KR101448134B1 - an blog prestige ranking method based on weighted indexing of terms - Google Patents

Abstract

The present invention relates to a term weighted index-based blog authority ranking method for further improving the quality of search results in a blog search system using term weighted indexes using structural characteristics of blog posts on the Internet. The present invention improves the quality of the blog search result by increasing the fitness of the blog post ranking and terminology by giving weight to important terms using the structural characteristics of the blog in the blog post ranking. When judging the rank of a blog post, the score of each term reflecting the relevance of the blog post to each term is indexed to the blog search system database, not the authority score of the blog post as a whole. The final ranking score reflected in the search result can be retrieved from the database and provided to the user, with a result list of posts considered up to date among the blogs having a high degree of relevance to the search word input by the user. The blog post itself can be expected to have high relevance of the search results while avoiding the problem that the results that are famous and authoritative but do not match the information desired by the user are ranked at the top of the search results.

Description

In this paper, we propose a blog ranking method based on weighted indexing of terms.

The present invention relates to a blog authority ranking method based on a term weighted index, and more particularly to a term weighted index for further improving a search result quality of a blog search system through a term weighted index using structural characteristics of a blog post on the Internet Based blog authority ranking method.

In general, a blog is a web site that records information about an individual's thoughts or interests on the Internet in a journal form, which allows them to share opinions and information with others. Through blogs, you can get a glimpse of people's interests and thoughts around the world, and see how they react to your products or services. In addition, popular blogs are visited by hundreds of thousands of visitors a day, so the power of the blog is as much as the existing media. In addition, thanks to the social network service like Twitter, the spread of information in blog posts has become more powerful. For this reason, the blog search field can make a good contribution to the study of the flow or impact of marketing, media, and Internet information.

A method for ranking blogs and ranking authority (hereinafter referred to as "blog ranking method") has been studied until recently. Especially, the characteristics of blogs with different characteristics and structure from general web documents have been studied variously. Unlike regular web pages, blogs are not effective enough to use links between pages as a key element in ranking, so link - based authority ranking such as PageRank is not effective.

Due to these problems, a blog ranking method has been developed that analyzes link similarity and accessibility and applies link ranking based on links. In addition, the blog ranking method that has been further developed here, considering the frequency with which blog posts are updated, and the number of comments on each article, is also considered. In the case of EigenRumor search and further improvement of Egocentric search among these conventional methods, the amount of trackbacks connected between blog posts and user comments or blog posts is used as an eigenvector, The user 's evaluation of each post of the blog was quantified and applied to the blog authority rankings.

EigenRumor Blog Ranking Method Focusing on the community among blogs Each blog site has an Authority score that quantifies how many popular blog articles are in the blog community and a hub (Hub score), and calculate the Reputation score by quantifying how many evaluations were received from the bloggers with high hub values for each blog post in each blog site.

FIG. 1 is a diagram illustrating a method of determining a blog ranking according to a conventional method such as an EigenRumor blog ranking method. As shown in FIG. 1, a comment of a blog post, a trackback, By summing links, measure the authority of the blog post and apply it to the ranking.

The blog search using the conventional blog ranking determination method frequently has a problem that the quality of the search result is lowered because the authority of the blog posts is high but the relevance to the search word is low. In the keyword-based search, a good matching between the search word and the search result is a good search system. Therefore, in addition to the authority of the blog post, the suitability of the blog post and the search word matching the user's desired result should also be reflected in the ranking determination method. However, the conventional method of determining the blog ranking is not considered.

Even if the user's search term is not an important part of the contents of the blog post, if the authority of the blog post itself is high and appears at the top of the search result, it is obvious that the user will not judge it as the desired result. In addition, considering the frequency of each term in the blog post, the quality of the search result is not guaranteed even if the importance of the index word is reflected. Thus, a more effective method reflecting the importance of the term in the blog post is required.

SUMMARY OF THE INVENTION The present invention has been made in view of the above-described needs, and a method and system for determining a blog ranking by a method of weighting terms using structural characteristics of a blog for an effective blog search system. By analyzing the structural characteristics of the blog, it is possible to determine the importance of the terms generated in the blog post, thereby determining the blog post ranking of each term and reflecting the result in the blog search result, The purpose of this study is to provide a blog ranking system based on the weighted index, which can improve the quality of blog search results by taking into account compatibility with terms closely related to the index.

According to an aspect of the present invention,

A pre-processing step of searching for blogs activated on the Internet through the web robot module, analyzing the structure of posts of each blog, and collecting and normalizing data necessary for blog ranking determination through the analyzed structure;

In the blog post data analyzed through the preprocessing step, weights are applied to the terms in the title, tag part and main text of the post, for each structure, to calculate weighted important term ranking score of the blog post Determining;

A weighted important term score in the ranking determination step and a general term score to which the frequency is applied for the remaining general terms in the blog post body are added to the total term scores for the post to calculate a term weighted ranking score, Storing a database in a database; And,

And a search step of calculating a blog list based on the term weighting ranking score of the blog important term corresponding to the search word requested by the user query server from the information stored in the database.

The present invention improves the quality of the blog search result by increasing the fitness of the blog post ranking and terminology by giving weight to important terms using the structural characteristics of the blog in the blog post ranking. When judging the rank of a blog post, the score of each term reflecting the relevance of the blog post to each term is indexed to the blog search system database, not the authority score of the blog post as a whole. The final ranking score reflected in the search result can be retrieved from the database and provided to the user, with a result list of posts considered up to date among the blogs having a high degree of relevance to the search word input by the user. The blog post itself can be expected to have high relevance of the search results while avoiding the problem that the results that are famous and authoritative but do not match the information desired by the user are ranked at the top of the search results.

Brief Description of the Drawings Fig. 1 is a diagram showing a calculation of authority scores in a conventional method for determining a blog ranking.
2 shows a structure of a general blog post.
3 is a diagram for explaining a term weighted blog post ranking method according to the present invention.
4 is a schematic diagram of a blog search system to which the term weighted blog post ranking method according to the present invention is applied.
5 is a diagram illustrating an example of a relational database schema of a blog search system according to the present invention.

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

FIG. 2 is a cross-sectional view showing the structure of a general blog post. A blog usually has a plurality of blog posts. The blog post data includes data such as title, category, and date, body data, . One post corresponds to one document in traditional IR.

The method for determining a blog ranking according to the present invention performs a preprocessing process for performing a blog and blog post data collection and analysis process. In the preprocessing process, blogs activated on the Internet are searched through the web robot module of the search system to analyze the structure of posts of each blog, and data necessary for blog ranking determination is collected and normalized through the analyzed structure, .

In the Ranking Judgment module, the processed data is used as a parameter to calculate a ranking score for each term, and the score is transmitted to the blog search system. The blog search system stores the score in the database according to the relation between the corresponding blog post and the term This is used for outputting a search result that a user searches in a search query server of the search system.

As shown in FIG. 3, in the blog posting data analyzed by the web robot module of the search system, the title, tag, We calculate the important term ranking score of the blog post by applying weight to each structure.

The important term ranking score has the highest weight in the title, followed by a differentiation in the order of tag and anchor text, respectively. Also, the frequency of general terms in the post body is calculated, and the term weighting ranking score is added to the important term ranking score to store the term weighting ranking score for each term of the blog post.

In other words, for important vocabulary extracted through the structural characteristics of the post, a weighted vocabulary score that is weighted differentially according to each structure is calculated, frequency is calculated for the remaining general terms of the text, The term score is added to the total term score for the post and the term weighted ranking score is stored in the database for each term in the blog post. The database indexes not only the relation between the post and the general term, but also the relation with the important term.

Here, the frequency of each term in the post body text is set to an upper limit to keep the reflection on the ranking lower than the score of the weighted vocabulary. In order to reflect the freshness of the post in the final ranking, the date of the creation of each post is also collected and indexed.

Since important terms representing blog posts can correspond to user queries with high relevance to blog posts, it is possible to increase the importance scores and post relevance ranking scores by assigning a higher weight than general terms. These ranking scores are reflected in the search results of the search system.

Categories in the blog structure are used to classify blog posts by bloggers according to themes and materials, so the terms used in categories can also be important terms in blog posts. However, because of the limitations of the generic term that is set in advance, the key term weight is set lower than the weight of the post title or tag. The post title is most relevant to the subject of the post content, so it is given a high weight. In addition, a tag that a user inserts in addition to the subject, material, classification, etc. of the post body is given a lower weight than the title. In the case of categories and tags, the filtering process is additionally needed because the number of cases that are not related to the post body is considerably large according to the usage pattern of the blogger. In the text, structurally important terms can be extracted from the subtitle or anchor text (AnchorText). The anchor text is the description part of the link connection for quoting an external web document in the body. In addition, weights are given as important terms, and the terms extracted from bold, underlined, and highlighted colors are also weighted differentially.

As shown in FIG. 4, the blog search system 100 to which the method of the present invention is applied includes a web robot 110 for collecting necessary data in a blog, a term weighting unit for ranking the collected data according to terms and blog posts, A database 130 for indexing and storing ranking scores and blog information, a user query server 140 for searching the database 130 for search results according to a user's request and outputting the search results in order of ranking, .

Basic information (title, date, blogger name) of the blog post is stored in the database 130, a term list corresponding to all the terms existing in the blog post is also stored, and authority scores and term scores Respectively. In addition, all the important terms of a post belonging to the same blog are indexed separately for each blog.

Figure 5 is an illustration of an overall schema when indexing such databases to a relational database system. The information stored in the database is primarily calculated on the basis of the ranking score of the blog important term associated with the term corresponding to the query term requested by the user query server, and the ranking score of the blog posts belonging to the blog, The scores for the final result are also summed to calculate the ranking score. The user query server fetches the blog post information corresponding to the blog post list in the order of the highest ranking ranking score from the database and outputs it as a search result.

In addition, in another embodiment of the present invention, the database 130 provides the latest query factor index obtained by converting the blog generation date record information into the normalized value, to the user query server 140, and the user query server 140 The final ranking score is calculated as an important term score corresponding to the search term and the latestness factor and is provided to the user query server. The user query server 140 firstly calculates a critical term ranking score and a blog list having a high degree of renewal factor among the blogs in the database 130 related to the important term index related to the search term requested by the user, The ranking list of the posts related to the search word actually input by the user is calculated secondarily and the result is finally divided into the first and second stages.

Therefore, in the blog authority ranking method based on the term weighted index according to the present invention and the system for performing the same, a result list of posts considered up to date among the blogs having a high degree of relevance to the search word input by the user is fetched from the database, It is possible to calculate the final blog ranking score that can be provided so that the blog post itself can avoid the problem that the results that are famous and high authority but do not match the information desired by the user are ranked at the top of the search result, Can be expected.

Claims (3)

I) a pre-processing step of searching blogs activated on the Internet through a web robot module, analyzing the structure of posts of each blog, and collecting and normalizing data necessary for blog ranking determination through the analyzed structure;
Ii) Ranking for calculating weighted critical term ranking scores of blog posts by applying weights to terms existing in the title, tag portion and important phrases in the body of the blog post data analyzed through the preprocessing step Determining;
Iii) summing the weighted critical term score in the ranking determination step and the general term score to which the frequency is applied for the remaining general terms in the blog post body, to the total term score for the post, and calculating the term weighted ranking score Storing a database for each term of the post; And,
And iv) a search step of calculating a blog list based on the term weighting ranking score of a blog important term corresponding to a search word requested by the user query server from the information stored in the database,
In the database storing step, the weighted important term score has a highest title, sequentially assigned a tag and an anchor text in order, and assigns a different weight to terms extracted from bold, underlined, and highlighted colors, Wherein the general term score applied is a lower limit than the weighted critical term score. delete 2. The method according to claim 1, wherein the searching step comprises the step of calculating a final ranking score using the term weighting ranking score and the blog generation date record information into a normalization value, And a ranking list of posts corresponding to a search word actually input by the user in the blog list is secondarily calculated and provided to provide a blog authority ranking method based on the weighted index.

Images