A Trust-Powered Technique to Facilitate Scientific Tool Discovery and Recommendation

Abstract

While the open science community engenders many similar scientific tools as services, how to differentiate them and help scientists select and reuse existing software services developed by peers remains a challenge. Most of the existing service discovery approaches focus on finding candidate services based on functional and non-functional requirements as well as historical usage analysis. Complementary to the existing methods, this paper proposes to leverage human trust to facilitate software service selection and recommendation. A trust model is presented that leverages the implicit human factor to help quantify the trustworthiness of candidate services. A hierarchical Knowledge-Social-Trust (KST) network model is established to extract hidden knowledge from various publication repositories (e.g., DBLP) and social networks (e.g., Twitter and DBLP). As a proof of concept, a prototyping service has been developed to help scientists evaluate and visualize trust of services. The performance factor is studied and experience is reported.