JP4818170B2 - Information search apparatus, information search method, information search program, and computer-readable recording medium recording the information search program - Google Patents

Description

  The present invention relates to a technology for searching for content.

  2. Description of the Related Art Conventionally, in an information search system provided on the Internet, contents such as sentences and drawings are accumulated in a database, contents related to a search keyword given by a user are taken out from the database, and presented as search results to the user. Technology was used.

  When presenting the content that is the search target, usually, the text relevance between the search formula and each content of the search result is obtained based on the frequency of use of each search term included in the search formula, and the text related A method of displaying contents in descending order is used (see Non-Patent Document 1).

  However, even if the search term is frequently used, the content is not necessarily expected by the user. Conversely, even content that is used infrequently may be expected content. Therefore, the search results ordered using the text relevance level do not match the search results displayed in the order desired by the user, and the user needs a lot of trial and error and time until the desired content is reached. There was a problem of giving to.

Therefore, Patent Document 1 discloses a technique for displaying the content desired by the user at a higher level by recording the number of browsing requests made by the user and correcting the above-described sentence relevance using this number. As a result, content with a high browsing request is displayed at a higher level, and content with a low browsing request is displayed at a lower level.
Japanese Patent Application No. 11-29501 Kitakenji, “Information Retrieval Algorithm”, Kyoritsu Shuppan, 2002 Dean MTO 'Brien, “Predictive modeling of first-click behavior in web-search, proceeding of WWW2006”, 2006, p.1031-1032.

  However, when making a browsing request by selecting the displayed content, it is known that more content displayed at the upper level tends to be selected and less content at the lower level tends to be selected (non-patent document). 2). Therefore, even if the content to be searched is ordered using the technique disclosed in Patent Document 1 due to the user's habit at the time of the browsing request, a so-called positive feedback phenomenon occurs, so that the search result is There is a problem that the order of displayed contents tends to be fixed.

  This invention is made | formed in view of the above, and makes it a subject to display a content in the order reflecting a user preference.

  The present invention according to claim 1 is a content storage means for storing content and an identifier of the content in association with each other, a search term used for a content search request, and browsing based on a search result based on the search term The browsing history storage means for storing the identifier of the content that has been identified in association with the display order indicating the order in which the identifier is displayed as a search result, and when the search term included in the search request is received, A plurality of contents including a search word are read from the content storage unit, sentence matching degrees indicating the frequency of appearance of the search word in the plurality of contents are calculated, and identifiers corresponding to the plurality of contents are respectively determined as the contents Further reading from the storage means to obtain the number of identifiers stored in the browsing history storage means matching the identifier, respectively The average value of the display order corresponding to the identifier stored in the browsing history storage means that matches the identifier is calculated, and stored in the browsing history storage means having a display order higher than the average value. The ratio of the number of identifiers stored in the browsing history storage means having the display rank equivalent to the average value with respect to the number of identifiers is obtained as the number of browsing requests for correction, and the number of browsing requests is the correction number Retrieval means for respectively calculating the browsing importance divided by the number of browsing requests, and a score obtained by combining the sentence suitability and the browsing importance for a plurality of identifiers read from the content storage means by the search means And a search result output means for outputting the plurality of identifiers based on the score.

  In the present invention, when a search word included in a search request is received, a plurality of contents including the search word are read from the content storage unit, and the frequency at which the search word appears in the plurality of contents is indicated. The sentence matching degree is calculated, the identifier corresponding to each of the plurality of contents is further read out from the content storage means, and the number of identifiers stored in the browsing history storage means that matches this identifier is obtained as the number of browsing requests. The average value of the display order corresponding to the identifier stored in the browsing history storage means that matches this identifier is calculated, and the number of identifiers stored in the browsing history storage means having a display order higher than this average value is calculated. On the other hand, the ratio of the number of identifiers stored in the browsing history storage means having the display order equivalent to the average value is obtained to correct the correction. Search means for calculating the browsing importance by dividing the number of browsing requests by the number of browsing requests for correction, and sentence matching and browsing for a plurality of identifiers read from the content storage means by the searching means Since there is a fitness synthesis unit that calculates the score obtained by combining the importance and a search result output unit that outputs the plurality of identifiers based on the score, the search result is displayed in a fixed manner. Content identifiers can be output in an order that reflects the user preferences.

  According to a second aspect of the present invention, in the first aspect of the present invention, the search means determines the number of search words stored in the browsing history storage means that matches the search word included in the search request. The search term usage count is obtained, and the browsing importance is summarized as a value obtained by dividing the browsing request count by the correction browsing request count and further dividing by the search term usage count.

  In the present invention, the search means obtains the number of search terms stored in the browsing history storage means that matches the search term included in the search request and sets the number of search terms used. Since the value obtained by dividing the number of times by the number of browsing requests for correction is further divided by the number of times the search word is used, it is possible to normalize the bias in the size of the number of browsing requests due to the difference in the search terms. That is, regardless of what search terms are given, a value obtained by dividing the number of browsing requests by the number of browsing requests for correction can be stored in a certain area.

  The gist of the present invention described in claim 3 is that, in the invention described in claim 1 or 2, the display rank higher than the average value is first.

  In the present invention, since the display order higher than the average value is first, it is possible to more reliably prevent the search result from being fixedly displayed, and the content identifiers in an order more reflecting the user's preference. Can be output.

  The present invention as set forth in claim 4 is a first step of associating a content and an identifier of the content and storing them in the content storage means, a search term used for the search request for the content, and a search by the search term A second step of associating the identifier of the content browsed based on the result with the display order indicating the order in which the identifier is displayed as the search result in association with each other and storing it in the browsing history storage means; When a search term is received, a plurality of contents including the search term are read out from the content storage means, and sentence matching degrees indicating the frequency of the search term appearing in the plurality of contents are calculated. Identifiers corresponding to the respective contents are further read out from the content storage means, and the identifiers stored in the browsing history storage means matching the identifiers are read. The number of children is obtained as the number of browsing requests, the average value of the display rank corresponding to the identifier stored in the browsing history storage means matching the identifier is calculated, and the display rank is higher than the average value. The ratio of the number of identifiers stored in the browsing history storage means having a display order equivalent to the average value to the number of identifiers stored in the browsing history storage means is determined as the number of browsing requests for correction. , A third step of calculating the browsing importance obtained by dividing the number of browsing requests by the number of browsing requests for correction, and a plurality of identifiers read from the content storage means in the third step, A fourth step of calculating a score obtained by combining the sentence suitability and the browsing importance, and a fifth step of outputting the plurality of identifiers based on the score. And it is required to.

  According to a fifth aspect of the present invention, in the invention according to the fourth aspect, the third step is a process of searching for a search word stored in the browsing history storage means that matches the search word included in the search request. The number is used as the number of search word usage, and the browsing importance is a value obtained by dividing the number of browsing request times by the number of correction browsing request times and further dividing by the number of search word usage times. .

  The gist of the present invention described in claim 6 is that, in the invention described in any one of claims 4 or 5, the display rank higher than the average value is first.

The present invention according to claim 7 is a first process for associating a content and an identifier of the content and storing them in the content storage means, a search word used for a content search request, and a search based on the search word The second process of storing the identifier of the content browsed based on the result and the display order indicating the order in which the identifier is displayed as the search result in association with each other and storing it in the browsing history storage means, and included in the search request When a search term is received, a plurality of contents including the search term are read out from the content storage means, and sentence matching degrees indicating the frequency of the search term appearing in the plurality of contents are calculated. The number of identifiers stored in the browsing history storage means that are further read out from the content storage means and correspond to the identifiers, respectively. Each of the obtained browsing requests is calculated, and an average value of display ranks corresponding to the identifiers stored in the browsing history storage means matching the identifiers is calculated, and the browsing history storage having a display rank higher than the average value. The number of identifiers stored in the browsing history storage unit having a display order equivalent to the average value with respect to the number of identifiers stored in the unit is determined as the number of browsing requests for correction, and the browsing request A third process for calculating the browsing importance obtained by dividing the number of times by the number of correction browsing requests, and the sentence suitability for the plurality of identifiers read from the content storage unit in the third process. a fourth process for calculating each score was synthesized with the viewing importance, causes executes a plurality of identifiers, a fifth process, to a computer to be output based on the score The gist of the door.

  According to an eighth aspect of the present invention, in the invention according to the seventh aspect of the invention, the third process includes the processing of the search term stored in the browsing history storage means that matches the search term included in the search request. The number is used as the number of search word usage, and the browsing importance is a value obtained by dividing the number of browsing request times by the number of correction browsing request times and further dividing by the number of search word usage times. .

  The gist of the present invention according to claim 9 is that, in the invention according to any one of claims 7 and 8, the display rank higher than the average value is first.

The present invention according to claim 10 is a first process for associating a content and an identifier of the content and storing them in the content storage means, a search word used for a search request for the content, and a search based on the search word The second process of storing the identifier of the content browsed based on the result and the display order indicating the order in which the identifier is displayed as the search result in association with each other and storing it in the browsing history storage means, and included in the search request When a search term is received, a plurality of contents including the search term are read out from the content storage means, and sentence matching degrees indicating the frequency of the search term appearing in the plurality of contents are calculated. The identifier corresponding to each of the contents is further read out from the content storage means, and the identifier stored in the browsing history storage means that matches the identifier And calculating the average number of display ranks corresponding to the identifiers stored in the browsing history storage means that match the identifiers, and the browsing history having a display rank higher than the average value. The number of identifiers stored in the browsing history storage means having the display order equivalent to the average value with respect to the number of identifiers stored in the storage means is determined as the number of browsing requests for correction, and the browsing A third process for calculating a browsing importance obtained by dividing a request count by the correction browse request count, and the sentence suitability for a plurality of identifiers read from the content storage means in the third process. said viewing a fourth process for calculating severity and the combined scores, respectively, a plurality of identifiers, is performed and a fifth process of outputting on the basis of the score in a computer and It is the gist of.

  According to an eleventh aspect of the present invention, in the invention according to the tenth aspect, in the third aspect, the third processing includes a search word stored in the browsing history storage means that matches the search word included in the search request. The number is used as the number of search word usage, and the browsing importance is a value obtained by dividing the number of browsing request times by the number of correction browsing request times and further dividing by the number of search word usage times. .

  The gist of the present invention according to claim 12 is that, in the invention according to any one of claims 10 or 11, the display rank higher than the average value is first.

  According to the present invention, it is possible to display contents in an order that reflects user preferences.

[First Embodiment]
FIG. 1 is a configuration diagram showing a configuration of an information search system 1 according to the first embodiment. The information search system 1 according to the present embodiment includes a user terminal 2 and an information search device 3 and is connected to each other via a network 4.

  The user terminal 2 includes input means such as a monitor and a keyboard, communication means for the network 4, etc., accepts a search request from a user (hereinafter simply referred to as “user”), and includes a search included in the search request. The word is transmitted to the search word receiving unit 31 of the information search device 3 to be described later via the network 4. Further, the user terminal 2 receives the search result for the search request transmitted from the search result output unit 35 of the information search device 3 via the network 4 and presents the search result to the user by displaying it on the screen.

  Furthermore, the user terminal 2 receives a browsing request for the desired content selected by the user who has browsed the search result, and transmits the browsing request to the browsing request receiving unit 36 of the information search device 3 via the network 4. Further, it receives the content corresponding to the browsing request transmitted from the content output unit 38 of the information search device 3 via the network 4 and displays the content on the screen.

  The information search device 3 includes a search word reception unit 31, a content search unit 32, a browsing history search unit 33, a matching level synthesis unit 34, a search result output unit 35, a browsing request reception unit 36, and a browsing history update. This includes a unit 37, a content output unit 38, a content storage unit 301 connected to the content search unit 32, and a browsing history storage unit 302 connected to the browsing history search unit 33. Hereinafter, the functions of these units will be described.

  The search word receiving unit 31 is connected to the network 4 and transmits the search word transmitted from the user terminal 2 to the content search unit 32 and the browsing history search unit 33.

  The content search unit 32 receives the search term transmitted from the search term receiving unit 31, searches the content storage unit 301 using the search term, calculates the sentence matching degree based on the search result, A search result including the content identifier and the sentence matching degree is transmitted to the browsing history searching unit 33 and the matching degree synthesizing unit 34. Also, the content search unit 32 searches the content storage unit 301 for the location of the corresponding content in response to the browsing request transmitted from the user terminal 2 via the browsing request receiving unit 36, and stores it in the searched location. Get the content that was recorded.

  The browsing history search unit 33 receives the search word transmitted from the search word receiving unit 31 and the search result transmitted from the content search unit 32, and calculates the browsing importance by referring to the browsing history storage unit 302. Then, the browsing importance is transmitted to the fitness composition unit 34.

  The matching level composition unit 34 receives the search result transmitted from the content search unit 32 and the browsing importance level transmitted from the browsing history search unit 33, and finally uses the sentence matching level and the browsing importance level. The score is calculated and transmitted to the search result output unit 35.

  The search result output unit 35 receives the score transmitted from the fitness synthesis unit 34, and outputs the content identifier of the content rearranged in the score order via the network 4 to the user terminal 2.

  The browsing request receiving unit 36 receives the browsing request transmitted from the user terminal 2 via the network 4 and transmits the browsing request to the content search unit 32 and the browsing history update unit 37.

  The browsing history update unit 37 adds the browsing history of the user to the browsing history storage unit 302 and updates the browsing history.

  The content output unit 38 receives the content acquired by the content search unit 32 and outputs it to the user terminal 2 via the network 4.

  As shown in FIG. 2, the content storage unit 301 stores content text information indicating the content itself, a content identifier that can identify the content, and a content location where the content is stored in association with each other. Yes. The content sentence information is a sentence indicating the content of the content. For example, if the content is a page stored on a server on the web, it means a sentence written on the page. The content identifier makes it possible to identify a certain content from other content. For example, if the content is a page stored in a server on the web, it means a URL (Uniform Resource Locator) of the page. The content location is a location where content files are stored. For example, when content is stored in a local computer, path information as shown in FIG. In the case of this content, the URL is stored.

  The browsing history storage unit 302 stores browsing histories for past browsing requests. As shown in FIG. 3, a content identifier, a search term included in the search request, and a content identifier are displayed as search results. The display order indicating the display order is stored in association with each other. For example, the second line from the top of FIG. 3 shows that “www.goo.ne.jp/” having the display rank “4” in the search result obtained by the search request with the search word “goo” is displayed. It means that browsing is requested. Note that the browsing history storage unit 302 stores content identifiers and display orders for search terms.

  Note that the above-described functions of the content search unit 32 and the browsing history search unit 33 can be configured to be included in one search unit, and the fitness level combining unit 34 is further added to the search unit. However, it does not affect the effect at all.

  As the network 4, for example, a LAN (Local Area Network), the Internet, a public line network, a cable television network, or the like can be used. The Internet is a collection of networks linked to each other based on a predetermined protocol.

  Next, the flow of processing of the information search device 3 in the embodiment of the present invention will be described. This processing flow mainly includes a search request stage until the user makes a search request and obtains a search result, and a browse request stage until the user requests and browses desired content based on the search result. It consists of two stages. Hereinafter, each step will be described with reference to flowcharts.

  First, the flow of processing at the search request stage will be described with reference to FIG.

  First, the search word receiving unit 31 receives a search word included in the search request transmitted from the user terminal 2 via the network 4 and transmits the search word to the content search unit 32 and the browsing history search unit 33. (S101).

  Next, the content search unit 32 receives the search term transmitted from the search term receiving unit 31, reads a plurality of contents including the search term and the content identifier from the content storage unit 301, and the search term is assigned to each content. The sentence matching degree indicating the appearance frequency is calculated, and the search result including the content identifier and the sentence matching degree is transmitted to the browsing history searching unit 33 and the matching degree synthesizing unit 34 (S102).

  Here, the sentence suitability means the frequency with which a search word appears in the content, and for example, TF (Term Frequency) method, TF / IDF (Term Frequency / Inverse Document Frequency) method, etc. are common. Has been used.

  Subsequently, the browsing history search unit 33 receives the search word transmitted from the search word receiving unit 31 and the search result transmitted from the content search unit 32, and calculates the browsing importance for each content identifier. Then, the browsing importance is transmitted to the fitness composition unit 34 (S103). The browsing importance calculation method will be described later.

  The matching level synthesis unit 34 receives the search result transmitted from the content search unit 32 and the browsing importance level transmitted from the browsing history search unit 33, and calculates a score obtained by synthesizing the sentence importance level and the browsing importance level. (S104). The score calculation method is not limited to addition, but may be multiplication or the like.

  The processes of (S103) and (S104) are repeated for all content identifiers searched in (S102) (S105).

  Finally, the search result output unit 35 receives the scores for the respective content identifiers transmitted from the fitness synthesis unit 34, rearranges the content identifiers in descending order, and sends them to the user terminal 2 via the network 4. Output (S106).

  Thereafter, in the user terminal 2, for example, a search result as shown in FIG. 5 is output on the screen.

  Next, the browsing importance calculation method in (S103) will be described. FIG. 6 is a flowchart illustrating a browsing importance calculation method. The browsing importance level is calculated by the browsing history search unit 33 with reference to the browsing history storage unit 302 for each content identifier obtained in (S102).

  First, the browsing history search unit 33 reads the content identifier from the browsing history storage unit 302, calculates the number of content identifiers that match the content identifier transmitted from the content search unit 32, and browses the number of the content identifiers. The number of requests (C) is set (S201).

  For example, when the content identifier transmitted from the content search unit 32 is “www.goo.ne.jp/” and the content identifier shown in FIG. The number of times is 3 (C = 3).

  Next, the browsing history search unit 33 calculates the average value of the display order corresponding to the content identifier, and the number of identifiers stored in the browsing history storage unit 302 having a display order higher than the average value. Then, the ratio of the number of content identifiers stored in the browsing history storage unit 302 having the display order equivalent to the average value is obtained, and the ratio is set as the correction browsing request count (R) (S202).

  For example, in the case of FIG. 3, the average value is (4 + 2 + 8) / 3 = about 4.6. Here, the fractional part is rounded off and the average value is set to 5 which is an integer value. Then, the number of content identifiers having a display rank higher than 5, which is the average value, for example, the first rank, and the number of content identifiers having the fifth rank, which is a display rank equivalent to the average value of 5, are obtained. Here, when the number of content identifiers with the first display order is 160 and the number of content identifiers with the fifth display order is 80, the correction browsing request count is 80/160 = 0.5 (R = 0). .5).

  Subsequently, the browsing history search unit 33 obtains the number of search terms stored in the browsing history storage unit 302 that matches the search term transmitted from the search term receiving unit 31 and sets the number of search terms used (N). (S203).

  The order of calculation in (S201), (S202), and (S203) is not limited to the above, and (S202) → (S201) → (S203) or (S203) → (S201) → (S202). ) Etc., and may be in any order, and does not affect the effects obtained.

  Finally, the browsing history search unit 33 performs the browsing request count (C) calculated in (S201), the correction browsing request count (R) calculated in (S202), and the search calculated in (S203). Using the word usage count (N), the browsing importance (S) based on the following equation is calculated (S204).

S = C / R / N (1)
The browsing importance (S) shown in the formula (1) is proportional to the number of browsing requests (C), and the value of the browsing requested content becomes larger, so that the user's preference can be taken into account. The content desired by the user can be output higher.

  Further, the browsing importance (S) in the same formula divides the number of browsing requests (C) by the number of browsing requests for correction (R), so that the browsing request for the content identifier displayed at a higher level is the browsing importance (S ) And a browsing request for a content identifier displayed at a lower level will contribute more greatly. That is, the search results ordered based on the browsing importance (S) indicated by the same formula are displayed, and after the browsing history storage unit 302 is updated by a browsing request made by the user, other users are the same. When a situation is assumed in which a plurality of search requests such as searching by a search word is performed, the order of content identifiers displayed as search results is not the order desired by the user, but a factor that is fixedly displayed is eliminated. be able to. More specific explanation will be given later.

  Further, the browsing importance level (S) in the above formula is obtained by dividing the value obtained by dividing the number of browsing requests (C) by the number of browsing requests for correction (R) by the number of times of search word usage (N). It is possible to normalize the deviation in the size of the number of browsing requests (C) due to the difference. That is, regardless of what search terms are given, the value (S / N) obtained by dividing the number of browsing requests by the number of browsing requests for correction can be stored in a certain range.

  Note that, as a modification of the formula (1), the formula (2) can also be used.

S = C / R (2)
In addition, the logarithm is used as shown in Expressions (3) to (5) without using the values of the number of browsing requests (C), the number of browsing requests for correction (R), and the number of times of using search words (N) as they are. Even in such a case, the same effect can be obtained.

S = log (C) / R / N (3)
S = log (C) / log (R) / log (N) (4)
S = C / R / log (N) (5)
Next, the flow of processing at the browsing request stage will be described with reference to FIG.

  First, the browsing request receiving unit 36 transmits the browsing request transmitted from the user terminal 2 via the network 4 to the content search unit 32 and the browsing history update unit 37, and the browsing history update unit 37 browses the browsing request. The browsing history is updated by adding to the history storage unit 302 (S301).

  Next, the content search unit 32 searches the content storage unit 301 for the content location corresponding to the content identifier included in the browsing request, and acquires the content stored in the searched location (S302).

  The content output unit 38 receives the content acquired by the content search unit 32 and outputs it to the user terminal 2 via the network 4 (S303).

  Finally, by using the browsing importance (S) shown in the formulas (1) to (5), the search result order is not the order desired by the user, but the factor that is fixedly displayed is eliminated. This will be described more specifically with reference to FIG.

  For example, it is assumed that there are four contents, the content identifiers corresponding to the contents are C1, C2, C3, and C4, and the content desired by the user is C4. Also, assuming that the user searches using a certain search term, the sentence suitability of each content for that search term is 10, 9, 8, 7 as shown in the initial state of FIG. And Since the number of browsing requests (C) for each content in this initial state is zero, the calculated score is determined only by the text suitability, and naturally, the display order of search results is 1, 2, 3, 4. .

  When the search results having this display order are presented to the user for a certain period and the number of browsing requests added in (S301) is observed, the number of browsing requests for C4 desired by the user is expected to be the largest. .

  However, since the search results are displayed in the order shown in FIG. 8A, as described above, as shown in FIG. 9, for example, as shown in FIG. Become. Accordingly, the actual number of browsing requests (C) is, for example, 10, 7, 5, 5.

  A search result when a search request is newly made by the same search word in a state where these browsing requests are stored in the browsing history storage unit 302 will be described. In the following, when the browsing request count is set as the browsing importance level (S = C), and when the browsing request count is divided by the correction browsing request count (R), the browsing importance level is set (S = C / The difference between the calculation in (R) and will be described.

  In the case of S = C, the score calculated by the fitness level synthesis unit 34 is the result of adding the text fitness level and the browsing importance level, and therefore, as shown in FIG. 12 and the display order of search results is 1, 2, 3, and 4. That is, the display order of C4 desired by the user is the same as the display order in the initial state. That is, the search result is displayed in a fixed manner, and the content desired by the user cannot rise to the top.

  On the other hand, when S = C / R, for example, when the value of R is 1, 0.7, 0.5, 0.4 in order from the first place as shown in FIG. Then, since the number of browsing requests is counted regardless of the difference in the search terms, the phenomenon that the user-desired C4 is in the fourth place can be leveled as a whole. Therefore, as shown in FIG. 8C, the scores calculated by the fitness composition unit 34 are 20, 19, 18, 19.5, and the display order of the search results is 1, 3, 4, 2. . Thus, in the case of S = C / R, the user's desired C4 can rise from the fourth place in the initial state to the second place, and the user can easily reach the desired content. .

  In this way, in the calculation of the browsing importance (S), by dividing the number of browsing requests (C) by the number of browsing requests for correction (R), browsing requests for content identifiers displayed higher in the search results are made. Relatively low evaluation, and conversely, high evaluation of browsing requests for content identifiers displayed at lower levels prevents the display order from being fixed and raises the user's desired content to higher levels. Is possible. Similarly, the display order of other C1, C2, and C3 rises or falls depending on the degree desired by the user. As a result, in the case of S = C / R, it is possible to change the display order of the search results to the order desired by the user regardless of the initial state.

  According to the present embodiment, when a search word included in a search request is received, a plurality of contents including the search word are read from the content storage unit 301, and the frequency at which the search word appears in the plurality of contents The content search unit 32 for calculating the sentence matching degree indicating each of the content identifiers, and the content identifier corresponding to each of the plurality of contents, and the number of content identifiers stored in the browsing history storage unit 302 matching the content identifier Each of the browsing requests is obtained and the average value of the display ranks corresponding to the content identifiers stored in the browsing history storage unit 302 matching the content identifiers is calculated, and the browsing history having a display rank higher than the average value. Display order equivalent to the average value for the number of content identifiers stored in the storage unit 302 A browsing history search unit that calculates the ratio of the number of content identifiers stored in the browsing history storage unit 302 and obtains the number of browsing requests for correction and calculates the degree of browsing importance divided by the number of browsing requests for correction. 33, and a fitness synthesis unit 34 for calculating scores obtained by synthesizing the text fitness and the browsing importance for the plurality of content identifiers read from the content storage unit 301 by the browsing history search unit 33, Since it has a search result output unit 35 that outputs a plurality of content identifiers based on this score, it prevents the search results from being displayed in a fixed manner, and sets the content identifiers of the content in an order that reflects the user's preferences. Can be output.

  According to the present embodiment, the browsing history search unit 33 obtains the number of search terms stored in the browsing history storage unit 302 that matches the search term included in the search request and sets the number of search terms used, and the browsing important The degree is a value obtained by dividing the number of browsing requests by the number of browsing requests for correction, and further divided by the number of times of use of search words, so that it is possible to normalize the bias in the number of browsing requests due to differences in search terms. That is, regardless of what search terms are given, a value obtained by dividing the number of browsing requests by the number of browsing requests for correction can be stored in a certain area.

  According to the present embodiment, since the display order higher than the average value is first, it is possible to more reliably prevent the search result from being displayed in a fixed manner, and the content identifiers of the contents in an order more reflecting the user's preference Can be output.

[Second Embodiment]
FIG. 10 is a configuration diagram showing a configuration of the information search system 1 in the second embodiment. The information search system 1 in the present embodiment is basically the same as that in the first embodiment, and the configuration of the information search apparatus 3 is partially different.

  The information search apparatus 3 includes a browsing history batch update unit 39, a browsing request count storage unit 303 connected to the browsing history search unit 33, a browsing history temporary unit connected to the browsing history update unit 37 and the browsing history batch update unit 39. The storage unit 304 is further provided. Since other configurations are the same as those described in the first embodiment, a duplicate description is omitted here.

  The browsing request count storage unit 303 is referred to by the browsing history search unit 33. As shown in FIG. 11, the first table that associates the display order with the correction browsing request count (R), the search term, and the search term use count A second table for associating (N) is stored.

  The browsing history temporary storage unit 304 stores a browsing history of browsing requests for a certain period of time by the user in response to an update request from the browsing history update unit 37, and the stored browsing request is browsed by the browsing history batch update unit 39. This is reflected in the browsing history of the history storage unit 302.

  The browsing history batch update unit 39 reflects the browsing history stored in the browsing history temporary storage unit 304 in the browsing history storage unit 302.

  Next, the flow of processing of the information search device 3 in the present embodiment will be described. As in the first embodiment, the search request stage and the browse request stage are composed of two stages. First, the flow of processing in the search request stage will be described.

  The flow of processing at the search request stage is basically the same as (S101) to (S106) described in the first embodiment, but the browsing importance calculation method in (S103) is different. The calculation method will be described with reference to FIG.

  First, the browsing history search unit 33 reads the content identifier from the browsing history storage unit 302, calculates the number of content identifiers that match the content identifier transmitted from the content search unit 32, and browses the number of the content identifiers. The number of requests (C) is set (S401).

  Next, the browsing history search unit 33 refers to the browsing history storage unit 302, calculates the average value of the display order corresponding to the content identifier, and refers to the first table stored in the browsing request count storage unit 303. Then, the correction browsing request count (R) corresponding to the display order equivalent to the average value is read (S402).

  Subsequently, the browsing history search unit 33 refers to the second table stored in the browsing request count storage unit 303, and uses the search term usage count of the search term that matches the search term transmitted from the search term reception unit 31 ( N) is read out (S403).

  Finally, the browsing history search unit 33 uses the browsing request count (C) calculated in (S401), the correction browsing request count (R) read in (S402), and the search word usage read in (S403). Using the number of times (N), the browsing importance level (S) is calculated based on the formula (1) described in the first embodiment (S404).

  Next, the flow of processing at the browsing request stage will be described with reference to FIG.

  First, the browsing request receiving unit 36 transmits the browsing request transmitted from the user terminal 2 via the network 4 to the content search unit 32 and the browsing history update unit 37, and the browsing history update unit 37 browses the browsing request. The information is added to the history temporary storage unit 304 (S501).

  Next, the browsing history batch update unit 39 adds the browsing history stored in the browsing history temporary storage unit 304 to the browsing history storage unit 302 when the predetermined time has elapsed, and the browsing history temporary storage unit 304 is emptied. (S502).

  Here, all of the browsing histories stored in the browsing history temporary storage unit 304 may be added to the browsing history storage unit 302 or may be a part thereof.

  Subsequently, the content search unit 32 searches the content storage unit 301 for the content location corresponding to the content identifier included in the browsing request, and acquires the content stored in the searched location (S503).

  The content output unit 38 receives the content acquired by the content search unit 32 and outputs it to the user terminal 2 via the network 4 (S504).

  According to the present embodiment, since the browsing request count storage unit 303 uses the correction browsing request count (R) and the search term usage count (N) provided in advance, the browsing importance level can be calculated at a higher speed. And the search result can be provided to the user faster.

  According to the present embodiment, since the browsing history temporary storage unit 304 and the browsing history batch update unit 39 are further used, a part of the browsing history stored in the browsing history temporary storage unit 304 for a certain period is used as the browsing history storage unit 302. Can be reflected.

[Third Embodiment]
FIG. 14 is a configuration diagram illustrating a configuration of the information search system 1 according to the third embodiment. The information search system 1 in the present embodiment is basically the same as that in the first embodiment, and the configuration of the information search apparatus 3 is partially different.

  The information search device 3 is configured to further include a browsing importance level update unit 40 that can refer to the content storage unit 301 and the browsing history storage unit 302 in place of the browsing history search unit 33. Other configurations are the same as those described in the first embodiment, and a duplicate description is omitted here.

  The browsing importance level update unit 40 refers to the browsing history storage unit 302, calculates the browsing importance level for each content, and stores it in the content storage unit 301.

  As shown in FIG. 15, the content storage unit 301 includes content text information indicating the content itself, a content identifier that can identify the content, a location where the content is stored, and a browsing importance level update unit. The browsing importance for each search word updated by 40 is stored in association with each other.

  Next, the flow of processing of the information search device 3 in the present embodiment will be described. As in the first embodiment, the search request stage and the browse request stage are composed of two stages. First, the flow of processing in the search request stage will be described.

  First, the search word receiving unit 31 receives a search word included in the search request transmitted from the user terminal 2 via the network 4, and transmits the search word to the content search unit 32 (S601).

  Next, the content search unit 32 receives the search term transmitted from the search term receiving unit 31, reads a plurality of contents including the search term and the content identifier from the content storage unit 301, and the search term is assigned to each content. A sentence matching degree indicating the frequency of appearance is calculated, and a search result including the content identifier and the sentence matching degree is transmitted to the matching degree synthesizing unit 34 (S602).

  Subsequently, the content search unit 32 reads, for each content identifier, the browsing importance level corresponding to the transmitted search word from the content storage unit 301, and transmits the browsing importance level to the relevance level synthesis unit 34 (S603). ).

  The matching level synthesis unit 34 receives the search result and the browsing importance level transmitted from the content search unit 32, and calculates a score obtained by synthesizing the sentence importance level and the browsing importance level (S604).

  The processes of (S603) and (S604) are repeated for all content identifiers searched in (S602) (S605).

  Finally, the search result output unit 35 receives the scores for the respective content identifiers transmitted from the fitness synthesis unit 34, rearranges the content identifiers in descending order, and sends them to the user terminal 2 via the network 4. It outputs (S606).

  The browsing importance level update unit 40 refers to the browsing history storage unit 302 and updates the browsing importance level of the content storage unit 301 when a predetermined time has elapsed. The calculation method of the browsing importance is the same as the calculation method described in the first embodiment.

  Further, since the flow of processing at the browsing request stage is the same as that in the first embodiment, a duplicate description is omitted here.

  According to this embodiment, since the browsing importance calculated in advance is used, the search result can be provided to the user more quickly.

It is a block diagram which shows the structure of the information search system in 1st Embodiment. It is a figure which shows the table stored in the content storage part. It is a figure which shows the browsing history stored in the browsing history storage part. It is a flowchart which shows the flow of a process of a search request | requirement step. It is a figure which shows an example of a search result. It is a flowchart which shows the calculation method of browsing importance. It is a flowchart which shows the flow of a process of a browsing request | requirement step. It is a comparison figure which shows the display order of the case where it is set as an initial state and S = C, and S = C / R. It is a figure which shows an example of a display order, the browsing request frequency, and the correction browsing request frequency. It is a block diagram which shows the structure of the information search system in 2nd Embodiment. It is a figure which shows the 1st table and 2nd table which were stored in the browsing request frequency storage part. It is a flowchart which shows the calculation method of browsing importance. It is a flowchart which shows the flow of a process of a browsing request | requirement step. It is a block diagram which shows the structure of the information search system in 3rd Embodiment. It is a figure which shows the table stored in the content storage part. It is a flowchart which shows the flow of a process of a search request | requirement step.

Explanation of symbols

DESCRIPTION OF SYMBOLS 1 ... Information search system 2 ... User terminal 3 ... Information search device 4 ... Network 31 ... Search term receiving part 32 ... Content search part 33 ... Browsing history search part 34 ... Matching degree synthesis part 35 ... Search result output part 36 ... Browsing request Receiving unit 37 ... browsing history update unit 38 ... content output unit 39 ... browsing history batch update unit 40 ... browsing importance update unit 301 ... content storage unit 302 ... browsing history storage unit 303 ... browsing request count storage unit 304 ... browsing history temporary Storage

Claims (12)

Content storage means for storing the content and an identifier of the content in association with each other;
The search term used for the content search request, the identifier of the content browsed based on the search result based on the search term, and the display order indicating the order in which the identifier is displayed as the search result are stored in association with each other. Browsing history storage means;
When a search term included in the search request is received, a plurality of contents including the search term are read from the content storage unit, and sentence matching degrees indicating how often the search term appears in the plurality of contents are respectively set. Calculate
The identifier corresponding to each of the plurality of contents is further read from the content storage unit, and the number of identifiers stored in the browsing history storage unit that matches the identifier is determined as the number of browsing requests,
The average number of display ranks corresponding to the identifiers stored in the browsing history storage means matching the identifiers is calculated, and the number of identifiers stored in the browsing history storage means having a display rank higher than the average value In contrast, the number of identifiers stored in the browsing history storage means having a display order equivalent to the average value is determined as the number of browsing requests for correction,
Retrieval means for calculating each browsing importance obtained by dividing the number of browsing requests by the number of browsing requests for correction;
Relevance combining means for calculating scores obtained by combining the sentence relevance and the browsing importance for a plurality of identifiers read from the content storage means by the search means;
Search result output means for outputting the plurality of identifiers based on the score;
An information retrieval apparatus comprising:   The search means obtains the number of search terms stored in the browsing history storage means that matches the search term included in the search request and sets the number of search terms used. The information search apparatus according to claim 1, wherein the value divided by the correction browsing request count is further divided by the search word usage count.   The information search apparatus according to claim 1, wherein a display rank higher than the average value is first. A first step of associating a content with an identifier of the content and storing the content in the content storage means;
A browsing history storage unit that associates a search term used for a content search request, an identifier of a content browsed based on a search result based on the search term, and a display rank indicating a rank in which the identifier is displayed as a search result. A second step of storing in
When a search term included in the search request is received, a plurality of contents including the search term are read from the content storage unit, and sentence matching degrees indicating how often the search term appears in the plurality of contents are respectively set. Calculate
The identifier corresponding to each of the plurality of contents is further read from the content storage unit, and the number of identifiers stored in the browsing history storage unit that matches the identifier is determined as the number of browsing requests,
The average number of display ranks corresponding to the identifiers stored in the browsing history storage means matching the identifiers is calculated, and the number of identifiers stored in the browsing history storage means having a display rank higher than the average value In contrast, the number of identifiers stored in the browsing history storage means having a display order equivalent to the average value is determined as the number of browsing requests for correction,
A third step of calculating each browsing importance obtained by dividing the number of browsing requests by the number of browsing requests for correction;
A fourth step of calculating a score obtained by combining the sentence suitability and the browsing importance for each of the plurality of identifiers read from the content storage unit in the third step;
A fifth step of outputting the plurality of identifiers based on the score;
A method for retrieving information, comprising:   In the third step, the number of search terms stored in the browsing history storage means that matches the search term included in the search request is obtained and used as the number of search term use. 5. The information search method according to claim 4, wherein a value obtained by dividing the number of times by the number of browsing requests for correction is a value obtained by further dividing the number of times by using the search word.   The information search method according to claim 4, wherein a display rank higher than the average value is first. A first process of associating the content with the content identifier and storing the content in the content storage means;
A browsing history storage unit that associates a search term used for a content search request, an identifier of a content browsed based on a search result based on the search term, and a display rank indicating a rank in which the identifier is displayed as a search result. A second process stored in
When a search term included in the search request is received, a plurality of contents including the search term are read from the content storage unit, and sentence matching degrees indicating how often the search term appears in the plurality of contents are respectively set. Calculate
The identifier corresponding to each of the plurality of contents is further read from the content storage unit, and the number of identifiers stored in the browsing history storage unit that matches the identifier is determined as the number of browsing requests,
The average number of display ranks corresponding to the identifiers stored in the browsing history storage means matching the identifiers is calculated, and the number of identifiers stored in the browsing history storage means having a display rank higher than the average value In contrast, the number of identifiers stored in the browsing history storage means having a display order equivalent to the average value is determined as the number of browsing requests for correction,
A third process for calculating each browsing importance obtained by dividing the number of browsing requests by the number of browsing requests for correction;
A fourth process for calculating a score obtained by combining the sentence suitability and the browsing importance for each of the plurality of identifiers read from the content storage unit in the third process;
A fifth process for outputting the plurality of identifiers based on the score;
An information search program for causing a computer to execute the above .   In the third process, the number of search terms stored in the browsing history storage unit that matches the search term included in the search request is obtained and used as the number of search term use. 8. The information search program according to claim 7, wherein a value obtained by dividing the number of times by the number of browsing requests for correction is a value obtained by further dividing by the number of times the search word is used.   9. The information search program according to claim 7, wherein a display rank higher than the average value is first. A first process of associating the content with the content identifier and storing the content in the content storage means;
A browsing history storage unit that associates a search term used for a content search request, an identifier of a content browsed based on a search result based on the search term, and a display rank indicating a rank in which the identifier is displayed as a search result. A second process stored in
When a search term included in the search request is received, a plurality of contents including the search term are read from the content storage unit, and sentence matching degrees indicating how often the search term appears in the plurality of contents are respectively set. Calculate
The identifier corresponding to each of the plurality of contents is further read from the content storage unit, and the number of identifiers stored in the browsing history storage unit that matches the identifier is determined as the number of browsing requests,
The average number of display ranks corresponding to the identifiers stored in the browsing history storage means matching the identifiers is calculated, and the number of identifiers stored in the browsing history storage means having a display rank higher than the average value In contrast, the number of identifiers stored in the browsing history storage means having a display order equivalent to the average value is determined as the number of browsing requests for correction,
A third process for calculating each browsing importance obtained by dividing the number of browsing requests by the number of browsing requests for correction;
A fourth process for calculating a score obtained by combining the sentence suitability and the browsing importance for each of the plurality of identifiers read from the content storage unit in the third process;
A fifth process for outputting the plurality of identifiers based on the score;
A computer-readable recording medium on which is recorded an information search program characterized in that the computer is executed .   In the third process, the number of search terms stored in the browsing history storage unit that matches the search term included in the search request is obtained and used as the number of search term use. 11. The computer-readable recording medium storing an information search program according to claim 10, wherein a value obtained by dividing the number of times by the number of correction browsing requests is further divided by the number of times of use of the search word.   The computer-readable recording medium recording the information search program according to any one of claims 10 and 11, wherein a display rank higher than the average value is first.

Images