Abstract
As service-oriented computing (SOC) technologies gradually mature, developing service-based systems (such as mashups) has become increasingly popular in recent years. Faced with the rapidly increasing number of Web services, recommending appropriate component services for developers on demand is a vital issue in the development of mashups. In particular, since a new mashup to develop contains no component services, it is a new “user” to a service recommender system. To address this new “user” cold-start problem, we propose a multiplex interaction-oriented service recommendation approach, named MISR, which incorporates three types of interactions between services and mashups into a deep neural network. In this article, we utilize the powerful representation learning abilities provided by deep learning to extract hidden structures and features from various types of interactions between mashups and services. Experiments conducted on a real-world dataset from ProgrammableWeb show that MISR outperforms several state-of-the-art approaches regarding commonly used evaluation metrics.
Highlights
W ITH the maturity of service-oriented computing (SOC), the development paradigm of software systems is shifting from component-based software development (CBSD) to service-oriented software development (SOSD)
2) We propose a novel multiplex interaction-oriented service recommendation approach, called MISR, by integrating three types of interactions between services and mashups into a deep neural network (DNN). 3) Experiments conducted on a real-world dataset crawled from the website ProgrammableWeb1 demonstrate that the proposed approach outperforms several state-of-the-art approaches regarding recommendation performance
After getting the feature vectors of mashup m, vseqm and vsetm, we calculate its content similarity to the existing mashups and obtain the neighbor mashup set N M. The complexity of this processing is O(P (H + D) + P log K), where H is the dimension of vseqm, D is the dimension of vsetm as well as that of word embeddings, P is the number of potential neighbor mashups, K is the size of N M, and P log K is the cost of searching top K values from a list that has P elements
Summary
W ITH the maturity of service-oriented computing (SOC), the development paradigm of software systems is shifting from component-based software development (CBSD) to service-oriented software development (SOSD). Xiong et al [13] presented a hybrid recommendation approach by integrating CF and deep learning for NLP Since most of these hybrid approaches leverage the interaction history between mashups and web application programming interfaces (APIs), they perform well in the normal recommendation process for services. From the perspective of the recommender system, the new mashup to be built does not contain any component service, which could be regarded as a new “user” to the recommender system In such a scenario, the traditional CF-based approach does not work well because no usage history is available to the new mashup. We propose a multiplex interaction-oriented service recommendation approach (referred to as MISR) to address the cold-start problem of developing new mashups.
Sign-in/Register to view highlights & summary Sign-in/Register to access full text options 
Free) 
Talk to us
Join us for a 30 min session where you can share your feedback and ask us any queries you have
Schedule a call
