JP2009223372A - Recommendation device, recommendation system, control method for recommendation device and control method for recommendation system - Google Patents

Abstract

<P>PROBLEM TO BE SOLVED: To solve the following problems: because it is necessary to previously determine a characteristic phrase (e.g. "a cool musical piece" or "tears overflow") inside a review in conventional recommendation technology wherein the predetermined characteristic phrase is extracted from the review such as an introduction sentence of a musical piece, and wherein a vector value according to appearance frequency thereof is used, trouble of deciding the characteristic phrase is taken; and when the review does not include the characteristic phrase at all, it is not possible to refer to the review. <P>SOLUTION: In this recommendation device, by morpheme-analyzing the review to each commodity, a plurality of keywords are automatically extracted, and a point is imparted according to the appearance frequency thereof. The recommendation device has a function of specifying the keywords performing accordance or the like between the commodities, and deciding similarity between both the commodities by a total value of the points attached to the keywords. <P>COPYRIGHT: (C)2010,JPO&INPIT

Description

  The present invention relates to a technique for recommending information that matches a user's hobbies and preferences to the user.

  Accumulating and analyzing user behavior history information on a conventional website, for example, for a user who purchased the product A, a product B that many other users who purchased the same product A prefer to purchase it. Services such as recommendation are provided. However, the above services refer to the purchase history of multiple other users, so even though there is a limitation of buyers of the same product, `` popular products '' bought by many users eventually tend to be recommended Is strong.

Therefore, Patent Document 1 discloses the following technique. That is, a predetermined characteristic phrase is extracted from a review such as an introduction sentence or an impression sentence of a musical piece, and a vector value for each musical piece is calculated based on the appearance frequency of the characteristic phrase. Then, if the angle between the vectors is small, the two music pieces have similar characteristics, and the other music piece is recommended to the user who purchased one piece of music. In this way, with the technique of Patent Document 1, it is possible to make a recommendation based on the similarity of the music itself, regardless of factors such as popularity (the number of purchasers is large).
JP 2007-280342 A

  However, in the above conventional technique, it is necessary to determine in advance a characteristic phrase (for example, “cool song”, “tears overflow”) in the review. Therefore, it takes time to determine the characteristic phrase. In addition, there is a problem that a review that does not include any characteristic phrase cannot be used as a reference.

  In order to solve the above-described problems, the present invention automatically extracts a plurality of keywords by performing a morphological analysis on reviews for each product and assigns points according to the appearance frequency. And a recommendation apparatus provided with the function which identifies the keyword etc. which match between goods, and judges the similarity between both goods with the total value of the score attached | subjected to the keyword is provided.

  Specifically, a sentence acquisition unit that acquires a sentence, an extraction part that extracts a word by associating the acquired sentence with a morphological analysis, and a weighting part that weights the extracted word with an appearance frequency in the sentence, , An accumulation unit that accumulates weighted results in association with sentences, an arithmetic expression storage unit that stores an arithmetic expression for comparing sentences using a weighted result product for each word, a weighted result of the accumulation unit, and an arithmetic expression storage It is a recommendation device which has a comparison part which compares sentences using an operation formula of a part, and an output part which outputs a comparison result.

  In addition, by combining the recommendation device and a content introduction server device having a function of accepting and publishing a posted review from a general user via a network, such as a so-called “review site”, posting to the content introduction server device is possible. A recommendation system is also provided which is configured so that the reviewed reviews can be used for comparison judgment processing such as similarity in a recommendation device.

  Specifically, an introduction section that allows browsing of content introduction information via a network, an impression sentence reception section that accepts impression sentences about the content via the network, a first browsing section that allows the received impression sentences to be viewed on the network, A recommendation system comprising: a content introduction server device comprising: and the above recommendation device having a comment sentence acquisition means for the sentence acquisition unit to acquire a comment sentence received by the comment sentence reception unit of the content introduction server device as a sentence. It is.

  FIG. 20 is a diagram illustrating an example of a movie recommendation screen by a conventional recommendation system and a movie recommendation screen by the recommendation system of the present invention having the above-described configuration. As shown in FIG. 20 (a), in the conventional recommendation system, for users who are interested in the movie “(not shown for public order and morals violation)” depicting the crisis of the destruction of the earth by meteorites flying from space, Popular movies include the pirate film “(Not listed for violations of public order and morals)”, the human drama “(Not published for violations of public order and morals)”, and the time science fiction product “(Not published for violations of public order and morals)” Because it is, it is recommended.

  On the other hand, as shown in FIG. 20B, in the recommendation system of the present invention having the above-described configuration, the same movie “(not shown for violation of public order and morals)” is a panic movie by Toru Soseki “( ) ", Panic movies due to changes in the Earth's internal core (" Not published regarding public order and morals violations "), and SF movie" (Not published regarding public order and morals violations) "depicting the invasion of aliens. You can see that it is recommended.

  According to the present invention configured as described above, it is possible to determine the similarity of sentences using the total value according to the appearance frequency of keywords automatically extracted by morphological analysis. Furthermore, by using introductory texts and post reviews on numerous products that are publicly available on the network, the content itself shown in the text is not affected by factors such as popularity (number of buyers). Recommendations can be made based on similarity.

  Hereinafter, embodiments of the present invention will be described with reference to the drawings. Note that the present invention is not limited to these embodiments, and can be implemented in various modes without departing from the spirit of the present invention.

  In the first embodiment, claims 1, 6, 7, and 12 will be mainly described. In the second embodiment, claims 2 and 8 will be mainly described. In the third embodiment, claims 3 and 9 will be mainly described. In the fourth embodiment, claims 4 and 10 will be mainly described. In the fifth embodiment, claims 5 and 11 will be mainly described.

  Example 1

    <Overview>

  FIG. 1 is a conceptual diagram for explaining an example of a recommendation by the recommendation device of this embodiment. As shown in FIG. 1A, the recommender of this embodiment performs a morphological analysis process on the sentence A written about “unification of force”, and includes, for example, “gravity”, “electromagnetic force” included in the sentence, Extract words such as “weak power”, “strong power”, “unified theory”. Then, according to the appearance frequency in the sentence, for example, “gravity 27”, “electromagnetic force 17”,.

  Similarly, for the sentence B to be compared with the sentence A, words are similarly extracted by morphological analysis, and a weighted score is assigned to each word such as “Big Bang 51”, “Unification 27”,.

  The same (synonymous) extracted words in both sentences, such as “gravity” and “weak boson”, are weighted product “gravity 27 × 12 = 324”, “glue boson”. 11 × 20 = 220 ”,... And the index value which shows the content similarity of both sentences is calculated | required by totaling the calculated value. Then, for example, using the sentence A as a reference, index values are similarly obtained for the sentences C, D,..., And as a result, the sentence B having the largest index value is most similar to the sentence A. Judge that.

  Then, in the recommendation device of the present embodiment, using such a determination result, for example, the browsing of the Web page B on which the sentence B is described is recommended to the viewer of the Web page A on which the sentence A is posted. That's it. In addition, if the text to be compared here is an introductory text or impression review of a product or service, it is considered that the similarity between the products is indicated by the similarity of the text, and the purchaser of one product etc. The recommendation is to recommend the purchase of the other product.

    <Functional configuration>

  FIG. 2 is a diagram illustrating an example of functional blocks in the recommendation device according to the present embodiment. As shown in this figure, the “recommendation device” (0200) of the present embodiment includes a “sentence acquisition unit” (0201), an “extraction unit” (0202), a “weighting unit” (0203), and an “accumulation”. Part "(0204)," comparison part "(0205)," arithmetic expression storage part "(0206), and" output part "(0207).

  The functional blocks of the present apparatus described below and other apparatuses described in the second embodiment can be realized as hardware, software, or both hardware and software. Specifically, if a computer is used, a CPU, a main memory, a bus, or a secondary storage device (a hard disk, a non-volatile memory, a storage medium such as a CD or a DVD, a read drive for the medium, etc.), information Input devices used for input, printing equipment, display devices, other hardware components such as external peripheral devices, interfaces for external peripheral devices, communication interfaces, driver programs for controlling these hardware, Other examples include application programs and user interface applications.

  Then, through the arithmetic processing of the CPU according to the program developed on the main memory, the data input from the input device or other interface etc. and stored in the memory or hard disk is processed and stored, or each of the above hardware and software An instruction for controlling the above is generated. In addition, the present invention can be realized not only as an apparatus or a system but also as a method. A part of the invention can be configured as software. Furthermore, a software product used for causing a computer to execute such software and a recording medium in which the product is fixed to a recording medium are naturally included in the technical scope of the present invention (the same applies throughout the present specification). Is).

  The “text acquisition unit” (step S0201) has a function of acquiring text, and can be realized by, for example, a communication IF (interface), a user input device, a GUI, a text acquisition program, and the like. Specifically, the sentence acquisition unit may be configured to acquire HTML document data and other document data via a network using, for example, a crawler application. When configured in this way, it is possible to acquire text on a Web page on the network.

  In addition, the sentence acquisition unit may be configured to acquire a sentence input by a user through keyboard input or the like via, for example, a GUI. In this case, a document input and created on the user terminal can be acquired. In addition, the text described in a paper document or the like may be automatically converted into digital data and acquired by an imaging device, a scanner, an OCR (optical character recognition) application, or the like.

  The “extractor” (0202) has a function of morphologically analyzing the acquired text and extracting a word in association with the text, and can be realized by an extraction program including an arithmetic device such as a CPU or a morphological analysis program, for example. . Specifically, the extraction unit performs morphological analysis processing of various sentences acquired by the sentence acquisition unit according to a morphological analysis program. As for morphological analysis processing, as in the past, for example, word decomposition processing of text by pattern matching using a word dictionary, syntax rule dictionary, etc., or word decomposition by scoring using a probabilistic language model such as a hidden Markov model It is good to do by processing. Of course, the extracted word may be an idiomatic phrase composed of a plurality of words depending on the setting of analysis rules for morphological analysis.

  Then, through the morphological analysis process, words such as “electromagnetic force” and “gravity” are extracted from the acquired sentences. Subsequently, for example, table data in which the sentence ID and the extracted word are associated with each other is generated according to the extraction program and stored in the flash memory or the like. As described above, in this embodiment, since a word is arbitrarily extracted from a sentence, it is not necessary to define a specific phrase (extracted word).

  The “weighting unit” (0203) has a function of weighting the extracted word by the appearance frequency in the sentence, and can be realized by an arithmetic unit such as a CPU or a weighting program, for example. “Weighted by appearance frequency” refers to a process of assigning a different value for each word according to the appearance frequency in a sentence. For example, the number of occurrences of all words in a sentence (for example, only a noun case may be used, and particles and auxiliary verbs may be excluded) is used as a parameter, and the number of occurrences of each of words A, B, C,. It is a condition that the ratio which shows is made into the weight value. In addition, the number of appearances may be used as a weight value.

  The extracted words may be configured such that a plurality of synonyms (synonyms) are collected as one word by referring to a synonym (synonyms) dictionary and the like and weighted according to the appearance frequency. Further, for example, according to the part of speech or the content of the character, a correction table or the like may be referred to correct the weight value for each word. Specifically, if the sentence is a review of meal impressions, etc., words such as “delicious” and “spicy” are corrected so that the weighted value is increased.

  Then, using the weighting values assigned to the words in the sentence as described above, the recommendation device according to the present embodiment compares the sentences with the following configuration.

  The “accumulation unit” (0204) has a function of accumulating the weighted result in association with the text, and can be realized by, for example, an HDD (hard disk drive), a flash memory, another recording medium, or a recording program.

  FIG. 3 is a diagram illustrating an example of weighting results stored in such a storage unit. As shown in this figure, for example, a “review ID” is assigned to each sentence, and the review ID, appearance keyword, and weight value thereof are shown as table data. Further, when the sentence is related to some content (for example, an introduction sentence or an impression sentence of a certain movie content), a “content ID” may be further associated with the review ID. That is, in the top record in the table data shown in FIG. 3, the introductory sentence regarding the movie (content) indicated by “content ID: id 328907” is “review ID: 111111”. In addition, “electromagnetic force (weight value 57)”, “big bang (weight value 1)”, and the like are shown as extracted words in the introduction sentence.

  The “arithmetic expression storage unit” (0205) has a function of storing arithmetic expressions, and is realized by storing and holding arithmetic expressions in, for example, an HDD (hard disk drive), a flash memory, other recording media, or a recording program. can do. The “arithmetic expression” is an expression for comparing sentences using a weighted result product for each word. For example, a product of weighted values of the same word between two sentences is calculated, and a calculated value of all words is calculated. Based on a formula that adds all, there is a formula that multiplies each parameter of the basic formula by a correction value or adds a correction value. Then, when the same word appears frequently in each of the two sentences, the calculation result of the above calculation formula takes a larger value than when the same word does not occur frequently. Therefore, the similarity of the appearance word between two sentences can be judged by comparing the calculation results by such an arithmetic expression. And the similarity of the content between two sentences can be guessed from the similarity of the appearance word.

In addition, the following formula | equation is mentioned as a specific example of a computing equation.

In addition, the said arithmetic formula has shown the following arithmetic processes. That is, first, an identification number n (1 to k) is assigned to a word extracted from the comparison source sentence α stored in the storage unit as described above, and a weight value of the word is set to P _n . Similarly, an identification number m (1 to 1) is assigned to a word extracted from the sentence β to be compared, and a weight value of the word is set to C _m . Then, whether or not the keyword n (1 to k) of the sentence α and the keyword m (1 to l) of the sentence β coincide (synonymous with each other) is determined brute-force.

As a result of the determination process, the product of the weights is calculated as f (n, m) = (P _n × (C _m × U)) if they match, and if they do not match, f (n, m m) = 0. In this way, a product of weighted values of words such as matching keywords between two sentences is obtained, and the products of the weighted values are integrated.

Further, the arithmetic expression stored here does not necessarily need to use a weighted result product for each word. For example, an arithmetic expression that uses a weighted result sum instead of a weighted result product, that is, f (n, m) = (P _n + (C _m × U)) or the like may be used. Alternatively, an arithmetic expression using an arithmetic average, a geometric average, a harmonic average, or the like of the weighted result instead of the weighted result product may be used.

  The “comparison unit” (0206) has a function of comparing sentences using the weighted result of the storage unit and the arithmetic expression of the arithmetic expression storage unit, and is realized by an arithmetic device such as a CPU or a comparison program. Can do. Specifically, the arithmetic expression stored in the flash memory or the like and the weight value for each word are stored in the main memory as described above. Then, the CPU executes the arithmetic processing of the arithmetic expression using the weight value as a variable, and calculates the arithmetic result. And the magnitude of the result of the calculation is performed by a logical calculation process of the CPU, and the similarity comparison between sentences is executed.

  FIG. 4 is a conceptual diagram illustrating an example of comparison processing using the calculation and the calculation result in the comparison unit. As shown in this figure, with respect to the word “electromagnetic force” appearing in both the sentence A and the sentence B, for example, referring to the table data as shown in FIG. 3, the weight values “57” and “18” are obtained, respectively. Is done. Then, the weighted result product “1026” is calculated by multiplying the values according to the arithmetic expression. In addition, the weighted result products “384”, “294”,... “16” are calculated for the other words “gravity”, “unification theory”,. A total value α “2346” obtained by adding all the values is calculated.

  Subsequently, the total value “1872” is calculated for each sentence A and sentence C, which is different from B, using a weight value and an arithmetic expression for each word. Then, when the total value is compared in size, it is determined that the total value of the sentence A and the sentence B is larger than the total value of the sentence A and the sentence C. Therefore, when the similarity of the appearance word with the sentence A is compared between the sentence B and the sentence C, the sentence B can be determined to have a higher similarity with the sentence A.

  The “output unit” (0207) has a function of outputting the comparison result, and can be realized by, for example, an I / O (input / output) connected to an external device or an output program. Specifically, an output indicating the determination result as described above is performed from “I / O” to an internal or external arithmetic device. And in the said arithmetic unit, the recommendation process, such as a sentence in which the comparison result with high similarity was given, or a product related to the sentence, is executed.

  In addition, the following processes are mentioned as a recommendation process according to the comparison result output from the output part. For example, as described above, another sentence itself that is considered highly similar to one sentence from the comparison result may be recommended. Alternatively, as long as the comparison result is about a sales product or an introduction / impression sentence of content media such as a movie, a game, a book, etc., a product having a similar introduction sentence or the like may be recommended.

  In this way, the recommendation device of the present embodiment can determine the similarity of sentences according to the words automatically extracted by the morphological analysis process. Various recommendations based on the similarity of the text content itself can be performed according to the determination result.

  Note that the recommendation device of the present embodiment having the above-described configuration may be configured to be incorporated in a server device on the Internet, for example, and to make a recommendation according to the determination result in response to a request from a client terminal. Alternatively, it is configured to perform the above-described recommendation for performing comparison processing on Web page document data on a network that is incorporated in a client terminal device and held in a cache, or text data held in a local storage device. May be.

    <Hardware configuration>

  FIG. 5 is a schematic diagram illustrating an example of a configuration of the recommendation device when the above functional components are realized as hardware. The operation of each hardware component in the sentence similarity determination process will be described with reference to FIG.

  As shown in this figure, the recommendation device is an extraction unit, a weighting unit, and a comparison unit, and includes a “CPU” (0501) that executes various other arithmetic processes and a “main memory” (0502). ing. Also, a “flash memory” (0503) that is an accumulation unit and an arithmetic expression storage unit, an “I / O” that is a sentence acquisition unit and an output unit, and the like are provided. They are connected to each other via a data communication path such as a “system bus” to transmit / receive information and process information.

  A program is read into the “main memory”, and the “CPU” executes various arithmetic processes by referring to and interpreting the read program. In addition, a plurality of addresses are assigned to each of the “main memory” and “flash memory”, and in the calculation processing of the “CPU”, the addresses are specified and accessed to store the data. It is possible to perform arithmetic processing using.

  Here, “CPU” interprets the text acquisition program read to “main memory”, and according to the program, the recommender of this embodiment is controlled to wait for text data input from “I / O”. The Then, for example, a character data string input by a keyboard connected to “I / O” is acquired as a sentence. Alternatively, Web document data on the network is periodically acquired from “I / O” in accordance with a so-called “crawler program” included in the document acquisition program. Then, the sentence data 1 obtained in this way is stored at address 1 of the “main memory”.

  Subsequently, the “CPU” interprets the extraction program, and executes a process of extracting a word from the sentence 1 stored in the address 1 of the “main memory” according to the interpretation result. Specifically, for example, the “CPU” refers to the word dictionary stored at the address B of the “flash memory”, and uses the longest match method to search for “words in the dictionary” and “words in the sentence”. Perform pattern matching processing. For the extracted word, the “CPU” refers to the grammar dictionary in the “flash memory” and executes a process of determining whether the word is correctly extracted from the word part-of-speech used in the grammar dictionary and the connection relationship. As for the extracted word determined to be incorrect as a result of the determination, another delimiter is found by the re-pattern matching process, and the word in the sentence is extracted in a grammatically correct form. Here, the “CPU” may refer to a synonym dictionary stored in the “flash memory” in accordance with the extraction program and perform a process of collecting synonyms out of the extracted words into one word. Then, the extracted words are stored as keywords at addresses 2,... Of “main memory”.

  Subsequently, the “CPU” interprets the weighting program, and executes weighting processing on the word stored at the addresses 2... Of the “main memory” according to the interpretation result. Specifically, for example, the number of appearances of all words extracted from one sentence is counted. Subsequently, the number of occurrences of the word (keyword 1) to be weighted is counted, and the appearance ratio (%) in the sentence of the keyword 1 is calculated by the calculation process of “CPU”. The calculated value is stored as a weighted value for keyword 1 at address 3 of “main memory”. And the same calculation process is performed also about the other word (keyword 2, keyword 3, ...) which appears in a sentence, and the weight value for every word is calculated.

  Then, the “CPU” writes the calculation result in association with the identification information of the sentence 1 in accordance with the accumulation recording program, and writes it into the address D of the “flash memory” as a weighted result DB (database) as shown in FIG. In addition, the processing related to the above-described sentence acquisition program, extraction program, weighting program, and storage record program is similarly executed for other sentences (sentence 2, sentence 3,...), And sentences 2, 3,. The weight value for each word is written in the weighted result DB of “flash memory” in association with the identification information of the sentence.

  Subsequently, the “CPU” interprets the comparison program, and compares the two sentences (for example, sentence 1 and sentence 2) with reference to the weighted result DB and the arithmetic expression stored in the “flash memory” according to the interpretation result. Execute. Specifically, for example, by comparing the extracted word group of sentence 1 and the extracted word group of sentence 2 with reference to the weighted result DB, the top 100 words that appear in common in both sentences are specified. Subsequently, as shown in FIG. 4, a product of the weight values of the common appearance words is calculated by a calculation process of “CPU”, and a value obtained by adding the products is calculated. Then, the integrated value of the calculated weighted result product is stored in the address 4 of the “main memory” as a comparison index value indicating the content similarity between the sentences 1 and 2. Also, the integrated value of the weighted result product is calculated in the same manner as described above for the other sentences such as sentence 1 and sentence 3, sentence 2 and sentence 3, and written in the comparison index value DB of “flash memory”.

  For example, when recommending a sentence whose content is similar to that of sentence 1, the “CPU” refers to the comparison index value DB according to the comparison program, and the sentence having the largest weighted result product integrated value with sentence 1, for example, The identification information of sentence 2 is acquired. Then, the acquired identification information of the sentence 2 is stored as a comparison determination result at the address 5 of the “main memory”, and the determination result is output from “I / O” according to the output program. .

  Then, using the determination result that is output, the recommendation device sends the recommendation of the product associated with the sentence 2 or the sentence 2 to a predetermined client terminal (by a CPU calculation process according to a recommendation program similar to the conventional one). Execute for the terminal that sent the request for recommendation).

    <Process flow>

  FIG. 6 is a flowchart illustrating an example of a process flow in the recommendation device according to the present exemplary embodiment. The steps shown below may be steps executed by each hardware configuration of the computer as described above, or may be processing steps that constitute a program for controlling the computer recorded on a medium. I do not care.

  As shown in this figure, first, when a sentence is acquired (step S0601), the acquired sentence is subjected to morphological analysis (step S0602). Then, the words in the sentence found as a result of the morphological analysis are extracted in association with the sentence (identification information thereof) (step S0603). Subsequently, the extracted word is weighted by the appearance frequency in the sentence (step S0604), and the weighted result is associated with the sentence (identification information thereof) and recorded in the storage unit (step S0605).

  Then, an arithmetic expression for comparing sentences using the weighted result product for each word is acquired from the storage unit or the like (step S0606), and the weighted result stored in the storage unit and the acquired arithmetic expression are used. The sentences are compared (step S0607). Then, the comparison result is output (step S0608) and used for recommendation.

    <Brief description of effect>

  As described above, the recommendation device according to the present embodiment can determine the similarity of sentences according to words automatically extracted by morphological analysis processing. Various recommendations based on the similarity of the text content itself can be performed according to the determination result.

  << Example 2 >>

    <Overview>

  FIG. 7 is a diagram for explaining the feature points of the recommendation system including the recommendation device according to the first embodiment. As shown in this figure, the recommendation apparatus according to the present embodiment is characterized in that an impression sentence on a network such as “movie review on a Web page” is acquired as a sentence used for comparison for recommendation.

  That is, on the network, there are, for example, a gourmet review site, a book review site, or a product review site in addition to a movie review site, and various impressions regarding contents and products are disclosed on these sites. Therefore, in this embodiment, by acquiring a huge number of impressions about various things on such a network, for example, with reference to the impressions of many users regarding one movie, the contents with high accuracy are obtained. Similarity comparison judgment can be made.

    <Functional configuration>

  FIG. 8 is a diagram illustrating an example of functional blocks in the recommendation system of the present embodiment. As shown in this figure, the recommendation system according to the present embodiment includes a “recommendation device” (0800) and a “content introduction server device” (0810).

  The “recommendation device” is basically the same as that described in the first embodiment. That is, as shown in FIG. 8, the “recommendation device” (0800) is based on the above-described embodiment, and “text acquisition unit” (0801), “extraction unit” (0802), and “weighting unit” (0803). And “accumulation unit” (0804), “comparison unit” (0805), “arithmetic expression storage unit” (0806), and “output unit” (0807). In addition, since these structural requirements have already been described, the description thereof will be omitted.

  And the difference with Example 1 in this recommendation apparatus is a point in which a text acquisition part further has an "impression text acquisition means" (0808).

  The “impression sentence acquisition means” (0808) has a function of acquiring the impression sentence received by the impression sentence reception unit of the content introduction server device as a sentence. “Comments” refers to sentences written by users for products and services that include content, such as movies, books, music, TV programs, games, foodstuffs, restaurants, hotels, and various other products and services. This is a sentence written by the user.

  Then, in the recommendation device of the present embodiment, sentences (impression sentences) are compared with each other using the weighted result product of extracted words as described in the first embodiment, using the impression sentences acquired here. Then, using the comparison result, a recommendation is made based on the similarity of the product or service that is the impression target of the impression sentence.

  Next, functional blocks in the “content introduction server device” that is the acquisition source of the impression text acquired by the recommendation device will be described below with reference to FIG. Note that the content introduction server device refers to a server device having a function of publishing a web page (HTML document data) for introducing content and other text data on the network. The “content” introduced in the server device is a concept including not only so-called “information content” represented by digital media content but also products / service contents provided by general products and services. .

  As shown in FIG. 8, the “content introduction server device” (0810) of the present embodiment includes an “introduction part” (0811), an “impression sentence reception part” (0812), and a “first browsing part” ( 0813).

  The “introduction unit” (0811) has a function of browsing content introduction information via a network. “Content introduction information” refers to information prepared to inform the user of matters related to the content (contents, content creator, creation date, etc.). For example, content introduction / provided by the content provider For example, a web page for explanation is given.

  The “introduction unit” can be realized by, for example, a communication IF, an introduction program, a Web page on which a content introduction sentence is described, or the like. Specifically, the content introduction server device holds a web page for content introduction / explanation in an HDD or the like in this introduction unit. When receiving an HTTP request from a client terminal via the Internet network or the like, the Web page is returned to the client terminal accordingly. Then, the browser performs predetermined processing at the client terminal and displays the received Web page on the display so that the content introduction information is browsed through the network.

  The “impression sentence reception unit” (0812) has a function of accepting an impression sentence with respect to the content via the network, and can be realized by, for example, a communication IF, an introduction program, a Web page provided with an impression sentence input transmission form, and the like. . Specifically, in the content introduction web page transmitted to the client terminal by the action of the introduction unit, a link to the web page having an impression sentence input transmission form for describing an impression on the introduced content is clearly indicated. For example, it may be configured to be displayed.

  When the link is clicked on the client terminal and an HTTP request is transmitted, the content introduction server device returns a web page including a comment input form as an HTTP response accordingly. Then, the client terminal inputs an impression sentence for the content to the impression sentence input form of the Web page, and the impression sentence is sent to the content introduction server device in response to a click on a transmission button on the Web page. Will be sent. As a result, the content introduction server device accepts impressions about the content via the network.

  The “first browsing unit” (0813) has a function of browsing the received comment text on the network, and can be realized by, for example, a communication IF, a first browsing program, or a Web page for reading the comment text.

  FIG. 9 is a diagram illustrating an example of a Web page for browsing impressions on the network in this way. As shown in this figure, an impression sentence for a certain movie content accepted at a movie review site is published on the network as a “work user review”. In the present embodiment, the above-mentioned recommendation device is configured to acquire impression sentences that are widely collected and disclosed via the network in this way, so that the content comparison process using many sentences can be easily performed. The recommendation based on the similarity determined as a result of the comparison can be performed.

    <Hardware configuration>

  FIG. 10 is a schematic diagram illustrating an example of a configuration of the content introduction server device of the recommendation system when the above functional components are realized as hardware. The operation of each hardware component in the impression text acceptance process will be described with reference to this figure. An example of the hardware configuration of the recommendation device of the recommendation system according to the present embodiment has been described in the first embodiment, and a description thereof will be omitted.

  As shown in this figure, the content introduction server device is an introduction part, an impression sentence reception part, and a first browsing part, and also executes a “CPU” (1011) that executes various other arithmetic processes, and a “main memory”. (1012). Also, an “HDD” (1013) that holds HTML document data for each Web page, a “communication IF” (1014) that is connected to a client terminal or a recommendation device via a network, and the like are provided. They are connected to each other via a data communication path such as a “system bus” to transmit / receive information and process information.

  Here, an HTTP request indicating access to the introduction Web page transmitted from the client terminal A (1020A) or B (1020B) via the network is received by the “communication IF”. Then, the “CPU” interprets the introduction program read to the “main memory”, and acquires the introduction sentence Web page indicated by the HTTP request from the address “A” of the “HDD” according to the program. Is stored at address 1 of. Subsequently, an HTTP response including the Web page is generated and transmitted from the “communication IF” to the transmission source indicated by the HTTP request.

  Then, the client terminal receives the introductory text Web page returned in response to the request, and displays it on its own display through processing such as a browser. In addition, if the content introduction Web page is configured to display a hyperlink to an impression text reception Web page including an impression text input form, an impression text reception Web page can be obtained by clicking the hyperlink. An HTTP request for accessing the URL is transmitted from the client terminal to the content introduction server device. Of course, the transmission of this HTTP request may be executed by obtaining the URL in the comment text reception Web page by a route other than the hyperlink click in the introduction Web page, for example, search or other means.

  In the content introduction server device, the HTTP request transmitted from the client terminal in this way is received by the “communication IF”. Then, the “CPU” interprets the comment text receiving program read to the “main memory”, and acquires the comment text receiving Web page indicated by the HTTP request from the address B of “HDD” according to the program. Store at address 2 of "main memory". Subsequently, an HTTP response including the Web page is generated and transmitted from the “communication IF” to the transmission source indicated by the HTTP request.

  Then, the client terminal receives the received comment text reception Web page, and displays it on its own display through processing such as a browser. Then, the user's text (impression text for the content) input to the input form in the comment text reception Web page by using a keyboard or the like is transmitted from the client terminal to the content introduction server device by pressing the transmission button in the Web page. .

  In the content introduction server device, the sent impression text is received by the “communication IF” and stored in the address 3 of the “main memory”. Then, the “CPU” interprets the first browsing program read to the “main memory”, and generates a comment introduction web page including the comment obtained according to the program. Then, the generated comment sentence introduction Web page is recorded at address C of “HDD”. When an HTTP request indicating access to the comment introduction web page from the client terminal via the network is received by “communication IF”, the comment introduction web page is transmitted from the address C of “HDD” according to the first browsing program. Acquire and store in address 4 of “main memory”. Subsequently, an HTTP response including the Web page is generated and transmitted from the “communication IF” to the transmission source indicated by the HTTP request.

  Then, the client terminal receives the sentinel comment introduction Web page and displays it on its own display through processing such as a browser.

  In addition, the content introduction server device transmits the impression text stored at the address 3 of the “main memory” together with the recommendation device (0800) via the network from the “communication IF”. Then, the recommendation device executes the comparison judgment process between sentences using the comment sentence as described in the embodiment. Then, according to the comparison determination result, the recommendation of the content that is the subject of the impression sentence is executed.

    <Process flow>

  FIG. 11 is a flowchart illustrating an example of a process flow in the content introduction server device of the recommendation system according to the present exemplary embodiment. The steps shown below may be steps executed by each hardware configuration of the computer as described above, or may be processing steps that constitute a program for controlling the computer recorded on a medium. I do not care.

  As shown in this figure, first, the content introduction information is transmitted to the terminal for browsing via the network (step S1111). Further, when an impression text for the content is received via the network (step S1112), the received impression text is transmitted to the terminal for browsing on the network (step S1113). In addition, the impression text received together with it is sent as a sentence to the recommendation device (step S1114). Note that the flow of processing in the recommendation device is the same as that described in the first embodiment, except that the text to be acquired is limited to the impression text transmitted from the content introduction server device in this way. Omitted.

    <Brief description of effect>

  As described above, in the recommendation system according to the present embodiment, it is possible to acquire the comment sentences widely collected and disclosed via the network by using the recommendation device. Accordingly, it is possible to easily compare contents using a large number of sentences and make a recommendation based on the similarity determined as a result of the comparison.

  Example 3

    <Overview>

  FIG. 12 is a diagram illustrating an example of a content recommendation screen by the recommendation system of the present embodiment. Here, as described in the second embodiment, for example, post impressions for each movie are collected on a Web page where a movie impression (review) is posted. Then, through the comparison and determination process using the comment text in the recommendation device, a comparison result indicating that the movie having the highest integrated value of the weighted result product with respect to the movie “に 対 し て × 子” is “Ekiden” is output. Then, in the recommendation system of the present embodiment, the comparison result is displayed as shown in FIG. 12 on a Web page that can be browsed from the content introduction server device. As a result, the movie “Ekiden” can be recommended to users who are interested in the movie “XX child”.

  Further, as described in detail in the fifth embodiment, the integrated value of the weighted result product may be displayed on the recommendation screen. Thus, by presenting to the user the integrated value of the weighted result product that is the basis for the recommendation determination, the user can make a more objective determination with respect to the recommended result.

    <Functional configuration>

  FIG. 13 is a diagram illustrating an example of functional blocks in the recommendation system of the present embodiment. As shown in this figure, the recommendation system according to the present embodiment includes a “recommendation device” (1300) and a “content introduction server device” (1310) as in the second embodiment.

  The functional blocks of the “recommendation device” are the same as those described in the first embodiment and the second embodiment, and thus illustration and description thereof are omitted.

  In addition, the “content introduction server device” also has an “introduction section” (1311), an “impression sentence reception section” (1312), and a “first browsing section” (1313) based on the second embodiment. Have. In addition, since these structural requirements have already been described in the above embodiment, the description thereof will be omitted. In the recommendation system according to the present embodiment, the content introduction server device further includes a “comparison result acquisition unit” (1314) and a “second browsing unit” (1315).

  The “comparison result acquisition unit” (1314) has a function of acquiring a comparison result output from the recommendation device, and can be realized by, for example, a communication IF or a comparison result acquisition program. The comparison result refers to a comparison result between impression sentences using a weighted result and an arithmetic expression. For example, identification of an impression sentence having the highest integrated value of the weighted result product as described in the first embodiment. Information. In addition, the integrated value itself of the weighted product may be acquired as the comparison result.

  The “second browsing unit” (1315) has a function of browsing the acquired comparison result on the network, and can be realized by, for example, a communication IF, a second browsing program, a comparison result browsing Web page, or the like. . Specifically, the obtained comparison result is applied to a predetermined web page template or the like to generate a web page for browsing the comparison result as shown in FIG. Then, by returning the Web page in response to an HTTP request indicating access from the client terminal, the acquired comparison result is browsed on the network.

  In this way, the user can know another content that is considered to be similar to the content introduced by the content introduction server device.

    <Process flow>

  FIG. 14 is a flowchart illustrating an example of a processing flow in the content introduction server device of the recommendation system according to the present exemplary embodiment. The steps shown below may be steps executed by each hardware configuration of the computer as described above, or may be processing steps that constitute a program for controlling the computer recorded on a medium. I do not care.

  As shown in FIG. 14A, first, content introduction information is transmitted to a terminal for browsing via a network (step S1411). Also, when an impression of the content is received via the network (step S1412), the received impression is transmitted to the terminal for browsing on the network (step S1413). In addition, the impression text received together with it is sent as a sentence to the recommendation device (step S1414).

  Then, in the recommendation device, the comparison determination process as described in the first embodiment is executed using the comment sentence transmitted from the content introduction server device, and the comparison result is output.

  Subsequently, as shown in FIG. 14B, the comparison result output from the recommendation device is acquired by the content introduction server device (step S1415), and the HTTP request is made to browse the acquired comparison result on the network. Is transmitted to the terminal in response to the reception (step S1416).

    <Brief description of effect>

  As described above, according to the recommendation system of the present embodiment, the user can know another content that seems to be similar to the content introduced by the content introduction server device.

  Example 4

    <Overview>

  The present embodiment is a recommendation system characterized in that, based on the above-described embodiment, the above-described comparison determination process and the Web page that enables browsing of the comparison determination result regarding the favorite content designated by the user are disclosed. . Therefore, the user can know the recommendation information regarding the content that the user likes.

    <Functional configuration>

  FIG. 15 is a diagram illustrating an example of functional blocks in the recommendation system of the present embodiment. As shown in this figure, the recommendation system according to the present embodiment includes a “recommendation device” (1500) and a “content introduction server device” (1510) as in the second embodiment.

  The functional blocks of the “recommendation device” are the same as those described in the first embodiment and the second embodiment, and thus illustration and description thereof are omitted.

  In addition, the “content introduction server device” also has an “introduction section” (1511), an “impression sentence reception section” (1512), and a “first browsing section” (1513) based on the second embodiment. Have. Further, on the basis of the third embodiment, a “comparison result acquisition unit”, a “second browsing unit”, or the like (not shown) may be included. In addition, since these structural requirements have already been described in the above embodiment, the description thereof will be omitted. In the recommendation system of the present embodiment, the content introduction server device further includes a “favorite content identification reception unit” (1514), a “favorite content comparison result acquisition unit” (1515), and a “third browsing unit” (1516). ).

  The “favorite content specifying unit” (1514) has a function of receiving, via a network, the specification of favorite content in a plurality of contents whose introduction information can be browsed by the introduction unit. Can be realized. Specifically, for example, a “favorite” button or the like is provided on a content introduction Web page transmitted and displayed on the terminal by the operation of the introduction unit described above. When the button is clicked on the terminal, the identification information of the content introduced on the page is acquired by the favorite content specifying unit via the network and specified as a favorite. Further, the identification information of the favorite content specified as described above may be stored in the recording apparatus in association with the identification information of the terminal that has designated the favorite.

  The “favorite content comparison result acquisition unit” (1515) has a function of acquiring, from the comparison result acquisition unit, the result of comparison with other content using the impression text for the favorite content received by the favorite content identification reception unit. For example, it can be realized by a communication IF or a favorite content comparison result acquisition program. The favorite content comparison result acquisition unit has the same basic processing operation as that of the comparison result acquisition unit described in the third embodiment, and thus description of common parts is omitted. The difference between the two is that the comparison result acquired by the favorite content comparison result acquisition unit is a comparison result regarding the specified favorite content. That is, for example, identification information of favorite content that is specified as favorite content and stored in the storage device is transmitted to the recommendation device, and the comparison result of the favorite content that is returned in response is acquired. .

  The “third browsing unit” (1516) has a function of browsing the acquired favorite content comparison result on the network. For example, in a communication IF, a third browsing program, a Web page for browsing the comparison result of the favorite content, etc. Can be realized. Specifically, the second browsing unit of the third embodiment and its basic processing action are also the same. The difference between them is that the comparison result browsing Web page generated for browsing on the network relates to the specified favorite content. Then, in the third browsing unit, when there is an access from the terminal, using the stored identification information of the terminal that has designated the favorite, the comparison result relating to the favorite content designated by the terminal A web page for browsing is returned.

    <Process flow>

  FIG. 16 is a flowchart illustrating an example of a processing flow in the content introduction server device of the recommendation system according to the present exemplary embodiment. The steps shown below may be steps executed by each hardware configuration of the computer as described above, or may be processing steps that constitute a program for controlling the computer recorded on a medium. I do not care.

  As shown in FIG. 16A, first, when specification of favorite content in a plurality of contents is received via the network (step S1611), an impression sentence related to the received favorite content is acquired (step S1612). And the acquired comment sentence is transmitted as a sentence with respect to a recommendation apparatus (step S1613).

  Then, in the recommendation device, the comparison determination process as described in the first embodiment is executed using the comment text of the favorite content transmitted from the content introduction server device, and the comparison result is output.

  Subsequently, as shown in FIG. 16B, the comparison result of the favorite content output from the recommendation device is acquired by the content introduction server device (step S1614), and the comparison result of the acquired favorite content is obtained on the network. In order to make it browse, it transmits to a terminal according to reception of an HTTP request, etc. (step S1615).

    <Brief description of effect>

  As described above, according to the recommendation system of the present embodiment, the user can know the content of the recommendation regarding the content designated by the terminal as being liked by the user.

  Example 5

    <Overview>

  In the present embodiment, based on the above-described embodiment, for example, as shown in FIG. 17, the weighted result product integrated value “4274” points (α) and “ The recommendation system is characterized by displaying the “3976” point (β) or the like as an index value of the comparison result in accordance with the content of the recommendation.

    <Functional configuration>

  FIG. 18 is a diagram illustrating an example of functional blocks in the recommendation system of the present embodiment. As shown in this figure, the recommendation system according to the present embodiment includes a “recommendation device” (1800) and a “content introduction server device” (1810) as in the fourth embodiment.

  The functional blocks of the “recommendation device” are the same as those described in the first embodiment and the second embodiment, and thus illustration and description thereof are omitted.

  In addition, the “content introduction server device” is also based on the above-described fourth embodiment, and includes an “introduction part” (1811), an “impression sentence reception part” (1812), a “first browsing part” (1813), “ It has a “favorite content identification reception unit” (1814), a “favorite content comparison result acquisition unit” (1815), and a “third browsing unit” (1816). Moreover, you may have a "comparison result acquisition part", a "second browsing part", etc. which are not illustrated like Example 3. FIG. In addition, since these structural requirements have already been described in the above embodiment, the description thereof will be omitted.

  The recommendation system according to the present embodiment is characterized in that the third browsing unit of the content introduction server device further includes “indexed browsing means” (1817).

  The “indexed browsing means” (1817) has a function of browsing the favorite content comparison result in which the indexed calculation result is associated with the content as the favorite content comparison result. For example, the communication IF, the indexed browsing program, the favorite content This can be realized by a web page for browsing the comparison results. Specifically, the integrated value of the weighted result product of the extracted words of the sentence A and the sentence B as described in the first embodiment is acquired as an index value indicating a comparison result such as the similarity between the sentence A and the sentence B. . Then, the comparison result browsing Web page as shown in FIG. 17 is generated by matching the value with a predetermined Web page template or the like. Then, the indexed browsing means uses the stored identification information of the terminal that has designated the favorite when accessed from the terminal, and relates to the favorite content designated by the terminal. For example, a Web page for browsing the comparison result indicating the index value is returned.

    <Process flow>

  FIG. 19 is a flowchart illustrating an example of the flow of processing in the recommendation system of the present embodiment. The steps shown below may be steps executed by each hardware configuration of the computer as described above, or may be processing steps that constitute a program for controlling the computer recorded on a medium. I do not care.

  As shown in FIG. 19A, first, when specification of favorite content in a plurality of contents is received via the network (step S1911), an impression sentence regarding the received favorite content is acquired (step S1912). And the acquired comment sentence is transmitted as a sentence with respect to a recommendation apparatus (step S1913).

  Then, in the recommendation device, the comparison determination process as described in the first embodiment is performed using the comment text of the favorite content transmitted from the content introduction server device, and an index value indicating the comparison result is output.

  Subsequently, as shown in FIG. 19B, an index value indicating the comparison result of the favorite content output from the recommendation device is acquired by the content introduction server device (step S1914), and the comparison result of the acquired favorite content is obtained. Is transmitted to the terminal in response to reception of an HTTP request or the like (step S1915).

    <Brief description of effect>

  As described above, by the recommendation system of this embodiment, the integrated value of the weighted result product that is the basis for the recommendation determination can be presented to the user. Therefore, the user can make a more objective judgment on the recommended result.

The conceptual diagram for demonstrating an example of the recommendation by the recommendation apparatus of Example 1. FIG. The figure showing an example of the functional block in the recommendation apparatus of Example 1. The figure showing an example of the weighting result accumulate | stored in the storage part of the recommendation apparatus of Example 1. The conceptual diagram showing an example of the comparison process using the calculation in the comparison part of the recommendation apparatus of Example 1, and a calculation result The figure showing an example of the hardware constitutions in the recommendation apparatus of Example 1. The flowchart showing an example of the flow of the process in the recommendation apparatus of Example 1. The figure for demonstrating the feature point of the recommendation system of Example 2. The figure showing an example of the functional block in the recommendation system of Example 2. The figure showing an example of the web page for making an impression sentence browse on a network in the content introduction server apparatus of the recommendation system of Example 2. The figure showing an example of the hardware constitutions in the contents introduction server apparatus of the recommendation system of Example 2. The flowchart showing an example of the flow of the process in the content introduction server apparatus of the recommendation system of Example 2. The figure showing an example of the recommendation screen of the content by the recommendation system of Example 3. The figure showing an example of the functional block in the recommendation system of Example 3. The flowchart showing an example of the flow of a process in the content introduction server apparatus of the recommendation system of Example 3. The figure showing an example of the functional block in the recommendation system of Example 4. The flowchart showing an example of the flow of a process in the content introduction server apparatus of the recommendation system of Example 4. The figure showing an example of the recommendation screen with the index value of the content by the recommendation system of Example 5. The figure showing an example of the functional block in the recommendation system of Example 5. The flowchart showing an example of the flow of the process in the content introduction server apparatus of the recommendation system of Example 5. The figure showing an example of the recommendation screen of the movie by a prior art, and the recommendation screen of the movie by this invention

Explanation of symbols

0200 Recommendation Device 0201 Sentence Acquisition Unit 0202 Extraction Unit 0203 Weighting Unit 0204 Accumulation Unit 0205 Comparison Unit 0206 Computation Expression Storage Unit 0207 Output Unit 0808 Impression Sentence Acquisition Unit 0810 Content Introduction Server Device 0811 Introduction Unit 0812 Impression Sentence Acceptance Unit 0813 First Browsing Unit

Claims (12)

A sentence acquisition unit for acquiring sentences;
An extraction unit that extracts words by morphologically analyzing the acquired sentences and associating them with the sentences;
A weighting unit that weights the extracted word by the appearance frequency in the sentence,
An accumulator that accumulates weighted results in association with sentences;
An arithmetic expression storage unit storing arithmetic expressions for comparing sentences using weighted result products for each word;
A comparison unit that compares sentences using a weighted result of the storage unit and an arithmetic expression of the arithmetic expression storage unit;
An output unit for outputting a comparison result;
A recommendation device having: Introducing content introduction information via the network,
Impression text accepting section for accepting impressions about the content via the network,
A first browsing section for browsing received impressions on the network;
A content introduction server device comprising:
The recommendation device according to claim 1, wherein the sentence acquisition unit includes an impression sentence acquisition unit that acquires an impression sentence received by an impression sentence reception unit of the content introduction server device, as a sentence.
A recommendation system consisting of The content introduction server device
A comparison result acquisition unit for acquiring a comparison result output from the recommendation device;
A second browsing unit for browsing the acquired comparison results on the network;
The recommendation system according to claim 2. The content introduction server device
A favorite content specification receiving unit that receives specification of favorite content in a plurality of contents whose introduction information can be browsed by the introduction unit via a network;
A favorite content comparison result acquisition unit that acquires a result of comparison with other content using an impression sentence for the favorite content received by the favorite content identification reception unit;
A third browsing unit for browsing the acquired favorite content comparison results on the network;
The recommendation system according to claim 2 or 3, further comprising:   The recommendation system according to claim 4, wherein the third browsing unit includes indexed browsing means for browsing a favorite content comparison result obtained by associating the indexed calculation result with the content as a favorite content comparison result. The recommendation system according to any one of claims 2 to 5, wherein the expression stored in the arithmetic expression storage unit is an expression expressed by the following equation (1).
A sentence acquisition step for acquiring a sentence;
An extraction step of extracting words by morphologically analyzing the acquired sentences and associating them with the sentences;
A weighting step for weighting the extracted word by the appearance frequency in the sentence;
A recording step for recording the weighted result in the storage unit in association with the sentence;
An arithmetic expression acquisition step for acquiring an arithmetic expression for comparing sentences using a weighted result product for each word;
A comparison step for comparing sentences using the weighted result of the storage unit and the obtained arithmetic expression;
An output step for outputting a comparison result;
A method for controlling a recommendation device that causes a computer to execute the process. A method for controlling a recommendation system comprising a content introduction server device and a recommendation device,
An introduction step for browsing content introduction information via the network on the content introduction server device,
Impression sentence reception step for accepting impressions about the content via the network;
A first browsing step for browsing received impressions on the network;
To the computer,
The computer further executes an impression sentence acquisition step of acquiring the impression sentence received in the impression sentence reception step in the content introduction server device in the sentence acquisition step of the recommendation device control method according to claim 7. Recommendation system control method. In the content introduction server device,
A comparison result acquisition step for acquiring a comparison result output from the recommendation device;
A second browsing step for browsing the acquired comparison result on the network;
The method of controlling a recommendation system according to claim 8, wherein the computer is executed. In the content introduction server device,
A favorite content specification accepting step for accepting specification of favorite content in a plurality of contents whose introduction information can be viewed in the introduction step via a network;
A favorite content comparison result acquisition step of acquiring a result of comparison with other content using a comment on the favorite content received in the favorite content identification reception step, from a comparison result acquisition step;
A third browsing step for browsing the acquired favorite content comparison result on the network;
10. The method of controlling a recommendation system according to claim 8, further comprising:   The control method for the recommendation system according to claim 10, wherein the third browsing step further causes the computer to execute an indexed browsing step for browsing the favorite content comparison result in which the indexed calculation result is associated with the content as the favorite content comparison result. . The method of controlling a recommendation system according to any one of claims 8 to 11, wherein the expression acquired in the operation expression acquiring step is an expression expressed by the following equation (1).