JP5736292B2 - Post message transmission program, apparatus, and method for changing post message to be recommended according to reaction of other contributor - Google Patents

Description

  The present invention relates to a recommendation service technology that recommends content that matches a user's preference, and a miniblog server technology that publishes posted text to a group of other users such as SNS (Social Networking Service).

  Conventionally, as a recommended service technology, there is a technology that distributes information related to a product / service to be recommended to the user based on purchase history information of products / services for each user, viewing / viewing history information, etc. Patent Document 1). In addition, when a user accesses a website from a terminal, there is a technique that determines an affiliate display advertisement for the user according to the number of clicks on the website (see, for example, Patent Document 1). According to this technique, an affiliate advertisement is automatically selected using a recommendation engine.

  Further, there is a technology for returning the stock status of products / services recommended to the user, weather information of the location requested by the user, etc. as a posted text from the content of the posted text transmitted from the user (for example, non-patented). References 2 and 3). According to this technique, the user's posted text is based on a communication service such as a bulletin board, chat, and SNS service. Here, the posted text to be recommended to the user is automatically posted by the bot technique. In particular, according to twitter (registered trademark) (asynchronous chat system), which is a typical SNS service, it is possible to periodically and automatically transmit a postal sentence imitating a real person using bot. it can.

JP 2011-39909 A

G. Linden, B. Smith, and Jeremy York, “Amazon.com Recommendations Item-to-Item Collaborative Filtering,” IEEE Internet Computing, IEEE Computer Society, 2003. “Collecam.net”, [online], [searched August 30, 2011], Internet <URL: http://korekamo.net/> "Weather bot", [online], [Search August 30, 2011], Internet <URL: http://twitter.com/otenki_bot/>

  However, according to the recommendation service technologies such as Patent Document 1 and Non-Patent Document 1, the information recommended to the user is advertisement information such as an affiliate on the Web site. For this reason, users who have a strong image as a commercial and have a low willingness to purchase are not only less interested but often do not want to distribute the advertisement information itself.

  Further, according to the communication service technology using bots as in Non-Patent Documents 2 and 3, useful advertisement information is provided by analyzing the search action history and the like in interactive communication with the user. be able to. However, the advertisement information is not recommended information that matches the user's preference. Of course, there is a service that analyzes the content of the posted posts of the twitter and leads to marketing, but it does not analyze the preferences of followers.

  Therefore, the present invention provides a posted message transmission program, apparatus, and method that can distribute a recommended posted message as a recommendation service using a bot technology in interactive communication with a user in an SNS service. With the goal.

According to the present invention, the group consisting customize posts projecting draft's from multiple other contributors viewable, and blog site server to publish posted messages posted by the author, via a network A posted message transmission program for causing a computer mounted on a communicable device to function,
A preference item table in which one or more preference items are registered;
Recommended preference item derivation means for sequentially deriving recommended preference items related to preference items in the preference item table;
Using a template sentence, a posted sentence generating means for sequentially generating a posted sentence including a recommended preference item in a part thereof,
Posting means for sequentially posting posted sentences to the blog site server;
Reaction information collection means for collecting reaction information of other contributors to the posted text from the blog site server,
It is characterized by deriving a recommended preference item with a high response from another poster from the response information, and causing the computer to function as a feedback means for newly registering the recommended preference item in the preference item table.

According to another embodiment of the posted message transmission program of the present invention,
The recommended preference item derivation means is:
For a large number of user preferences, each user has a user preference database representing the presence or absence of preference for each preference item,
It is also preferable to make the computer function so as to sequentially derive recommended preference items that cooperate with all preference items in the preference item table by collaborative filtering using the user preference database.

According to another embodiment of the posted message transmission program of the present invention,
The reaction information collecting means counts the number of reaction operations of other contributors to the posted text from the blog site server,
The feedback means also preferably causes the computer to function so as to newly register a recommended preference item included in a posted sentence having a predetermined number of reaction operations or more in the preference item table.

  According to another embodiment of the posted message transmission program of the present invention, it is also preferable to cause the computer to function so that the number of reaction operations is the number of link clicks, the number of spread transmissions, and / or the number of favorite registrations.

According to another embodiment of the posted message transmission program of the present invention,
The posted message generation means further includes the address to the redirect page provided by the posted message transmission device in the posted message,
The reaction information collection means counts the number of redirects for each posted sentence,
The feedback means also preferably causes the computer to function so as to newly register a recommended preference item included in a posted sentence having a predetermined number of redirects or more in the preference item table.

According to another embodiment of the posted message transmission program of the present invention,
The reaction information collecting means collects the posted texts of a plurality of other contributors who can view the posted texts of the corresponding posters,
Feedback means
It has a preference item search engine in which various keywords are associated with preference items,
A keyword is detected from a plurality of collected postings by morphological analysis or N-gram index character string extraction method, and the favorite item is searched by inputting the keyword into a favorite item search engine, and the favorite item is recommended. It is also preferable to make the computer function so as to be newly registered in the preference item table as a preference item.

According to another embodiment of the posted message transmission program of the present invention,
At the initial stage, the preference item table registers initial preference items in advance,
It is also preferable to cause the computer to function so that the initial preference item is set based on a user operation.

According to another embodiment of the posted message transmission program of the present invention,
The preference item table registers the time stamp of the registration date for each preference item,
Feedback means, the preference item with elapsed past time stamps for a certain period of time, it is also preferable to remove from the preference item table.

According to the present invention, the group consisting customize posts projecting draft's from multiple other contributors viewable, and blog site server to publish posted messages posted by the author, via a network A communicable posted message sending device,
A preference item table in which one or more preference items are registered;
Recommended preference item derivation means for sequentially deriving recommended preference items related to preference items in the preference item table;
Using a template sentence, a posted sentence generating means for sequentially generating a posted sentence including a recommended preference item in a part thereof,
Posting means for sequentially posting posted sentences to the blog site server;
Reaction information collection means for collecting reaction information of other contributors to the posted text from the blog site server,
It is characterized by having a feedback means for deriving a recommended preference item with a lot of responses from other posters from the response information and newly registering the recommended preference item in the preference item table.

According to the present invention, the group consisting customize posts projecting draft's from multiple other contributors viewable, and blog site server to publish posted messages posted by the author, via a network A method for sending posted text in a communicable device,
Having a preference item table in which one or more preference items are registered;
A first step of deriving a recommended preference item related to the preference item in the preference item table;
A second step of using a template sentence to generate a posted sentence including a recommended preference item in a part thereof;
A third step of posting a post to the blog site server;
While repeating the first step to the third step sequentially,
A fourth step of collecting reaction information of other contributors to the posted text from the blog site server;
And a fifth step of deriving a recommended preference item having a high response from another poster from the response information and newly registering the recommended preference item in the preference item table.

  According to the posted message transmission program, apparatus, and method of the present invention, a recommended posted message as a recommendation service can be distributed using the bot technique in interactive communication with a user in an SNS service.

It is a system configuration diagram in the present invention. It is a functional block diagram of the 1st contribution transmission apparatus in this invention. It is a sequence diagram in the present invention. It is explanatory drawing which represented the operation | movement of collaborative filtering simply. It is a functional block diagram of the 2nd posting text transmission apparatus in this invention. It is a function block diagram of the 3rd posted message transmission apparatus in this invention.

  Hereinafter, embodiments of the present invention will be described in detail with reference to the drawings.

  FIG. 1 is a system configuration diagram according to the present invention.

  According to FIG. 1, a blog site server 3 is connected to the Internet. “Blog site server” is an SNS (Social Networking Service) server, for example, which is posted to a group of a plurality of other contributors who can view the posted text of the contributor. Publish your post. The SNS server is a miniblog server such as twitter (registered trademark), facebook (registered trademark), or mixi (registered trademark).

  According to FIG. 1, a large number of terminals 2 access the blog site server 3 via an access network and the Internet. For example, it is assumed that a group composed of a plurality of other contributors B and C (accounts B and C) that can view the posted text of the contributor A (account A) is configured. In this case, the posted text posted by the poster A is viewed only on the terminal operated by the posters B and C. According to FIG. 1, for example, a posted message transmitting apparatus 1 accessed as an account A and a terminal 2 accessed as accounts B and C are connected to the blog site server 3.

  The account A assigned to the posted message transmitting apparatus 1 is for bot. Here, it is also preferable that accounts for a plurality of bots such as contents, news, and topics are provided in accordance with the preference of the user who follows the account.

  The posted message transmitting apparatus 1 according to the present invention distributes a recommended posted message (tweet) as a recommendation service in the interactive communication with the user in the SNS service by using the bot technology. Depending on the reaction (preference) of other users to the recommended posted text, the posted content of the recommended posted text is automatically updated.

FIG. 2 is a functional configuration diagram of the first posted message transmitting apparatus according to the present invention.
FIG. 3 is a sequence diagram in the present invention.

  The posted message transmitting apparatus 1 includes a communication interface 10 connected to the Internet, an API (Application Programming Interface) unit 11, a preference item table 12, a recommended favorite item derivation unit 13, a posted message generation unit 14, and a posting unit 15. And a reaction information collecting unit 16 and a feedback unit 17. These functional components excluding the communication interface 10 are realized by executing a posted message transmission program that causes a computer installed in the terminal to function. In the following description, it is assumed that the miniblog server is a twitter server. Also, it is assumed that the recommendation target is “movie”, for example, and a large number of users are following the account A of the posted message transmission apparatus 1.

  The API unit 11 is a rule interface for the application of the posting unit 15 and the reaction information collection unit 16 to use the function of the application service of twitter. The API unit 11 is prepared differently for each miniblog server.

  The preference item table 12 is a table in which one or more preference items are registered. Moreover, it is also preferable that the time stamp of the registration date is registered for each preference item in the preference item table.

  An arbitrary preference item (initial preference item) is registered in the preference item table 12 as an initial setting in advance. According to FIG. 2, when recommending a movie, for example, “movie A” and “movie B” are registered as movie titles. The initial preference item is set based on a user operation.

  The recommended preference item derivation unit 13 sequentially derives recommended preference items related to the preference item in the preference item table 12 (see S13 in FIG. 3). Specifically, for example, using a collaborative filtering algorithm using a user preference database, other preference items that are collaborative (similar) to the preference items registered in the preference item table 12 are found.

  Collaborative filtering is an engine technology that uses a user preference database that accumulates preference information of a large number of users and automatically infers using information of other users with similar preferences to a certain user. Based on the “word-of-mouth” principle of referring to opinions of others with similar hobbies.

  According to the present invention, as this collaborative filtering, it is assumed that an existing recommendation engine and an existing user preference database are used. That is, the user preference database is assumed to be a profile of numerical values of purchase browsing history for each user collected by other services, for example.

  FIG. 4 is an explanatory diagram simply showing the operation of collaborative filtering.

  According to FIG. 4, for example, a virtual user who likes movie A and movie B registered in the preference item table is assumed. Moreover, the presence or absence of preference of each preference item is represented in the user preference database for each of a large number of existing users a to d. It searches for “movie C” that other users a to d who like movie A and / or movie B like, and infers that the virtual user also likes “movie C”, and sets it as a recommended preference item. According to FIG. 4, although only described with ○, in terms of implementation, the analogy between users is analogized by representing the correlation coefficient of evaluation given to the same item. The virtual user behavior and the existing other user behavior accumulated in the user preference database are analyzed in correlation, and a user with a high correlation is assumed to be “close to preference”. If “preference is close”, it is predicted that the user likes the same action, and recommends an action (purchasing a product or viewing an article) that exists in the existing other user action but not in the virtual user action. This algorithm is implemented as a standard function in advance in spreadsheet software, for example.

  The posted sentence generation unit 14 sequentially generates a posted sentence including a recommended preference item as a part of the template sentence (see S14 in FIG. 3). In the posted text, the posted text sent from the account A for bots appears to be spoken by an arbitrary character, so that the viewer may feel human friendliness with respect to the bot character. it can.

Examples of template sentences include the following, and several types are prepared.
“My recommendation is <A>!
http://www.lismovideo.jp/content_id= <B>"
In <A>, the title found in the recommendation process is embedded.
In <B>, the ID of the recommended content is embedded.
For example, in the case of a movie, it is preferable to use different templates according to meta information such as genres such as suspense and comedy.
In addition, it is preferable that the posted text generation unit 14 further embeds meta information of content to be recommended in the template text.

  The posting unit 15 posts posted sentences to the blog site server 3 sequentially (for example, twitter tweets) via the API unit 11 (see S15 in FIG. 3). Here, it is also preferable that the time for transmitting the posted text is scheduled. For example, the date and time for sending a posted message is set. For example, the cycle may be “once every hour” “once every hour from 9:00 to 16:00” “once every hour from 16:00 to 23:00” or “12 A time setting such as “hour 5 minutes” may be used.

  The reaction information collection unit 16 collects reaction information of other contributors to the posted text from the blog site server 3 via the API unit 11 (see S16 in FIG. 3). Specifically, the number of reaction operations of other contributors to the posted text is counted from the blog site server 3 via the API unit 11. The number of reaction operations is, for example, the number of clicks on a link, the number of spread transmissions (for example, ReTweet in twitter), and / or the number of favorite registrations. The number of reaction operations is counted for each posted sentence. Further, the reaction operation as the reaction information is based on the reaction operation in the user who follows the account A of the posted message transmission device 1.

  The feedback unit 17 derives a recommended preference item with a high response from other posters from the response information, and newly registers the recommended preference item in the preference item table (see S17 in FIG. 3). The feedback unit 17 may newly register, for example, a recommended preference item included in a posted sentence having a predetermined number of reaction operations or more in the preference item table. For example, based on the reaction information tabulated on a daily basis, the top several items may be used as recommended preference items.

  Further, the feedback unit 17 deletes a preference item having a past time stamp after a certain period (for example, one week) from the preference item table. Furthermore, it is also preferable that the feedback unit 17 deletes a lower number of unpreferred preference items from the preference item table. By such an operation, the preference item table is constantly updated (tuned) in interactive communication, so that the recommended preference item similar to the registered preference item is always updated.

  FIG. 5 is a functional configuration diagram of the second posted message transmitting apparatus according to the present invention.

  The second posted message transmitting device in FIG. 5 differs from the first posted message transmitting device in FIG. 2 only in the functions of the posted message generating unit 14 and the reaction information collecting unit 16. The posted message generation unit 14 includes a link of an address to a redirect page whose redirect source is the posted message transmission device in the posted message. When another user who has viewed the posted text clicks on this link, the user is redirected to the content detailed information or the like via the posted text transmitting apparatus 1. By clicking on a link in the posted text (tweet), the posted text transmitting device is accessed, and a detailed content information page corresponding to the posted text in the redirect is displayed on the terminal. The reaction information collection unit 16 can count the number of times the posted message transmission device 1 has redirected for each posted message. Thereby, the reaction information collection unit 16 can collect reaction information of other contributors to the posted text from the blog site server 3. Then, the feedback unit 17 newly registers a recommended preference item included in a posted sentence having a predetermined number of redirects or more in the preference item table 11.

  FIG. 6 is a functional configuration diagram of the third posted message transmitting apparatus according to the present invention.

  6 differs from the first posted message transmission device of FIG. 2 only in the functions of the reaction information collection unit 16 and the feedback unit 17, and further includes a preference item search engine 18. Have. The reaction information collection unit 16 collects posted messages (tweets) of other users who follow the account A of the posted message transmission apparatus. Such a function is generally referred to as a “crawler”. “Crawler” refers to a function that automatically and periodically obtains a page that can be viewed by the user, and is provided as, for example, a Twitter API.

  The preference item search engine 18 stores items in which various keywords are associated with preference items, inputs a keyword (query), and outputs an appropriate preference item. When the preference item is a movie name, the preference item search engine 18 associates a keyword extracted from the summary text such as the title, character, contents (abstract) of the movie with the preference item of the movie name. Yes. Note that the search engine is an existing general one.

  The feedback unit 17 detects a keyword from a plurality of collected posted sentences by morphological analysis or N-gram index character string extraction method. Among the keywords, the upper-ranked keywords having a high appearance frequency are input to the preference item search engine 18. The preference item search engine 18 searches for a preference item optimum for the input keyword. The feedback unit 17 newly registers the preference item output from the preference item search engine 18 in the preference item table 11 as a recommended preference item.

  Morphological analysis refers to a technique in which a natural language sentence in a posted sentence is divided into columns of morphemes (the smallest unit having meaning in the language) using a word dictionary and a dependency dictionary, and each part of speech is discriminated. By this analysis, for example, only noun morphemes can be extracted as keywords, and keywords with a high appearance frequency can be detected.

  In addition, N-gram is not based on dictionary or syntactic analysis for key extraction of transposed index (combination of keywords and pages), but simply counts words cut out for every fixed number of characters (for example, every two characters). The technology to do. N means that there can be a plurality of character units to be cut out. For example, the key of the table of contents is created by always cutting out a two-letter word without considering the meaning of the word.

  Of course, it is also preferable to use some filtering function so that only meaningful keys are extracted keys, for example, in accordance with movie contents, without using morphological analysis or N-gram index character string extraction methods as they are.

  As described above in detail, according to the posted message transmission program, apparatus, and method of the present invention, the recommended posted message as the recommendation service is used for the recommended service in the interactive communication with the user in the SNS service. Can be delivered. In particular, according to the present invention, it is possible to distribute a recommended posting according to, for example, a follower preference in twitter. In addition, the posted text does not feel an inorganic image as a general advertisement, and an organic image that imitates a real person in the natural exchange of posted text with the user. You can feel it.

  Various changes, modifications, and omissions of the above-described various embodiments of the present invention can be easily made by those skilled in the art. The above description is merely an example, and is not intended to be restrictive. The invention is limited only as defined in the following claims and the equivalents thereto.

DESCRIPTION OF SYMBOLS 1 Posting text transmission apparatus 10 Communication interface 11 API part 12 Preference item table 13 Recommended preference item derivation part 14 Posting sentence generation part 15 Posting part 16 Reaction information collection part 17 Feedback part 18 Preference item search engine 2 Terminal 3 Blog site server

Claims (10)

Mounting the group consisting customize posts projecting draft's from multiple other contributors viewable, and blog site server to publish posted messages posted by the author, the possible communication apparatus via a network A post message sending program for causing a computer to function,
A preference item table in which one or more preference items are registered;
Recommended preference item derivation means for sequentially deriving recommended preference items related to the preference item in the preference item table;
Using a template sentence, a posted sentence generating means for sequentially generating a posted sentence including the recommended preference item in a part thereof,
Posting means for sequentially posting the posted text to the blog site server;
Reaction information collecting means for collecting reaction information of other contributors to the posted text from the blog site server;
A posted sentence transmission program characterized by deriving recommended preference items with a high response from the other posters from the reaction information, and causing a computer to function as feedback means for newly registering the recommended preference items in the preference item table . The recommended preference item derivation means includes:
For a large number of user preferences, each user has a user preference database representing the presence or absence of preference for each preference item,
2. The posting according to claim 1, wherein the computer functions so as to sequentially derive recommended preference items that cooperate with all the preference items of the preference item table by collaborative filtering using the user preference database. Sentence sending program. The reaction information collecting means counts the number of reaction operations of other contributors to the posted text from the blog site server,
The feedback means causes a computer to function so as to newly register the recommended preference item included in the posted sentence in which the number of reaction operations is equal to or greater than a predetermined number in the preference item table. The posted message transmission program according to 2.   The posted message transmission program according to claim 3, wherein the computer functions so that the number of times of the reaction operation is the number of clicks of a link, the number of spread transmissions, and / or the number of favorite registrations. The posted message generation means further includes an address to a redirect page provided by the posted message transmission device in the posted message,
The reaction information collecting means counts the number of redirects for each posted sentence,
The feedback means causes a computer to function so as to newly register the recommended preference item included in the posted sentence whose number of redirects is a predetermined number or more in the preference item table. The posted text transmission program described in. The reaction information collecting means collects a plurality of other posters that can view the posted text of the poster,
The feedback means includes
It has a preference item search engine in which various keywords are associated with preference items,
A keyword is detected from a plurality of collected postings by a morphological analysis or an N-gram index character string extraction method, and the preference item is searched by inputting the keyword to the preference item search engine. The posted message transmission program according to claim 1, wherein the computer is caused to function as a recommended preference item and newly registered in the preference item table. In the initial stage, the preference item table registers initial preference items in advance,
The posted text transmission program according to any one of claims 1 to 6, wherein the initial preference item causes a computer to function so as to be set based on a user operation. The preference item table registers a time stamp of registration date and time for each preference item,
Before notated fed back means, the preference item with elapsed past time stamps of a certain period, the posted message transmission according to any one of claims 1 to 7, characterized in that to remove from the preference item table program. For a group consisting customize posts projecting draft's from multiple other contributors viewable, and blog site server to publish posted messages posted by the author, communicable posted message sent over the network A device,
A preference item table in which one or more preference items are registered;
Recommended preference item derivation means for sequentially deriving recommended preference items related to the preference item in the preference item table;
Using a template sentence, a posted sentence generating means for sequentially generating a posted sentence including the recommended preference item in a part thereof,
Posting means for sequentially posting the posted text to the blog site server;
Reaction information collecting means for collecting reaction information of other contributors to the posted text from the blog site server;
A posted message transmission apparatus comprising: feedback means for deriving a recommended preference item with a high response from the other posters from the response information and newly registering the recommended preference item in the preference item table. For a group consisting customize posts projecting draft's from multiple other contributors viewable, and blog site server to publish posted messages posted by the author, posts in communicable devices via a network A sentence transmission method,
Having a preference item table in which one or more preference items are registered;
A first step of deriving a recommended preference item related to the preference item in the preference item table;
A second step of generating a posted sentence including the recommended preference item in a part thereof using a template sentence;
A third step of posting the post to the blog site server;
While repeating the first step to the third step sequentially,
A fourth step of collecting reaction information of other contributors to the posted text from the blog site server;
And a fifth step of deriving a recommended preference item with a high response of the other poster from the response information and newly registering the recommended preference item in the preference item table. .

Images