Combining hyperlink-induced topic search and Bayesian approach for personalised manufacturing service recommendation

Abstract

The explosive growth of Web service technologies for integrating virtual enterprises has resulted in the serious information overload. Thus the users are facing increasing difficulty in selecting the correct manufacturing services from the vast amount provided or recommended by collaborative partners for service-oriented supply chain deployment. Therefore, in this paper, a novel approach is presented for recommending personalised manufacturing services by combining a Hyperlink-Induced Topic Search (HITS) algorithm and the Bayesian approach. The personalised service recommendation problem is modelled to determine the optimal manufacturing services that are most probably the best selections to user preferences for some known manufacturing services. Further, the Bayesian approach decomposes such a problem of posterior probability into two sub-problems: the prior probability of a manufacturing service hypothesis before considering any user preferences, and the conditional probability of considering the aggregated preferences for some known manufacturing services as evidences when the user actually wants an unknown most preferable manufacturing service. Next, the personalised HITS algorithm is adapted to the network of service-oriented supply chain to rank the authority scores of manufacturing services that determine the relative probabilities of service execution through personalised trust propagation. The experimental results show that the proposed method can produce more accurate manufacturing service recommendation results than the existing approaches.