JP6136794B2 - Information processing method, program, and information processing apparatus - Google Patents

Description

  The present invention relates to an information processing method, a program, and an information processing apparatus using a computer.

  Conventionally, there has been proposed a relationship extraction device that calculates a plurality of types of similarity for an arbitrary word pair in a text and generates a feature vector having each similarity as an element (see, for example, Patent Document 1).

JP 2011-118526 A

  However, the conventional technique has a problem that the amount of calculation at the time of information analysis is enormous.

  In one aspect, an object of the present invention is to provide an information processing method and the like that can reduce the amount of calculation based on terms having a large amount of calculation and the number of terms.

  The information processing method disclosed in the present application refers to a category associated with a user in an information processing method using a computer, calculates the number of combinations between categories, and based on the calculated number of combinations between categories, The first similarity is calculated, the combination of categories having the first similarity exceeding the threshold is extracted, and the second similarity between the extracted categories is calculated based on the terms and the number of terms corresponding to each extracted category To do.

  In one aspect, the amount of calculation can be reduced.

It is explanatory drawing which shows the outline | summary of an information processing system. It is a block diagram which shows the hardware group of a computer. It is explanatory drawing which shows the example of a blog article. It is explanatory drawing which shows the record layout of a category list file. It is explanatory drawing which shows the record layout of a user blog article list file. It is explanatory drawing which shows the record layout of a user category list file. It is explanatory drawing which shows the record layout of a user article list file. It is explanatory drawing which shows the record layout of a category word list file. It is explanatory drawing which shows the record layout of a category article list file. It is explanatory drawing which shows a category co-occurrence file. It is explanatory drawing which shows the record layout of a category co-occurrence similarity file. It is explanatory drawing which shows a content similarity file. It is explanatory drawing which shows the record layout of the category article list file after deletion. It is a flowchart which shows the production | generation procedure of a category article list file. It is a flowchart which shows the production | generation procedure of a category co-occurrence file. It is a flowchart which shows the procedure of a 1st similarity calculation process. It is a flowchart which shows the procedure of a 2nd similarity calculation process. It is a flowchart which shows the procedure of a 2nd similarity calculation process. It is a flowchart which shows the procedure of a deletion process. It is a functional block diagram which shows operation | movement of the computer of the form mentioned above. FIG. 6 is a block diagram illustrating a hardware group of a computer according to a second embodiment.

Embodiment 1
Hereinafter, embodiments will be described with reference to the drawings. FIG. 1 is an explanatory diagram showing an outline of an information processing system. The information processing system includes an information processing apparatus 1, a server computer 2, and the like. The information processing apparatus 1 and the server computer 2 are connected via a communication network N such as the Internet. The information processing apparatus 1 is, for example, a personal computer, a server computer, a mobile phone, a PDA (Personal Digital Assistant) or the like. Hereinafter, the information processing apparatus 1 will be described as being replaced with the computer 1. The server computer 2 stores articles including a plurality of words, lyrics, tweets, product descriptions, store descriptions, or articles such as papers, and information with categories for the sentences. The server computer 2 receives sentences and categories from another computer (not shown) and stores the received sentences and categories.

  In the following, an example in which a plurality of users attach categories to blog articles and upload them to the server computer 2 will be described as an example. The computer 1 periodically accesses the server computer 2 and downloads each user's blog articles and categories. The computer 1 performs information processing on the downloaded article and category. Details will be described below.

  FIG. 2 is a block diagram showing a hardware group of the computer 1. The computer 1 includes a central processing unit (CPU) 11 as a control unit, a random access memory (RAM) 12, an input unit 13, a display unit 14, a storage unit 15, a communication unit 16, and the like. The CPU 11 is connected to each part of the hardware via the bus 17. The CPU 11 controls each part of the hardware according to the control program 15P stored in the storage unit 15. The RAM 12 is, for example, SRAM (Static RAM), DRAM (Dynamic RAM), flash memory, or the like. The RAM 12 also functions as a storage unit, and temporarily stores various data generated when the CPU 11 executes various programs.

  The input unit 13 is an input device such as a mouse or a keyboard, a mouse or a touch panel, and outputs received operation information to the CPU 11. The display unit 14 is a liquid crystal display, an organic EL (electroluminescence) display, or the like, and displays various information according to instructions from the CPU 11. The communication unit 16 is a communication module, and transmits and receives information to and from the server computer 2 via the communication network N.

  The storage unit 15 is a hard disk or a large-capacity memory, and stores a category list file 151, a user blog article list file 152, a user category list file 153, a user article list file 154, in addition to the control program 15P described above. In addition, the storage unit 15 stores a category phrase list file 155, a category article list file 156, a category co-occurrence file 157, a category co-occurrence similarity file 158, a content similarity file 159, and the like. In the present embodiment, an example in which the category list file 151 and the like are stored in the storage unit 15 of the computer 1 has been described. For example, the various files described above may be appropriately stored in a database server (not shown) connected via the communication network N. In this case, the CPU 11 accesses the database server as necessary, and writes and reads data.

  FIG. 3 is an explanatory diagram showing an example of a blog article. The example of FIG. 3 is a blog article of user A, and four blog articles are described. A001 is identification information (hereinafter referred to as article ID) for specifying the user's blog article. Each blog post has a category. For example, “Gourmet” category is assigned to “I ate ramen” with article ID A001, and “Fashion” category is assigned to “I bought clothes in Shibuya” with article ID A002. Yes. In the present embodiment, information for identifying the user is also attached to the article ID. For example, the article ID “A001” is the article of the user A, and the article ID “B001” is the article of the user B.

  FIG. 4 is an explanatory diagram showing a record layout of the category list file 151. As shown in FIG. 4, the category list file 151 stores categories to be analyzed in advance. In the example of FIG. 4, a total of 10 categories such as gourmet, alcohol, and fashion are stored. In this embodiment, an example in which ten categories are analyzed will be described. However, this is merely an example, and the number of categories and the contents of the categories are not limited.

  FIG. 5 is an explanatory diagram showing a record layout of the user blog article list file 152. The user blog article list file 152 includes a user field, an article ID field, a category field, a body field, and the like. User information is stored in the user field. In the article ID field, an article ID for specifying the user's blog article is stored. When the blog article is downloaded from the server computer 2, the CPU 11 newly generates an article ID and stores the user, article ID, category, and text in the user blog article list file 152.

  The category field stores a category assigned in association with the article ID. In the body field, the body of the blog article is stored in text format in association with the article ID. In the example of FIG. 5, “I ate ramen” whose user ID is A001 is stored in the category “Gourmet”, and “I drank beer” whose user ID is B001 is “ It is remembered in “Sake”. In the present embodiment, information of the blog article text is stored, but the title of the blog article may be stored and analyzed.

  FIG. 6 is an explanatory diagram showing a record layout of the user category list file 153. The user category list file 153 includes user fields and category fields. In the category field, a category is stored in association with the user. The CPU 11 refers to the user blog article list file 152 and extracts the category of each user. The CPU 11 stores the extracted category of each user in the user category list file 153. In the example of FIG. 6, the category of the user A is gourmet, fashion, and liquor. The category of user B is only alcohol.

  FIG. 7 is an explanatory diagram showing a record layout of the user article list file 154. The user article list file 154 includes a user field, an article ID field, a category field, a body text field, and a phrase field. In the phrase field, a phrase (term) extracted from the blog article body is stored. CPU11 performs well-known morphological analysis with respect to the content of a text, and extracts a phrase. The CPU 11 stores the extracted word / phrase in association with the text. In the example of FIG. 7, the phrases “ramen” and “eat” are extracted for “eating ramen”. In the present embodiment, an example using morphological analysis is shown, but a phrase may be extracted by another method. For example, a dictionary may be prepared in the storage unit in advance, and words in the text that match the words in the dictionary may be extracted.

  FIG. 8 is an explanatory diagram showing a record layout of the category word / phrase list file 155. The category phrase list file 155 includes a category field, a phrase x appearance frequency (number of terms) field, and the like. The word x appearance frequency field stores words that belong to a category in association with the category and the appearance frequency of the word in the blog article body. The CPU 11 refers to the category field and the phrase field of the user article list file 154, extracts the phrase used in the category, and counts the appearance frequency of the phrase in the category. In the example of FIG. 8, “Ramen” has an appearance frequency of 300 and “Eat” has an appearance frequency of 900 for the category “Gourmet”. In addition, regarding the category “alcohol”, “eating” has an appearance frequency of “300”. Thus, there are overlapping phrases between categories such as “drink” and “eat”.

  FIG. 9 is an explanatory diagram showing a record layout of the category article list file 156. As shown in FIG. The category article list file 156 includes a category field, an article list field, and the like. In the article list field, the article ID of each user is stored in association with the category. The CPU 11 refers to the user article list file 154 and extracts the article ID of each user belonging to the category. The CPU 11 stores the user name and the extracted article ID in the category article list file 156 in association with the category. In the example of FIG. 9, two articles specified by the article IDs “A001” and “A004” of the user A for the category “gourmet” are stored. In addition, two articles specified by the user A's article IDs “A002” and “A003” are stored for the category “fashion”. In the present embodiment, the article ID is stored as an example. However, when the information for specifying the user is not described in the article ID, the article ID may be stored together with the user information.

  FIG. 10 is an explanatory diagram showing the category co-occurrence file 157. The category co-occurrence file 157 stores the number of co-occurrence between two categories. That is, it is determined that one user co-occurs when two categories of articles are described. The CPU 11 refers to the user category list file 153 and reads the category of a specific user. When the CPU 11 detects a plurality of categories, the CPU 11 increments the number of co-occurrence for all combinations of categories. The CPU 11 counts the combinations of categories for all other users and stores them in the category co-occurrence file 157. In the example of FIGS. 6 and 10, the user A has three categories of “gourmet”, “fashion”, and “alcohol”. The CPU 11 detects the combination of “gourmet” and “fashion” and increments the number of combinations of “gourmet” and “fashion” in the category co-occurrence file 157.

  Similarly, the CPU 11 detects a combination of “gourmet” and “alcohol” and increments the number of combinations of “gourmet” and “alcohol” in the category co-occurrence file 157. Further, the CPU 11 detects the combination of “fashion” and “alcohol”, and increments the number of combinations of “fashion” and “alcohol” in the category co-occurrence file 157. Subsequently, the CPU 11 reads the category list of the user B from the user category list file 153. Since user B has only the category “alcohol”, the number of co-occurrence is zero. The above processing is performed for all users to be analyzed.

  In the example of FIG. 10, 30 users refer to both “gourmet” and “alcohol” categories, and 100 users refer to both “gourmet” and “fashion” categories. A combination of “gourmet” and “alcohol”, and a combination that differs only in the order of categories such as “alcohol” and “gourmet” are the same combination. In this embodiment, the category list file 153 is referred to as an example, but the category co-occurrence file 157 may be generated with reference to the user article list file 154.

  FIG. 11 is an explanatory diagram showing a record layout of the category co-occurrence similarity file 158. The category co-occurrence similarity file 158 stores the first similarity between two categories. For example, the first similarity between the categories “gourmet” and “alcohol” is 0.99. Here, a procedure for calculating the first similarity by setting the first category as “gourmet” and the second category as “alcohol” will be described. The CPU 11 refers to the category co-occurrence similarity file 158 and reads the number of combinations between “gourmet” as the first category and other categories other than “alcohol” as the second category. In the example of FIG. 10, fashion is 100, car is 12, music is 70, movie is 45, animation is 40, game is 25, baseball is 15, and soccer is 15.

  Similarly, the CPU 11 refers to the category co-occurrence similarity file 158 and determines the number of combinations between “alcohol” as the second category and other categories other than “gourmet” as the first category. read out. In the example of FIG. 10, fashion is 45, car is 8, music is 50, movie is 20, animation is 24, game is 20, baseball is 10, and soccer is 5. The CPU 11 sets the vector of the first category “gourmet” to {100, 12, 70, 45, 40, 25, 15, 15}. Further, the CPU 11 sets the vector of the second category “alcohol” to {45, 8, 50, 20, 24, 20, 10, 5}. The column direction of the vector is the other category {fashion, car, music, movie, animation, game, baseball, soccer} for both the first vector and the second vector. In the present embodiment, an example is given in which a vector is obtained based on the number of combinations between the first category and other categories other than the second category, but the present invention is not limited to this. The vector may be obtained based on the number of combinations between the first category and other categories including the second category. Further, it is not always necessary to use the number of combinations of all other categories for other categories, and the number of combinations of some other categories may be used.

  The CPU 11 calculates the first similarity from the first category vector and the second category vector. The first similarity may be calculated using a cosine similarity, an inner product, a correlation function, or the like. For example, the first similarity may be calculated based on the sum of squares of the differences between the vectors. In this embodiment, an example using cosine similarity will be described as an example. The cosine similarity is calculated by the following formula 1 stored in the storage unit 15.

（式１）

(Formula 1)

  V is the number of columns, and is 8 in the above-described example. The CPU 11 substitutes the vector of the first category and the vector of the second category into Equation 1, and calculates the first similarity 0.99. The CPU 11 performs the same process for other combinations, calculates the first similarity for each category combination, and stores the calculated first similarity in the category co-occurrence similarity file 158. For example, the first similarity between “gourmet” and “fashion” is as low as 0.57. Note that the numerical values shown in FIG. 11 are appropriate values for ease of explanation.

  Next, the CPU 11 reads the first threshold value stored in the storage unit 15. The CPU 11 refers to the category co-occurrence similarity file 158 and extracts a combination of categories having a first similarity exceeding the first threshold. The first threshold is whether the user sets an appropriate value from the input unit 13. In this embodiment, the first threshold value is 0.74 as an example. The CPU 11 extracts four combinations of “gourmet” and “fashion”, “music” and “movie”, “anime” and “game”, “baseball” and “soccer” as combinations exceeding the first threshold.

  FIG. 12 is an explanatory diagram showing the content similarity file 159. The CPU 11 calculates a content similarity (hereinafter referred to as a second similarity) for the extracted category combination. The CPU 11 stores the second similarity calculated in association with the category combination in the content similarity file 159. The procedure for calculating the second similarity is as follows. The CPU 11 reads the extracted words and appearance frequencies of one category from the category word / phrase list file 155. For example, in the category “Gourmet”, “Ramen” is 300 times, “Eat” is 900 times, “Delicious” is 600 times, “Wine” is 200 times, “Drink” is 60 times, “Beer” is 45 times, Is read out. The CPU 11 generates a vector of one category. In the above example, the vector of one category is {300, 900, 600, 200, 60, 45. The column direction of the vector is determined in advance and is {ramen, eat, delicious, wine, drink, beer ...}.

  The CPU 11 reads the words and appearance frequencies of other categories extracted in the same manner from the category word / phrase list file 155. For example, in the category “Sake”, “Beer” 320 times, “Drink” 600 times, “Eat” 300 times, “Delicious” 400 times, “Wine” 280 times, “Ramen” 80 times Read as CPU11 produces | generates the vector of another category. In the above example, the vectors of other categories are {320, 600, 300, 400, 280, 80. The CPU 11 substitutes the vector of one category “gourmet” and the vector of the other category “sake” into Equation 1, and calculates the second similarity based on the words and appearance frequencies of the first category and the second category. To do. In addition, although the example which uses Formula 1 gives the calculation of the 2nd similarity degree, you may calculate by another system.

  In the example of FIG. 12, the second similarity between “gourmet” and “alcohol” is 0.8. The second similarity between “music” and “movie” is 0.3, the second similarity between “anime” and “game” is 0.7, and the second similarity between “baseball” and “soccer” is 0.2. Subsequently, the CPU 11 reads the second threshold value from the storage unit 15. In the present embodiment, the second threshold value is set to the same value as the first threshold value, but may be another value. CPU11 extracts the combination of the category which has the 2nd similarity exceeding a 2nd threshold value. In the example of FIG. 12, two combinations of “gourmet” and “alcohol” and “animation” and “game” are extracted.

  The CPU 11 refers to the category article list file 156 and performs processing for deleting user information that does not match between the extracted category combinations. FIG. 13 is an explanatory diagram showing the record layout of the category article list file 156 after deletion. The CPU 11 reads the extracted article ID of the user in one category. Moreover, CPU11 reads the article ID of the user of the other category extracted. The CPU 11 refers to the user information attached to the article ID, and deletes the article ID of the user whose user does not match between the categories. In the example of FIG. 9, the user H who describes the article “gourmet” does not describe the article “alcohol”, so the article IDs (H001, H005) related to the article of the user H are deleted. Similarly, the article of user B is also deleted from the category “alcohol”. On the other hand, since the articles of user A and user C are both stored in both the categories “gourmet” and “sake”, they are not deleted. The CPU 11 performs the same processing for another set of combinations “animation” and “game”. The articles of users E and H are not deleted, and the articles of users F and H in the category “game” are deleted.

  Various processing contents in the above hardware group will be described with reference to flowcharts. FIG. 14 is a flowchart showing a procedure for generating the category article list file 156. The CPU 11 downloads a blog article from the server computer 2 (step S141). CPU11 produces | generates article ID, when a new blog article exists (step S142). CPU11 extracts a user name, a category, and a text from a blog article. The CPU 11 stores the user name, article ID, category, and text in the user blog article list file 152 (step S143). The CPU 11 extracts the category of each user and stores the category extracted for each user, thereby generating the category word / phrase list file 155 (step S144).

  The CPU 11 refers to the user blog article list file 152 and extracts words / phrases in the text by morphological analysis (step S145). The CPU 11 stores the word / phrase extracted in association with the article ID in the user article list file 154 (step S146). The CPU 11 refers to the user article list file 154, extracts words for each category, counts the appearance frequency of each word, and stores the words and appearance frequency in the category word list file 155 (step S147). The CPU 11 refers to the user blog article list file 152 and stores the article ID of each user in the category article list file 156 for each category (step S148).

  FIG. 15 is a flowchart showing a procedure for generating the category co-occurrence file 157. The CPU 11 reads out a category corresponding to the user from the user category list file 153 (step S151). The CPU 11 determines whether there are a plurality of read categories (step S152). If the CPU 11 determines that there are a plurality (YES in step S152), the process proceeds to step S153. CPU11 extracts the combination of a some category (step S153). In the example of user A in FIG. 6, three combinations of “gourmet” and “fashion”, “gourmet” and “alcohol”, and “fashion” and “alcohol” are extracted.

  The CPU 11 increments the number of corresponding category combinations in the category co-occurrence file 157 for the extracted category combinations (step S154). In the example of user A in FIG. 6, the number of category co-occurrence files 157 corresponding to “gourmet” and “fashion”, “gourmet” and “alcohol”, and “fashion” and “alcohol” are respectively incremented. The If the CPU 11 determines that there are not a plurality of categories (NO in step S152), the CPU 11 skips steps S153 and S154 and shifts the processing to step S155.

  The CPU 11 determines whether or not the processing described above has been completed for all users (step S155). If the CPU 11 determines that the process has not ended (NO in step S155), the process proceeds to step S156. The CPU 11 reads out other user categories from the user category list file 153 (step S156). Thereafter, the process returns to step S152. By repeating the above processing, the category co-occurrence file 157 based on the combination of categories of each user is completed. If the CPU 11 determines that the processing has been completed for all users (YES in step S155), the series of processing ends.

  FIG. 16 is a flowchart showing the procedure of the first similarity calculation process. CPU11 extracts the 1st category and the 2nd category used as the object which computes the 1st similarity (Step S161). The CPU 11 refers to the category co-occurrence file 157 and reads the number of combinations between the first category and other categories other than the second category (step S162). The CPU 11 generates a first vector based on the read number of combinations (step S163). The CPU 11 refers to the category co-occurrence file 157 and reads the number of combinations between the second category and other categories other than the first category (step S164). The CPU 11 generates a second vector based on the read combination number (step S165).

  CPU11 reads Formula 1 memorize | stored in the memory | storage part 15 (step S166). The CPU 11 substitutes the first vector and the second vector into Equation 1 to calculate the first similarity (step S167). The CPU 11 stores the calculated first similarity in association with the first category and the second category in the category co-occurrence similarity file 158 (step S168). The CPU 11 determines whether or not the processing has been completed for all the combinations of categories stored in the category co-occurrence file 157 (step S169). If the CPU 11 determines that the process has not been completed for all combinations of categories (NO in step S169), the process proceeds to step S1610.

  CPU11 extracts the combination of the other 1st category and 2nd category different from step S161 (step S1610). Thereafter, the process returns to step S162. By repeating the above processing, the first similarity can be calculated for all combinations of categories. If the CPU 11 determines that the process has been completed for all the combinations of categories (YES in step S169), the series of processes is terminated.

  17 and 18 are flowcharts showing the procedure of the second similarity calculation process. The CPU 11 reads the first threshold value stored in the storage unit 15 (step S171). The CPU 11 refers to the category co-occurrence similarity file 158 and extracts a combination of categories having the first similarity exceeding the first threshold (step S172). The CPU 11 performs the following processing for the extracted category combination. The CPU 11 refers to the category word / phrase list file 155 and reads the words and appearance frequencies of one category (step S173). The CPU 11 rearranges the appearance frequencies in the order of words (step S174). This word order may be determined in advance, for example, in the order of the Japanese syllabary. CPU11 produces | generates the vector of one category which uses the appearance frequency of a phrase as a column value (step S175).

  The CPU 11 refers to the category word / phrase list file 155 and reads the words / occurrence frequencies of other categories paired with one category (step S176). The CPU 11 rearranges the appearance frequencies in the order of words (step S177). CPU11 produces | generates the vector of the other category which makes the appearance frequency of a phrase a column value (step S178). CPU11 reads Formula 1 from the memory | storage part 15 (step S179). CPU11 substitutes the vector of one category and the vector of another category to Formula 1, and calculates the 2nd similarity (Step S181).

The CPU 11 stores the calculated second similarity in the content similarity file 159 in association with one category and another category (step S182). The CPU 11 determines whether or not the processing has been completed for all the combinations of categories extracted in step S172 (step S183). If the CPU 11 determines that the process has not ended (NO in step S183), the process proceeds to step S184. The CPU 11 extracts one category and another category that are another combination (step S184). Thereafter, the process returns to step S173. By repeating the above processing, the second similarity can be calculated for all the combinations of categories having the first similarity exceeding the first threshold. If the CPU 11 determines that the process has been completed for all the combinations of categories (YES in step S183), the series of processes is terminated.
In the present embodiment, the amount of calculation of the second similarity is reduced by narrowing down the combinations for calculating the second similarity by the first similarity having a smaller amount of calculation than the second similarity by the above processing. . As a result, in this embodiment, it is possible to obtain a category in which the model is likely to fluctuate with other categories within a practical time.

  FIG. 19 is a flowchart showing the procedure of the deletion process. CPU11 reads the 2nd threshold value which is remembered in memory section 15 (step S191). The CPU 11 refers to the content similarity file 159 and extracts a combination of categories having the second similarity exceeding the second threshold (step S192). The CPU 11 refers to the category article list file 156 and reads the article ID of one category (step S193). CPU11 specifies the user corresponding to article ID (step S194). Specifically, the user identification information given to the article ID is referred to, or the user corresponding to the article ID is specified with reference to the user article list file 154.

  The CPU 11 refers to the category article list file 156 and reads the article ID of another category corresponding to the one category extracted in step S192 (step S195). CPU11 specifies the user corresponding to article ID (step S196). The CPU 11 deletes the article ID that does not match the identified user from the category article list file 156 (step S197). In other words, the CPU 11 stores only the article ID that the identified user has in common between categories. The CPU 11 determines whether or not all combinations of categories extracted in step S192 have been processed (step S198).

  If the CPU 11 determines that all combinations of categories have not been processed (NO in step S198), the process proceeds to step S199. The CPU 11 selects another category combination (step S199). Thereafter, the process proceeds to step S193. By repeating the above processing, the deletion processing ends for all the combinations of categories having the second similarity exceeding the second threshold. If the CPU 11 determines that all combinations of categories have been processed (YES in step S198), the series of processing ends. As described above, it is possible to significantly reduce the amount of calculation by previously narrowing down combinations of categories having high first similarity between categories.

Embodiment 2
FIG. 20 is a functional block diagram showing the operation of the computer 1 of the above-described form. When the CPU 11 executes the control program 15P, the computer 1 operates as follows. The calculation unit 201 refers to the category associated with the user and calculates the number of combinations between categories. The first similarity calculation unit 202 calculates the first similarity between categories based on the calculated number of combinations between categories. The extraction unit 203 extracts a combination of categories having a first similarity exceeding a threshold value. The second similarity calculation unit 204 calculates a second similarity between the extracted categories based on the terms corresponding to each extracted category and the number of terms.

  FIG. 21 is a block diagram illustrating a hardware group of the computer 1 according to the second embodiment. A program for operating the computer 1 reads a portable recording medium 1A such as a CD-ROM, a DVD (Digital Versatile Disc) disk, a memory card, or a USB (Universal Serial Bus) memory into a reading unit 10A such as a disk drive. It may be stored in the storage unit 15. Further, a semiconductor memory 1B such as a flash memory storing the program may be mounted in the computer 1. Further, the program can be downloaded from another server computer (not shown) connected via a communication network N such as the Internet. The contents will be described below.

  The computer 1 shown in FIG. 21 reads a program for executing the above-described various software processes from the portable recording medium 1A or the semiconductor memory 1B or downloads it from another server computer (not shown) via the communication network N. To do. The program is installed as the control program 15P, loaded into the RAM 12, and executed. Thereby, it functions as the computer 1 described above.

  The second embodiment is as described above, and the other parts are the same as those of the first embodiment. Therefore, the corresponding parts are denoted by the same reference numerals, and detailed description thereof is omitted.

  With respect to the embodiments including the first and second embodiments, the following additional notes are disclosed.

(Appendix 1)
In an information processing method using a computer,
Refer to the category associated with the user, calculate the number of combinations between categories,
Based on the calculated number of combinations between categories, calculate the first similarity between categories,
Extracting a combination of categories having a first similarity exceeding a threshold;
An information processing method for calculating a second similarity between extracted categories based on terms and the number of terms corresponding to each extracted category.

(Appendix 2)
The information processing method according to claim 1, wherein the number of combinations between categories of a plurality of users is calculated by referring to a storage unit that stores categories in association with each user and counting combinations between categories for each user.

(Appendix 3)
The calculated number of combinations between the first category and other categories other than the second category, and the calculated number of combinations between the second category and other categories other than the first category; The information processing method according to attachment 1 or 2, wherein a first similarity between the first category and the second category is calculated based on

(Appendix 4)
Refer to the storage unit storing the term and the number of terms in association with the category, and read out the term and the number of terms in one extracted category,
Refer to the storage unit, read the terms and the number of terms in other categories,
The second similarity between the one category and the other category is calculated based on the read terms and the number of terms in the one category and the terms and the number of terms in the other category. The information processing method as described in any one.

(Appendix 5)
Extracting a combination of categories having a second similarity exceeding a threshold;
The information processing method according to any one of supplementary notes 1 to 4, wherein information relating to a user that does not match between the extracted categories is deleted with reference to a storage unit that stores information relating to the user in association with the category.

(Appendix 6)
On the computer,
Refer to the category associated with the user, calculate the number of combinations between categories,
Based on the calculated number of combinations between categories, calculate the first similarity between categories,
Extracting a combination of categories having a first similarity exceeding a threshold;
A program that executes a process of calculating a second similarity between extracted categories based on terms and the number of terms corresponding to each extracted category.

(Appendix 7)
A calculation unit that refers to the category associated with the user and calculates the number of combinations between categories;
A first similarity calculator that calculates a first similarity between categories based on the calculated number of combinations between categories;
An extraction unit for extracting a combination of categories having a first similarity exceeding a threshold;
An information processing apparatus comprising: a second similarity calculation unit that calculates a second similarity between extracted categories based on terms and the number of terms corresponding to each extracted category.

DESCRIPTION OF SYMBOLS 1 Computer 1A Portable recording medium 1B Semiconductor memory 2 Server computer 10A Reading part 11 CPU
12 RAM
13 Input unit 14 Display unit 15 Storage unit 15P Control program 16 Communication unit 151 Category list file 152 User blog article list file 153 User category list file 154 User article list file 155 Category phrase list file 156 Category article list file 157 Category co-occurrence file 158 Category co-occurrence similarity file 159 Content similarity file 201 Calculation unit 202 First similarity calculation unit 203 Extraction unit 204 Second similarity calculation unit N Communication network

Claims (5)

In an information processing method using a computer,
Refer to the category associated with the user, calculate the number of combinations between categories,
Based on the calculated number of combinations between categories, calculate the first similarity between categories,
Extracting a combination of categories having a first similarity exceeding a threshold;
An information processing method for calculating a second similarity between extracted categories based on terms and the number of terms corresponding to each extracted category. The information processing method according to claim 1, wherein the number of combinations between categories of a plurality of users is calculated by referring to a storage unit that stores categories in association with each user and counting combinations between categories for each user. . The calculated number of combinations between the first category and other categories other than the second category, and the calculated number of combinations between the second category and other categories other than the first category; The information processing method according to claim 1 or 2, wherein a first similarity between the first category and the second category is calculated based on the first category. On the computer,
Refer to the category associated with the user, calculate the number of combinations between categories,
Based on the calculated number of combinations between categories, calculate the first similarity between categories,
Extracting a combination of categories having a first similarity exceeding a threshold;
A program that executes a process of calculating a second similarity between extracted categories based on terms and the number of terms corresponding to each extracted category. A calculation unit that refers to the category associated with the user and calculates the number of combinations between categories;
A first similarity calculator that calculates a first similarity between categories based on the calculated number of combinations between categories;
An extraction unit for extracting a combination of categories having a first similarity exceeding a threshold;
An information processing apparatus comprising: a second similarity calculation unit that calculates a second similarity between extracted categories based on terms and the number of terms corresponding to each extracted category.

Images