the preference evaluation item,
The evaluation score for each evaluation item related to the information to be searched is weighted according to the priority for each evaluation item stored in the priority management unit that stores the priority indicating the priority ratio of the evaluation score. An evaluation output procedure for outputting an evaluation result related to the information to be searched based on the evaluated score;
An operation detection procedure for detecting a user operation on the information of the evaluation result;
Each evaluation score is extracted from the information of the evaluation result related to the user operation detected in the operation detection procedure, the extracted evaluation scores are compared, and the attention corresponding to the evaluation item of user attention is based on the comparison result Attention evaluation detection procedure for detecting evaluation items;
A priority setting procedure for calculating the priority for each evaluation item based on the attention evaluation item detected in the attention evaluation detection procedure, and updating the priority for the calculated evaluation item in the priority management unit; ,
An information retrieval program for executing a process including:

Images