Extraction of Complementary Topics based on Phrase Importance and Co-occurrence in Technical Blogs
Masaru Hakii, Daisuke Kitayama
The 24th International Conference on Information Integration and Web Intelligence (iiWAS2022), pp.435-440, 2022年11月, 査読有
論文PDF
概要
When acquiring knowledge in a certain field, it is important to obtain comprehensive information. However, it is difficult for users to extract missing information in unknown fields by themselves. Therefore, this study proposes a subtopic extraction method that can supplement missing information step-wise by presenting the user with important unknown phrases related to the topic the user is currently learning about by performing Web searches; the phrases are such that they are highly relevant to the user’s browsing history. We then conducted small-scale experiments and found that topic extraction using phrase co-occurrence as a reference for the next search keyword to be entered is effective compared to conventional topic extraction using LexRank.