Clustering Mashups by Integrating Structural and Semantic Similarities Using Fuzzy AHP

Abstract

Mashup technology has become a promising way to develop and deliver applications on the web. Automatically organizing Mashups into functionally similar clusters helps improve the performance of Mashup discovery. Although there are many approaches aiming to cluster Mashups, they solely focus on utilizing semantic similarities to guide the Mashup clustering process and are unable to utilize both the structural and semantic information in Mashup profiles. In this paper, a novel approach to cluster Mashups into groups is proposed, which integrates structural similarity and semantic similarity using fuzzy AHP (fuzzy analytic hierarchy process). The structural similarity is computed from usage histories between Mashups and Web APIs using SimRank algorithm. The semantic similarity is computed from the descriptions and tags of Mashups using LDA (latent dirichlet allocation). A clustering algorithm based on the genetic algorithm is employed to cluster Mashups. Comprehensive experiments are performed on a real data set collected from ProgrammableWeb. The results show the effectiveness of the approach when compared with two kinds of conventional approaches.