Service Package Recommendation for Mashup Creation via Mashup Textual Description Mining

Abstract

Mashup is a developer-centric technique which allows developers to compose existing Web services together to create innovative or consolidated web applications. However, the rapid growth in the number of services and the myriad of functionally similar services make it difficult for developers to select appropriate ones to develop new applications. Therefore, it is vital to recommend a set of suitable services for mashup creation based on functionalities of services and their relationships. To this end, we propose a service package recommendation approach for mashup development, which is based on mashup textual description mining to discover semantic relationships among services. Specifically, discourse analysis of computational linguistics is utilized to uncover the structures underneath mashups' functional specifications, then the semantic relationships between services can be learned from their appearances and the constructed structures in mashup specifications. Accordingly, we are able to recommend a package of services that can be used together with high compatibility for a new mashup to be developed. We evaluate our approach on a real-world dataset. Experimental results show that our approach achieves higher accuracy and outperforms other comparative ones.