JP5658939B2 - Evaluation system for written articles with hierarchical structure - Google Patents

Description

  The present invention relates to a technique for evaluating the value of written articles in, for example, consumer generated media (CGM).

  In CGM such as word-of-mouth communication, impressions, opinions, additional information, etc. for one product or service (hereinafter referred to as “product etc.”) are written, and the service provider operating CGM stores these writings. And provided to users. The subject of writing differs for EC sites, blogs, and Twitter. The target on the EC site is, for example, a product. The subject in the blog is, for example, a critique of social phenomena. The target in Twitter (registered trademark) is, for example, written personal thoughts and claims.

  Such writing may end at the first writing (first generation), or may continue from the first writing to other writings (second generation, third generation,...). When writing continues from the origin, these writings form a tree structure or a hierarchical structure.

  Patent Document 1 “Bulletin Board System” discloses a bulletin board system that hierarchically displays articles. In this system, the citation relationship by trackback of the blog article is displayed in a hierarchy.

  Thus, while the rise in the field of consumer generated media (CGM) in which consumers use the Internet and the like to generate content is remarkable, there is also an aspect of information flooding.

  Therefore, there is a need for a technique for extracting only useful information for a user in a system (for example, a bulletin board, a word of mouth, a blog, Twitter, etc.) that manages a written article that reacts by quoting other articles.

JP 2007-272330 A

  An object of the present invention is to evaluate the value of written articles based on the characteristics of the hierarchical structure of articles written hierarchically.

  Another object of the present invention is to select and extract valuable written articles from a flooded information group by paying attention to the hierarchical structure, thereby evaluating based on individual written subordinate articles.

  Yet another object of the present invention is to present the extracted writings in an appropriate order to those seeking the writing.

  The evaluation system for written articles according to the present invention is:
Accept written articles and written articles subordinate to the written article from the user terminal, ThatFor each written articleOf written articles to be evaluatedAn evaluation system,
An article writing unit that accepts written articles from the user terminal;
An article data storage unit that stores the written article and stores a dependency relationship between the written articles;
With reference to the article data storage unit,Dependent / subordinatedWrite articleEach ofAboutThe bookInclusive articleBelowIt is inallWriting articlesBy counting in a predetermined way,ThisThe bookInclusive articleEach ofAn article value determination unit that calculates an evaluation value of
An article value table for storing the evaluation value for each written article;
It is provided with.
The article value determination unitAll of the aboveWrite articlesThe total numberCalculate as the evaluation value of written articlesTheFeatures.

Moreover, the article value determination unit, and calculates the number of unique users of the all writing articles as the evaluation value of this該書inclusive article.
In addition, the written article includes a citation description that can identify an upper written article,
The article value determination unit sequentially identifies, for one written article, a written article that is higher than the other written articles based on the citation description included in the other written articles. It is characterized in that all written articles below the one written article are identified by determining whether or not they match the one written article.

The evaluation system further includes an evaluation weight table that stores evaluation weights for each user.
The article writing unit accepts user-specific information together with the written article, and the article data storage unit stores each written article in association with the user-specific information,
The article value determination unit refers to the article data storage unit for one written article, extracts an evaluation weight from the evaluation weight table, and uses the result of the count and the evaluation weight to evaluate the one written article and calculating the value.
The method for evaluating written articles according to the present invention is as follows:
The user terminal receives a write articles that are dependent on the write articles and the write article, a method for evaluating writing articles for evaluating a writing article their respective,
A written article receiving step for receiving a written article from the user terminal; and
An article data storage step of storing the written article in the article data storage unit and storing a dependency relationship between the written articles;
With reference to the article data storage unit, for each of the write articles in a relationship of dependent-be dependent by all writes articles at the bottom of those 該書 included articles counted in a predetermined manner, those 該書 An article value determination step for calculating the evaluation value of each included article;
An evaluation value storage step for storing the evaluation value in the article value table for each written article;
It is provided with.

The program according to the present invention is:
The evaluation system with an article data storage unit and the article value table, a program for executing the evaluation process of writing articles that are dependent on the write articles and the writing article,
A written article receiving step for receiving a written article from the user terminal; and
An article data storage step of storing the written article in the article data storage unit and storing a dependency relationship between the written articles;
With reference to the article data storage unit, for each of the write articles in a relationship of dependent-be dependent by all writes articles at the bottom of those 該書 included articles counted in a predetermined manner, those 該書 An article value determination step for calculating the evaluation value of each included article;
An evaluation value storage step for storing the evaluation value in the article value table for each written article;
This is a written article evaluation program for executing the above.

The computer-readable recording medium according to the present invention is:
The evaluation system with an article data storage unit and the article value table, a program for executing the evaluation process of writing articles that are dependent on the write articles and the writing article,
A written article receiving step for receiving a written article from the user terminal; and
An article data storage step of storing the written article in the article data storage unit and storing a dependency relationship between the written articles;
With reference to the article data storage unit, for each of the write articles in a relationship of dependent-be dependent by all writes articles at the bottom of those 該書 included articles counted in a predetermined manner, those 該書 An article value determination step for calculating the evaluation value of each included article;
An evaluation value storage step for storing the evaluation value in the article value table for each written article;
It is characterized by recording a written article evaluation program for executing.

  Since the evaluation value for extracting the written article is calculated for each written article based on the count value of the lower written article, the user pays attention even in the intermediate writing in the hierarchical structure It is possible to identify a written article with many writings and responses, and a written article with a high appeal / attraction and advertising effect from a large number of written articles. In particular, it also functions effectively in SGM such as Twitter RT (QT).

  The count value in the present invention refers to the number of articles written immediately under one written article and the number of all written articles subordinate to the one written article when attention is paid to one written article. , The number of unique users of the immediately written article subordinate to the one written article, the number of unique users of all the written articles subordinate to the one written article, and the evaluation weights to the existence number and the number of unique users It means a value taken into account (for example, multiplied).

  In other embodiments, the number of written articles directly underneath is counted, so that the magnitude of the direct response can be evaluated for each written article. In this way, written articles with a high direct response can gather many opinions and questions by themselves, and can be said to be highly capable of attracting the attention of the surroundings.

  In addition, by counting the number of unique users instead of the number of articles, it is possible to eliminate writing of self-produced performances, and to extract credible written articles by removing noise.

  Furthermore, by extracting a chain with the maximum number of writings and the number of unique users, it is possible to extract writing articles with very little noise.

  In other embodiments, since all lower-level written articles are counted, the overall response magnitude can be evaluated for the original written article. Such written articles with many overall responses can be said to be highly capable of providing topics that are easy to develop and inducing mutual exchange of opinions.

  In addition, by counting the number of unique users instead of the number of articles, it is possible to eliminate writing of self-produced performances, and to extract credible written articles by removing noise.

  In other embodiments, the evaluation is based on the evaluation weights assigned to the writers of the subordinate writing articles, so that the ability to appropriately respond to those who want to respond can be evaluated. For example, when the reaction of people belonging to a predetermined purchase layer can be obtained like an alpha blogger, the advertising significance is high. Also, if you can get a celebrity reaction, it means that it is highly effective in attracting public attention.

FIG. 1 is a diagram illustrating a network configuration of an article evaluation system. FIG. 2 is a diagram illustrating a tree display example of written articles. FIG. 3 is a diagram illustrating an example of the article data storage unit. FIG. 4 is a diagram showing the configuration of the article management server that constitutes the evaluation system. FIG. 5 shows an article value table. FIG. 6 is a diagram showing a direct reaction number calculation processing flow. FIG. 7 is a diagram showing a total reaction number calculation processing flow. FIG. 8 is a diagram illustrating an example of the article data storage unit according to the second embodiment. FIG. 9 is a diagram showing a direct reaction number calculation processing flow according to the second embodiment. FIG. 10 is a diagram showing a total reaction number calculation processing flow according to the second embodiment. FIG. 11 is a diagram illustrating an example of the evaluation weight table. FIG. 12 is a diagram illustrating an example of the article value table according to the third embodiment. FIG. 13 is a diagram showing a direct reaction weight total calculation output processing flow. FIG. 14 is a diagram showing a total reaction weight total calculation output processing flow.

  FIG. 1 is a diagram illustrating a network configuration of an article evaluation system. The electronic commerce server 101 is an example of a server that manages written articles. The electronic commerce server 101 is connected to the user terminal 102 and the store terminal 103 via the Internet. The store terminal 103 is a terminal of a merchant exhibiting in the virtual shopping mall. The user terminal 102 is a buyer's terminal, and writes word-of-mouth information from the product screen displayed on the browser.

  Hereinafter, word-of-mouth is described as an example of article writing. However, it may be a written article having a hierarchical structure that accepts written articles associated with one article, such as a bulletin board, a blog, and Twitter.

  FIG. 2 is a diagram illustrating a tree display example of written articles. The written article has a hierarchical structure as shown in the figure. Articles (secondary written articles, tertiary written articles,...) Written in response to the articles registered earlier (primary written articles) are registered in the lower layer. For example, writings such as answers, opinions, impressions, etc. for the upper article are displayed in the lower part. For example, article # 02 is a writing for article # 01.

  FIG. 3 is a diagram illustrating an example of the article data storage unit. In association with the article ID, a parent article ID, a writing date and time, a writer code, and article data (article contents) are stored in association with each other. The article ID is identification information of the article, and the parent article ID is the ID of the upper article, that is, the article ID of the reaction source. The writer code is an identification code of a person who writes an article from the user terminal 102. The article data is the contents of writing.

  FIG. 4 is a diagram showing the configuration of the article management server that constitutes the evaluation system. The electronic commerce server (article management server) 101 includes an article data storage unit 401, an article screen transmission unit 402, an article writing unit 403, an article value determination unit 404, and an article value table 405. The article screen transmission unit 402 is configured to transmit an article screen such as a product screen introducing products to the user terminal 102. The writer code of each article, the date and time of writing, and the article data are displayed according to the configuration of FIG.

  The article writing unit 403 receives article data (word-of-mouth content) designating a parent article ID from the user terminal 102, sets the logged-in user's code as a writer code, and sets the current date and time as a new date and time. Such information is written in association with the article ID.

  The article value determination unit 404 determines the value of the article based on the article ID for each article, the parent article ID, the writer code, and the article data, and stores the result in the article value table 405.

  FIG. 5 shows an article value table. The article value table 405 stores evaluation values in association with article IDs. In this example, the number of direct reactions and the total number of reactions are used as evaluation values. The number of direct responses is the number of articles written in the immediate lower layer (the number of articles that have reacted directly), and the total number of responses is the number of articles written in all the lower layers (the number of articles that have reacted directly and indirectly). For example, the number of direct reactions of article # 01 is two articles # 02 and article # 04, and the total number of reactions is six articles # 02 to article # 07.

  The direct reaction number calculation process by the article value determination unit 404 will be described. FIG. 6 is a diagram showing a direct reaction number calculation processing flow. The following processing is repeated for each article to be judged (S601). As a result, the number of reactions is obtained directly in order from article # 01 to article # 07. Hereinafter, a flow for obtaining the number of direct reactions to article # 01 will be described as an example.

  The following processing is repeated for each other article (S602). As a result, it is determined and counted whether article # 02 to article # 07 other than article # 01 are located in the immediately lower layer.

  It is determined whether the parent article ID of another article matches the determination target article ID (S603). If they match, 1 is added to the number of direct reactions (initial value 0 for each determination target article) (S604). . For example, as shown in FIG. 3, the parent article ID of the other article # 02 is # 01, which matches the determination target article ID # 01, so the number of direct reactions is counted up. Since the parent article ID of the other article # 03 is # 02, it is not counted up.

  When all other articles are processed (S605), the number of direct responses to the determination target article is written in the article value table 405 (S606).

  By processing this for all articles (S607), the number of direct reactions for all articles is calculated.

  Next, the total reaction number calculation process by the article value determination unit 404 will be described. FIG. 7 is a diagram showing a total reaction number calculation processing flow. The following processing is repeated for each article to be judged (S701). Similar to the above, the total number of responses is obtained in order from article # 01 to article # 07. Hereinafter, a flow for obtaining the total number of responses to the article # 01 will be described as an example.

  The following processing is repeated for each other article (S702). Thereby, it is determined and counted whether article # 02 to article # 07 other than article # 01 are located in any lower layer.

  As described above, it is determined whether the parent article ID of another article matches the determination target article ID (S703). If they match, 1 is added to the total number of reactions (initial value 0 for each determination target article). (S706). Article # 02 and article # 04 are thereby counted up.

  For example, in the case of article # 03, since there is no match at $ 703, it is determined whether the upper article has # 01 as the parent. For this purpose, an upper article corresponding to the parent article ID is specified (S704). If the article is # 03, the parent article is # 02, so the upper article is # 02. If the parent article ID of the upper article is a determination target article ID (S705), 1 is added to the total number of reactions (S706). Since the parent article of the article # 02 is # 01, the article # 03 is counted up at this stage.

  When the parent article ID of the upper article is not the determination target article ID, the upper article is further specified and the same determination is repeated. Then, if the parent article ID of the upper article is the root, there is no determination target article above the article, and the process ends without counting up.

  If this process is performed for all the other articles (S707), the total number of responses is obtained, so the total number of responses to the determination target article is written in the article value table 405 (S708).

  The process ends when all articles are determined (S709).

Embodiment 2. FIG.
In the above-described form, an example is shown in which the hierarchical relationship is managed by the parent article ID, but the hierarchical relationship can also be determined based on the citation description in the article data.

  FIG. 8 is a diagram illustrating an example of the article data storage unit according to the second embodiment. In this example, a citation description is seen in the article data (written content). Following the quote sign of “RT”, the writer code of the article quoted with “@” at the head is described. Furthermore, article data (written contents) of the cited article is described by delimiting with “:”. By this citation description, it is possible to determine the upper article.

  In the direct reaction number calculation process according to the present embodiment, instead of determining whether the parent article ID of another article matches the determination target article ID in S603 in FIG.

  FIG. 9 is a diagram showing a direct reaction number calculation processing flow according to the second embodiment. Instead of S603, the processing of S901 to S903 is performed. First, it is determined whether another article has a quotation mark (“RT”) (S901). If not, finish it. If there is, the citation writer code and the cited article data are specified (S902). That is, the writer code following “@” and the article data following “:” are extracted. When this matches the writer code and article data (excluding the quote description part) of the article to be judged (S903), 1 is directly added to the number of reactions (S605).

  For example, in the article data of article # 02, there is “RT”, the writer code following “@” is “SAM”, and the article data following “:” is “XX yogurt, refreshing best! Therefore, it can be seen that the article directly reacts to article # 01.

  Also, in the total reaction number calculation process, it is determined whether or not the cited description matches.

  FIG. 10 is a diagram showing a total reaction number calculation processing flow according to the second embodiment. Instead of S703 to S706 in FIG. 7, the processing of S1001 to S1007 is performed. S1001 to S1003 are the same as S901 to S903 described above.

  If the cited writer code and the cited article data do not match the writer code and the article data to be judged, in order to search further upward, sequentially match the cited writer code and the cited article data. The top article to be identified is identified (S1004). For example, the article # 02 that matches the article writer's quote writer code “JUN” and the article article data “Taste is refreshing. Refreshing” is the first top article.

  Then, it is determined whether or not there is a citation code in the upper article (S1005), and similarly, the citation writer code and the cited article data are specified (S1006). If there is no quote, finish here.

  When the identified writer code and citation article data match the writer code and article data (excluding the citation description part) of the article to be judged (S1007), 1 is added to the total number of reactions (S706). . Since the writer code “SAM” and the cited article data “XX yogurt, refreshing best!” Of the upper article # 03 of the article # 02 match the judgment target article # 01, the article # 02 is counted up. Is done.

  The quotation mark may be another code such as “QT”. In addition, as long as the format conforms to a predetermined rule, the cited writer code and the cited article data are exchanged, the cited article data is described after the citation code, and the citation writer code is described thereafter. It may be in a format other than the above example.

Embodiment 3 FIG.
In the above example, each written article is equivalent, but the evaluation can be changed by the writer. That is, it is possible to weight articles according to the writer.

  For this purpose, the electronic commerce server (article management server) 101 has an evaluation weight table. FIG. 11 is a diagram illustrating an example of the evaluation weight table. The evaluation weight corresponding to each writer code is set in advance.

  The article value table generated in this embodiment uses a total weight, which is the sum of evaluation weights, as a value index. FIG. 12 is a diagram illustrating an example of the article value table according to the third embodiment. For each article ID, a direct reaction weight total and a total reaction weight total are stored. The direct reaction weight sum is the sum of the evaluation weights of the writing articles in the immediately lower layer, and the total reaction weight sum is the sum of the evaluation weights of all the lower layer writing articles. Since the weight of the article # 07 written by “SUS” is 10 and the weight of the article # 03 written by “TOM” is 5, the evaluation of the articles # 05 and # 02 which are the reaction sources is increasing. .

  In this form of direct response weight total calculation processing, the following processing is performed instead of S604 “add 1 to the direct response number” and S606 “write the direct response number for the article to be determined” in FIG.

  FIG. 13 is a diagram showing a direct reaction weight total calculation output processing flow. Instead of S604, the writer of the other article whose parent article ID matches in S603 is specified (S1301), and the evaluation weight of the writer is acquired from the evaluation weight table (FIG. 11) (S1302). Then, an evaluation weight is added to the total direct reaction weight (initial value 0 for each article to be judged) (S1303).

  Also, instead of S606, the total direct reaction weight for the article to be determined is written in the article value table 405 (S1304).

  In the total reaction weight total calculation output processing, the following processing is performed instead of S706 “Add 1 to the total number of responses” and S708 “Write the total number of responses to the article to be judged” in FIG.

  FIG. 14 is a diagram showing a total reaction weight total calculation output processing flow. Instead of S706, the writer of another article in S703 or the upper article of S705 is specified (S1401), the evaluation weight of the writer is further specified (S1402), and the total reaction weight total (for each article to be judged) To the initial value 0) is added an evaluation weight (S1403).

  Also, instead of S708, the total reaction weight total for the article to be determined is written in the article value table 405 (S1404).

  The evaluation based on the weight can be applied not only to the first embodiment but also to the second embodiment.

  The above-described article evaluation system is not limited to an example of product evaluation in electronic commerce such as a virtual shopping mall, and is effective for other bulletin boards and Twitter. In particular, it is expected to be applied to CGM (Consumer Generated Media) and applied to word-of-mouth sites, Q & A communities, social networking services (SNS), blog portals, BBS portals, COI (Community Of Interest) sites, etc. It is possible.

  An electronic commerce server (article management server) 101 is a computer, and each element can execute processing by a program. Further, the program can be stored in a storage medium so that the computer can read the program from the storage medium.

101 e-commerce server (article management server)
102 user terminal 103 store terminal 401 article data storage unit 402 article screen transmission unit 403 article writing unit 404 article value determining unit 405 article value table

Claims (8)

The user terminal receives a write articles that are dependent on the write articles and the write article, an evaluation system write articles for evaluating a writing article their respective,
An article writing unit that accepts written articles from the user terminal;
An article data storage unit that stores the written article and stores a dependency relationship between the written articles;
With reference to the article data storage unit, for each of the write articles in a relationship of dependent-be dependent by all writes articles at the bottom of those 該書 included articles counted in a predetermined manner, those 該書 An article value determination unit that calculates the evaluation value of each included article;
An article value table for storing the evaluation value for each written article;
A system for evaluating written articles characterized by comprising: The article value determination unit, evaluation system according to claim 1, wherein calculating the total number of all said writing article as the evaluation value of the write articles. The article value determination unit, evaluation system according to claim 1, wherein calculating the number of unique users of the all writing articles as the evaluation value of this該書inclusive article.   The written article includes a citation description that can identify an upper written article;
  The article value determination unit sequentially identifies, for one written article, a written article that is higher than the other written articles based on the citation description included in the other written articles. The evaluation according to any one of claims 1 to 3, wherein all the written articles below the one written article are identified by determining whether the written article matches the one written article. system. The evaluation system further includes an evaluation weight table that stores evaluation weights for each user.
The article writing unit accepts user-specific information together with the written article, and the article data storage unit stores each written article in association with the user-specific information,
The article value determination unit refers to the article data storage unit for one written article, extracts an evaluation weight from the evaluation weight table, and uses the result of the count and the evaluation weight to evaluate the one written article evaluation system according to claim 1 in any one of 4, characterized in that to calculate the value. The user terminal receives a write articles that are dependent on the write articles and the write article, a method for evaluating writing articles for evaluating a writing article their respective,
A written article receiving step for receiving a written article from the user terminal; and
An article data storage step of storing the written article in the article data storage unit and storing a dependency relationship between the written articles;
With reference to the article data storage unit, for each of the write articles in a relationship of dependent-be dependent by all writes articles at the bottom of those 該書 included articles counted in a predetermined manner, those 該書 An article value determination step for calculating the evaluation value of each included article;
An evaluation value storage step for storing the evaluation value in the article value table for each written article;
A method for evaluating a written article characterized by comprising: The evaluation system with an article data storage unit and the article value table, a program for executing the evaluation process of writing articles that are dependent on the write articles and the writing article,
A written article receiving step for receiving a written article from the user terminal; and
An article data storage step of storing the written article in the article data storage unit and storing a dependency relationship between the written articles;
With reference to the article data storage unit, for each of the write articles in a relationship of dependent-be dependent by all writes articles at the bottom of those 該書 included articles counted in a predetermined manner, those 該書 An article value determination step for calculating the evaluation value of each included article;
An evaluation value storage step for storing the evaluation value in the article value table for each written article;
Write article evaluation program to execute. The evaluation system with an article data storage unit and the article value table, a program for executing the evaluation process of writing articles that are dependent on the write articles and the writing article,
A written article receiving step for receiving a written article from the user terminal; and
An article data storage step of storing the written article in the article data storage unit and storing a dependency relationship between the written articles;
With reference to the article data storage unit, for each of the write articles in a relationship of dependent-be dependent by all writes articles at the bottom of those 該書 included articles counted in a predetermined manner, those 該書 An article value determination step for calculating the evaluation value of each included article;
An evaluation value storage step for storing the evaluation value in the article value table for each written article;
The computer-readable recording medium which recorded the evaluation program of the written article for performing this.

Images