JP2015005091A - Information collection device and information collection program - Google Patents

Abstract

PROBLEM TO BE SOLVED: To collect information about the user's taste and preference or the like from the Internet.SOLUTION: An information collection device of the present invention includes: acquisition means that obtains a retrieval query which is input by a first user; extraction means that extracts a contributed content which is related to the retrieval query; and association means that associates the first user with a second user who contributed the extracted contributed content.

Description

  The present invention relates to a technology of an information collection device and an information collection program.

  2. Description of the Related Art Conventionally, a technique for collecting information related to a user's hobbies and preferences based on an action history related to the user's Web is known. For example, it is possible to collect information related to the user's hobbies and preferences by accumulating information on the user's past search queries (search terms) and websites referred to in the past and analyzing the contents thereof. Also, based on the past behavior of a community and the past behavior of another community, the similarity between the two communities is determined, and the preference of the other community is estimated based on the preference of one community There is also a technique (see, for example, Patent Document 1).

  Information on the user's hobbies and preferences collected in this way is used, for example, for display / distribution of a targeting advertisement or the like for the user.

JP 2008-102846 A

  In recent years, techniques for displaying and distributing advertisements and contents on the Internet have become increasingly sophisticated and diversified. For this reason, in order to present more effective information to the user, wide and detailed information about the user is indispensable.

  The present invention has been proposed in view of the above points, and it is an object of the present invention to propose a technique for collecting information related to a user's hobbies and preferences from the Internet.

  In order to solve the above problems, in the information collection device according to the present invention, an acquisition unit that acquires a search query input by a first user, and a post related to the search query based on the search query Extraction means for extracting content, and association means for associating the first user with the second user who has posted the extracted posted content.

  According to the embodiment of the present invention, it is possible to collect information related to a user's hobbies and preferences from the Internet.

It is a network block diagram of an information collection system. 2 is a diagram illustrating a hardware configuration example of an information collection server 3. FIG. 2 is a diagram illustrating an example of a software configuration of an information collection server 3. FIG. It is a figure which shows the data structural example of user information DB301c. It is a figure which shows the data structural example of the search query log 301a. It is a figure which shows the data structural example of the contribution content log 301b. It is a figure showing an example of data composition of user relevance information 301d. 5 is a flowchart showing information collection processing of the information collection server 3. It is a figure which shows a user's relevance.

  Hereinafter, preferred embodiments of the present invention will be described.

<Configuration>
(Network configuration)
FIG. 1 is a network configuration diagram of an information collection system according to an embodiment of the present invention. In the information collection system, a search server 1, a blog server 2, an information collection server 3, and a terminal 4 are connected via a network 5.

  The search server 1 provides the terminal 4 with an information search service for searching for information on the Internet. Thus, the search server 1 is configured as a so-called search engine, for example. When a search query (search term) is input from the user terminal 4, the search server 1 searches the Internet for information (for example, a website) that matches the search query.

  The blog server 2 is a server that provides a blog service called a mini blog, a micro blog, a tweet blog, or the like. The user posts posted content such as his / her situation and miscellaneous information in the blog server 2 in a short sentence. It is also possible to communicate through the exchange of text posted between users in the blog. In most cases, the content posted to the blog server 2 is a short text and can be easily updated. For this reason, real-time communication is often performed. The blog server 2 can also post information such as images, videos, and URLs (Uniform Resource Locator).

  The information collection server 3 collects information about the user (for example, hobbies and preferences). For collection, for example, based on an action history related to the search user's Web in the search server 1, information related to the search user's hobbies and preferences is collected and analyzed. Furthermore, in this embodiment, as will be described later, based on the posted content of the blog server 2, information related to the search user's hobbies and preferences is collected and analyzed. This will be described later in detail again.

  The terminal 4 is a user terminal used by the user. A user accesses the search server 1 using the terminal 4 and inputs a search word from the search screen to search information on the Internet (for example, a website). Further, the user accesses the blog server 2 using the terminal 4 and posts posted content such as his / her own situation and miscellaneous notes. Note that the terminal 4 includes information processing devices such as a PC (Personal Computer), a smartphone, and a mobile phone. Further, in order to access the search server 1 and browse the search screen, for example, an application such as a Web browser is provided.

  The network 5 is a network including wired and wireless, and is, for example, an Internet network that connects the search server 1, the blog server 2, and the terminal 4.

(Hardware configuration)
FIG. 2 is a diagram illustrating a hardware configuration example of the information collection server 3 according to the embodiment of the present invention. The information collecting server 3 includes a central processing unit (CPU) 31, a read only memory (ROM) 32, a random access memory (RAM) 33, an input device 34, an output device 35, a communication device 36, and a hard disk drive (HDD) 37. Have.

  The CPU 31 executes various programs and performs arithmetic processing. The ROM 32 stores a program necessary for starting up. The RAM 33 temporarily stores processing performed by the CPU 31 and stores data. The input device 34 is a keyboard or a mouse (including a touch panel). The output device 35 is a display that displays and outputs video and images, and a speaker that outputs sound and the like. The communication device 36 communicates with other devices such as the terminal 4 via the network 5. The HDD 37 stores various data and programs.

(Software configuration)
FIG. 3 is a diagram showing a software configuration example of the information collection server 3 according to the embodiment of the present invention. The information collection server 3 includes a storage unit 301, a search query acquisition unit 302, a posted content extraction unit 303, a user association unit 304, and a user information collection unit 305 as main functional units.

  The storage unit 301 has a function of storing a search query log 301a, a posted content log 301b, a user information DB (Database) 301c, and user relevance information 301d. These pieces of information will be described later along with specific examples.

  The search query acquisition unit 302 has a function of acquiring a search query input by a search user and an input time of the search query.

  The posted content extraction unit 303 calculates a relevance value between the search query and the posted content based on the search query input by the search user and the input time of the search query, thereby posting related to the search query. It has a function of extracting content (posted content highly relevant to a search query).

  When the relevance between the search query input by the search user and the posted content posted by the blog user is high, the user association unit 304 includes the search user who input the search query and the blog user who posted the extracted posted content. Has the function of associating with. The search query is entered by the search user, and the posted content is posted by the blog user. If the search query is highly related to the posted content (the more relevant), the search query This is because the input search user and the blog user who posted the posted content are likely to be the same user or a user with similar hobbies and preferences. This will be described later in detail again.

  The user information collection unit 305 obtains information related to the blog user (for example, hobby preferences) from the posted content posted by the blog user associated with the search user input by the search query, that is, information related to the search user (for example, hobby). It has a function to collect preferences, etc.).

  The main functional configuration of the information collection server 3 has been described above. Each of these functions is actually realized by causing a computer to implement the program by the CPU 31 of the information collection server 3.

(User information DB 301c)
FIG. 4 is a diagram showing a data configuration example of the user information DB 301c according to the embodiment of the present invention. The user information DB 301c includes data items such as “search user ID”, “user name”, “age”, “address”, “hobby preference”, “similar blog user ID”, and “hobby preference (extended)”.

  “Search user ID” is an ID (unique identifier) of a search user who uses the search server 1. The “search user ID” is used as an identifier for identifying a search query input person (searcher).

  “User name”, “age”, and “address” are the user name, age, and address of the search user. The search user attribute information is registered when the search user who uses the search server 1 performs user registration, for example. Further, other attribute information may be registered.

  “Hobby preference” is information relating to a search user's hobbies and preferences. The “hobby preference” can be registered when the search user himself performs user registration. Further, based on the search user's behavior history regarding the Web, information related to the search user's hobbies and preferences may be collected and analyzed, and the result may be registered.

  The “similar blog user ID” is a blog user who is likely to be the same person as the search user or a blog user who is highly likely to have similar hobbies and tastes after the information collection process described later (with similar blog user User ID) is stored. Therefore, it can be said that the blog user stored in the “similar blog user ID” column is the same hobby preference group member who has the same hobby preference as the search user.

  “Hobby preferences (extended)” is information on hobby preferences of similar blog users collected from posted content posted by similar blog users. In addition, since the search user of “search user ID” and the blog user of “similar blog user ID” have similar hobbies and preferences, “hobby preferences (extended)” is simultaneously the hobbies and preferences of the search users themselves. Information is likely to be. Therefore, in this embodiment, the information regarding the hobby preference of the similar blog user collected from the posted content posted by the similar blog user is handled as the hobby preference of the search user itself.

  For example, “search user ID” y-taro is interested in watching soccer games, cars, and traveling as “hobby preferences”. This “hobby preference” is information that is registered when the search user himself / herself performs user registration, or obtained based on the user's behavior history regarding the Web.

  Here, according to FIG. 4, since “search user ID” y-taro is associated with “similar blog user ID” taro12 and taro332, “search user ID” y-taro and “similar blog user” IDs “taro12 and taro332” are similar in “hobby preference”. Therefore, the “hobby preference” of the “search user ID” y-taro can be extended to the sake of the same “hobby preference” as the “similar blog user IDs” taro12 and taro332.

(Search query log 301a)
FIG. 5 is a diagram showing a data configuration example of the search query log 301a according to the embodiment of the present invention. The search query log 301a includes log data items such as “time” when the search query is input, “search user ID” of the search user who inputs the search query, and “search query” that is input. The “time” includes year / month / day / hour / minute / second.

  The search user accesses the search server 1 using the terminal 4 and inputs a search query (search word) from the search screen to search for information on the Internet (for example, a website). At this time, the information collection server 3 (or the search server 1) accumulates logs such as “time”, “search user ID”, and “search query” in the search query log 301a as the search query log.

  In this embodiment, in order to identify the search user who has input the search query, the search user (for example, “search user ID” y-taro) is already logged in on the search screen. Of course, as long as the search user who input the search query can be identified, other identifiers that can identify the search user may be used. For example, a search user can be identified by using a browser cookie and a browser identifier. In this case, there is no need to be logged in.

(Posted content log 301b)
FIG. 6 is a diagram showing a data configuration example of the posted content log 301b according to the embodiment of the present invention. The posted content log 301b includes log data items such as “time” of posting, “blog user ID” of the blog user who posted the posted content, and “posted content” posted as posted content information.

  The blog user accesses the blog server 2 using the terminal 4 and posts posted content such as his / her situation and miscellaneous notes. On the other hand, the information collection server 3 acquires information such as “time”, “blog user ID”, “posted content”, etc. from the posted content that is published by the blog server 2 at regular time intervals or at a predetermined timing. This is accumulated in the posted content log 301b as a log. In addition, when the user information details (profile details) of the posted blog user are made public, “blog user ID”, “nickname”, “age”, “address”, “hobby”, etc. within the disclosure range Are stored in the posted content log 301b.

(User relevance information 301d)
FIG. 7 is a diagram showing a data configuration example of the user association degree information 301d according to the embodiment of the present invention. The user relevance information 301d includes data items such as “search user ID”, “blog user ID”, and “relevance value”.

  As described above, the posted content extraction unit 303 determines the relevance value (similarity value) between the search query and each posted content based on the search query input by the search user and the input time of the search query. ) To extract post content related to the search query (highly related to the search query). At this time, the higher the relevance value between the search query and the posted content, the more likely that the search user who entered the search query and the blog user who posted the posted content are the same person, or a hobby preference Is likely to be a member of the same hobby group.

  Accordingly, the posted content extraction unit 303 posts the search query and the relevance value (similarity value) between the posted content and the search user who has input the search query in the user relevance information 301d and the posted content. Adopted as a relevance value indicating the degree of relevance with the blog user. This is because the higher the relevance value (similarity value) between the search query and the posted content, the higher the possibility that they are the same user, or it can be said that the users are very close to each other even if they are not the same user. .

  Note that the relevance value is not determined by only the relevance value between one search query and one post content, but a plurality of search queries over a certain period of time in order to increase the reliability of the relevance value. And a relevance value with a plurality of posted contents. Thus, for example, based on the search query newly entered by the search user and the relevance value between the newly posted content, the search user who entered the search query, and the blog user who posted the posted content The relevance value is updated each time.

<Operation>
FIG. 8 is a flowchart showing information collection processing of the information collection server 3 according to the embodiment of the present invention. Hereinafter, it will be described in detail with reference to the drawings. As a premise, the information collection server 3 has data of a search query log 301a, a posted content log 301b, and a user information DB (Database) 301c in advance.

  S1: The search query acquisition unit 302 of the information collection server 3 acquires the search query input by the search user and the input time of the search query from the search query log 301a.

  For example, in the case of FIG. 5, the search query “world cup” and “Japan” input by “search user ID” y-taro and the input time “3/10 22:18” of the search query are acquired.

  S2: The posted content extraction unit 303 extracts the posted content related to the search query (highly related to the search query) from the posted content log 301b based on the search query input by the search user.

  Specifically, the search query input by the search user is compared with the posted content (content), and the relevance value is calculated. The relevance value between the search query and the posted content (content) can be calculated using, for example, morphological analysis and cosine similarity. If the relevance between the search query and the posted content is high, the posted content is extracted. Note that the low relevance to the search query is determined based on the relevance value. If the relevance value is equal to or higher than a predetermined threshold (for example, 0.7), the posted content is highly relevant to the search query. Is determined.

  For example, the search queries “World Cup” and “Japan” are compared with the posted contents (contents) of “Nippon, World Cup participation decision congratulations”, and the relevance value is calculated. For example, when the relevance value is 0.800, the posted content is extracted.

  Also, for example, compare the search queries “World Cup” and “Japan” with the posted content (contents) of “I haven't been to watch the game recently. Is calculated. For example, if the relevance value is 0.846, the posted content is extracted.

  Further, for example, the search queries “World Cup” and “Japan” are compared with the posted content (contents), “The content was determined, but the Japanese World Cup was good”, and the relevance value is calculated. For example, when the relevance value is 0.723, the posted content is extracted.

  On the other hand, for example, the search queries “World Cup” and “Japan” are compared with the posted content (contents) of “Now we will make a hotel reservation for GW.”, And the relevance value is calculated. For example, when the relevance value is 0.023, the posted content is not extracted.

  S3: Next, the posted content extraction unit 303 extracts the posted content related to the search query (higher related to the search query) from the posted content extracted in S2 based on the input time of the search query.

  Specifically, the posted content extraction unit 303 extracts the posted content posted at the time before and after the input time of the search query (for example, 30 minutes before and after) from the posted content extracted in S2. The input time of the search query is close to the posting time of the posted content. For example, there is a possibility that the search user who inputs the search query and the blog user who posted the posted content have a degree of relevance higher than a predetermined level. Because it is expensive.

For example, in the case of FIG. 5, the input time of the search query is “3/10 22:18”. Therefore, the posted content posted within 30 minutes before and after 3/10 22:18 is extracted. Specifically, in the case of FIG. 5, the following posted content is extracted.
“20xx / 3/10 22:25 taro12 Nippon, congratulations on the decision to participate in the World Cup”
“20xx / 3/10 22:35 taro12 I have n’t been to watch recently. I want to watch the game against Brazil live!”
“20xx / 3/10 22:30 taro 332 Regardless of the content, the Japanese World Cup has been decided”
Note that S2 and S3 may be processed at the same time in the same step. That is, the posted content related to the search query is extracted based on the search query and the input time of the search query.

  S4: Next, the user association unit 304 associates the search user who has input the search query with the blog user who has posted the extracted posted content based on the relevance value.

  FIG. 9 is a diagram showing the relationship of users according to an embodiment of the present invention. In S2, the relevance value between the search query and the posted content is calculated. If the relevance between the search query and the posted content is high (for example, the relevance value is equal to or greater than a predetermined threshold), the search entered by the search query There is a high possibility that the user and the blog user who posted the posted content are the same user or a user with a similar hobby preference. Therefore, the relevance value calculated in S2 is used, and the relevance value is adopted as a numerical value indicating the closeness of relevance between the search user who input the search query and the blog user who posted the posted content.

  For example, if the relevance value between the search queries “World Cup” and “Japan” and the posted content (contents) of “Nippon, World Cup participation congratulations” is 0.800, this relevance value is 0. .800 is registered in the user relevance information 301d as the relevance value with the “blog user ID” taro12 for the “search user ID” y-taro.

  In addition, for example, the relevance value between the search queries “World Cup” and “Japan” and the posted content (contents) of “I haven't been to watch the game recently. In this case, the relevance value 0.846 is registered in the user relevance information 301d as a relevance value with the “blog user ID” taro12 for the “search user ID” y-taro. However, if there is already registration in the user association degree information 301d, the existing association degree value is updated by incorporating the current association degree value. For example, if the existing relevance value is 0.800 and the current relevance value is 0.846, for example, an average is taken and updated to 0.823 (FIG. 7). Thus, the reliability of the relevance value is increased by considering a plurality of relevance values between the search query and the posted content.

  Also, for example, when the relevance value between the search query “World Cup” and “Japan” and the posted content (content) is “It was good to have decided the World Cup in Japan” is 0.723. The relevance value 0.723 is registered in the user relevance information 301d as a relevance value with the “blog user ID” taro 332 for the “search user ID” y-taro (FIG. 7).

  Here, referring to FIG. 7, “search user ID” y-taro and “blog user ID” taro 12 are users who have the same user or similar hobbies in the degree of association indicated by the association degree value 0.823. I understand that there is. Similarly, it can be seen that “search user ID” y-taro and “blog user ID” taro 332 are the same users or users with similar hobbies in terms of the degree of association indicated by the association degree value 0.723.

  In the present embodiment, the description has been made on the assumption that there is a relatively high degree of relevance of posted content. However, there may be cases where only posted content having a low relevance level exists. In such a case, even if the blog user has a low relevance level, considering the case where the relevance level increases due to subsequent posted content, in the user relevance level information 301d, a low relevance level value is used together with the search user. Make an association. However, in the case of a blog user with a very low degree of association, it is not necessary to associate with a search user who has input a search query.

  Next, the user association unit 304 refers to the user relevance information 301d, and a blog user whose relevance value is equal to or greater than a predetermined threshold with respect to the search user who has input the search query is the same user or a user with a similar hobby preference. Is registered in the “similar blog user ID” of the user information DB 301c.

  For example, for the “search user ID” y-taro that has entered the search query, the “blog user ID” taro12 and taro332 whose relevance value is a predetermined threshold value 0.7 or more are the same users as the “search user ID” y-taro. Or it considers that it is a user with a similar hobby preference, and it registers into "similar blog user ID" of user information DB301c (FIG. 4).

  The “search user ID” y-taro is a person who has searched a website related to “world cup” and “Japan” by inputting a search query “world cup” and “Japan”. That is, there are images of people who are interested in “World Cup” and “Japan”. On the other hand, in the posted content of the blog, there are “blog user IDs” taro12 and taro332 that posted the posted content related to “World Cup” and “Japan” in a close time zone. Therefore, the “search user ID” y-taro and the “blog user IDs” taro12 and taro332 have different user IDs on the system of the search server 1 and the blog server 2, but may be the same person. Alternatively, it can be seen that even if they are not the same person, they are users who have similar hobbies in terms of the keywords “World Cup” and “Japan”. As described above, the “search user ID” y-taro and the “blog user ID” taro12 and taro332 in this embodiment are at least persons having the same or similar hobbies regardless of whether or not they are the same person ( )).

  S5: The user information collection unit 305, from the posted content posted by the blog user associated with the search user input by the search query, information related to the blog user (for example, hobby preferences), that is, information related to the search user (for example, Hobbies and preferences).

  For example, referring to the user information DB 301c (FIG. 4), the “search user ID” y-taro has the same or similar hobby preference as the “similar blog user IDs” taro12 and taro332. Therefore, the user information collection unit 305 searches the posted content posted by the “similar blog user ID” taro12 and taro332 in the past, and analyzes the hobby preferences of the taro12 and taro332. Then, the user information collection unit 305 registers the analyzed hobbies and preferences of taro12 and taro332 in “hobby preferences (extended)” of the user information DB 301c (FIG. 4).

  For example, there is a posted content of taro332 that says “20xx / 3/11 21:04 YYY I had a delicious sake. When the user information collection unit 305 analyzes that the blog user taro 332 has a hobby preference of sake, the user information collection unit 305 registers the “hobby preference (extended)” of the search user y-taro in the user information DB 301c (FIG. 4). According to the “hobby preference” of the user information DB 301c, the hobby preference of the search user y-taro is only the soccer watching, the car, and the travel. The hobby preference can be further treated as having a hobby preference of sake.

  Conventionally, for example, when distributing advertisements related to hobbies and preferences to “search user ID” y-taro, advertisements related to watching soccer games, cars, and travel are distributed. However, according to the present embodiment, hobbies and preferences can be expanded, so it is also possible to distribute advertisements related to sake to “search user ID” y-taro. Of course, in addition to the distribution of the advertisement, various controls can be performed on the “search user ID” y-taro based on the taste preference.

<Supplement>
In S4 described above, the relevance value between the search query and the posted content (content) is registered in the user relevance information 301d as the relevance value with the blog user for the search user (FIG. 7). Here, in general, the user information details (profile details) of the posted blog user may be disclosed together with the posted content. As described above, such user information details are accumulated in the posted content log 301b within the disclosure range.

  Therefore, the user association unit 304 includes user attribute information (for example, “user name”, “age”, “address”) in the user information DB 301c and user information details (for example, “blog user ID”) in the posted content log 301b. , “Nickname”, “age”, “address”, “hobby”), and the degree of association with the blog user with respect to the search user can be corrected.

  That is, as the user attribute information of the search user and the user information details of the blog user match (similar), the search user and the blog user are more likely to be the same person, so the relevance value is higher. Correct to the value. By doing in this way, the reliability of a relevance value can be raised more.

<Summary>
As described above, according to the embodiment of the present invention, it is possible to collect information on the user's hobbies and preferences from the Internet. The present invention has been described with reference to preferred embodiments of the present invention. While the invention has been described with reference to specific embodiments, various modifications and changes may be made to the embodiments without departing from the broad spirit and scope of the invention as defined in the claims. Obviously you can. In other words, the present invention should not be construed as being limited by the details of the specific examples and the accompanying drawings.

1 Search Server 2 Blog Server 3 Information Collection Server 4 Terminal 5 Network 31 CPU
32 ROM
33 RAM
34 Input device 35 Output device 36 Communication device 37 HDD
301 Storage Unit 302 Search Query Acquisition Unit 303 Post Content Extraction Unit 304 User Association Unit 305 User Information Collection Unit

Claims (7)

Obtaining means for obtaining a search query input by a first user;
Extraction means for extracting post content related to the search query based on the search query;
Association means for associating the first user with the second user who has posted the extracted posted content;
An information collecting apparatus comprising: The extraction means includes
Calculating a relevance value between the search query and the post content, and extracting post content having a relevance value greater than or equal to a predetermined value;
The information collecting apparatus according to claim 1. The extraction means includes
Extracting the input time of the search query and posted content posted at a predetermined time before and after,
The information collecting apparatus according to claim 2. The association means includes
Associating the first user with the second user based on a relevance value between the search query and the posted content;
The information collection device according to claim 2 or 3, The extraction means includes
Calculating the relevance value including a relevance value between user information of the first user and user information of the second user;
The information collection device according to claim 1, wherein the information collection device is an information collection device. Information collecting means for collecting information about the first user from the posted content of the second user associated with the first user;
6. The information collecting apparatus according to claim 1, further comprising: On the computer,
Obtaining means for obtaining a search query input by a first user;
Extraction means for extracting post content related to the search query based on the search query;
An information collection program for causing the first user to function as an association means for associating the extracted second posted content with the second user.

Images