A Rapid Mining Model for Extracting Sparse Distribution Association Semantic Link from Large-scale Web Resources

Abstract

Association semantic link (ASL) can provide theoretical support for many web intelligent activities. However, when we extract the keywords-level association semantic link (k-ASL), some sparse distribution k-ASL are easily discarded. To solve this problem, this paper proposes a rapid mining model for extracting sparse distribution k-ASL from large-scale web resources. First, the time validity for three types of k-ASL is analysed to clear their semantic characteristics. Second, three existing problems for mining sparse distribution k-ASL are presented to analyse why this kind of k-ASL is easily discarded. After that, we present the rapid mining theoretical foundation for mining sparse distribution k-ASL. Furthermore, the rapid mining model for extracting sparse distribution k-ASL is proposed, which is based on the presented theory and set computation such as 'difference computation', 'union computation'. At last, the evaluation method is presented and the correctness of the proposed model is validated by experiments.