Self-adaptive mobile web service discovery framework for dynamic mobile environment

The advancement in mobile technologies has undoubtedly turned mobile web service (MWS) into a significant computing resource in a dynamic mobile environment (DME). The discovery is one of the critical stages in the MWS life cycle to identify the most relevant MWS for a particular task as per the request's context needs. While the traditional service discovery frameworks that assume the world is static with predetermined context are constrained in DME, the adaptive solutions show potential. Unfortunately, the effectiveness of these frameworks is plagued by three problems. Firstly, the coarse-grained MWS categorization approach that fails to deal with the proliferation of functionally similar MWS. Secondly, context models constricted by insufficient expressiveness and inadequate extensibility confound the difficulty in describing the DME, MWS, and the user’s MWS needs. Thirdly, matchmaking requires manual adjustment and disregard context information that triggers self-adaptation, leading to the ineffective and inaccurate discovery of relevant MWS. Therefore, to address these challenges, a self-adaptive MWS discovery framework for DME comprises an enhanced MWS categorization approach, an extensible meta-context ontology model, and a self-adaptive MWS matchmaker is proposed. In this research, the MWS categorization is achieved by extracting the goals and tags from the functional description of MWS and then subsuming k-means in the modified negative selection algorithm (M-NSA) to create categories that contain similar MWS. The designing of meta-context ontology is conducted using the lightweight unified process for ontology building (UPON-Lite) in collaboration with the feature-oriented domain analysis (FODA). The self-adaptive MWS matchmaking is achieved by enabling the self-adaptive matchmaker to learn MWS relevance using a Modified-Negative Selection Algorithm (M-NSA) and retrieve the most relevant MWS based on the current context of the discovery. The MWS categorization approach was evaluated, and its impact on the effectiveness of the framework is assessed. The meta-context ontology was evaluated using case studies, and its impact on the service relevance learning was assessed. The proposed framework was evaluated using a case study and the ProgrammableWeb dataset. It exhibits significant improvements in terms of binary relevance, graded relevance, and statistical significance, with the highest average precision value of 0.9167. This study demonstrates that the proposed framework is accurate and effective for service-based application designers and other MWS clients