Generation of Overview-Oriented Search Results Using Ordered Structure of Word Occurrences in Technical Blogs
Masaru Hakii, Daisuke Kitayama
Proceedings of The International MultiConference of Engineers and Computer Scientists 2021(IMECS2021),pp.162–167,Hong Kong (online),2021年10月,査読有
論文PDF
概要
For a given search topic, a user without prior knowledge may have difficulty fully understanding the topic in the search results. In addition, when a user wants to gain a deeper knowledge of the topic, the user can not create appropriate search queries with the existing search engines because the user does not know the surrounding topics. To solve this problem, we propose a method for presenting search results to help users grasp the full scope of a topic. Specifically, we present the user with search results, based on the tree structure and using words with prerequisite knowledge relations from the set of search results. By presenting search results using prerequisite knowledge relationships the user can traverse the tree structure when they want to know more about the topic. Thus, the user can grasp the full topic. This paper proposes a method of presentation that searches for results to help provide the full scope of a topic to the user. Proposed method is that obtaining a set of articles from search results and create a tree structure of words from the set of articles and generating overview oriented search results on the basis of a tree structure. As a result of the evaluation experiment, the precision of the detailed relation, meaning the depth direction of the tree was 70.5% and the Spearman’s rank correlation coefficient of the order of appearance, meaning the width direction of the tree, was 0.41. This result indicates that it may help the user fully understanding, but there is a problem with the extraction of the title sentence.