Using Sequential Pattern Mining and Interactive Recommendation to Assist Pipe-like Mashup Development

Xinyi Liu, Hailong Sun, Hanxiong Wu, Richong Zhang, Xudong Liu

Abstract: Mashups represent a typical type of service oriented applications targeting end-user development. However, due to lack of development expertise, end-users usually find it hard to build a mashup. Therefore, it is of paramount importance to provide effective assistance to achieve efficient mashup development. In this work, we aim at leveraging the expertise that can be mined from voluminous mashups on Internet to recommend appropriate mashup modules and their composition patterns to facilitate pipe-like mashup development. First, we crawl all the mashups available in Yahoo!Pipes and extract the meta-data of each mashup from original JSON data. Second, we use GSP (Generalized Sequential Pattern) algorithm to mine the frequent composition pattern of mashup modules, and design an interactive recommendation algorithm to assist mashup development. Third, we implement a system prototype based on the proposed method and evaluate its effectiveness with 848 Yahoo! mashups through cross-validation.

External IDs:dblp:conf/sose/LiuSWZL14