Abstract:
Recommender systems are thought to be one of the most successful and useful applications in the domain of machine learning and artificial intelligence. These systems save users the hassle of going through a range of products to select the most suitable for their needs. Rather, they employe the use of the historical data about the users and their experiences of the available items / products to recommend the products most suited for them statistical [1]. With quick advancement in web technology, quantity of web-services offered on the internet is ever growing. This makes it a challenging task to choose a web-service fit to the objective needs of a user [2]. To cater to this aspect, numerous research efforts have proposed several methods of web-service recommendation in which multiple QoS related attributes play a role such as response time, throughput and web-service delivery [3]. Other than these attributes, there are some hidden attributes that affect the QoS prediction such as the context of users and web-services for example: user trustworthiness and web-services reputation [4]. The intent of proposed research work is to develop and design a web-service recommendation framework which employs the QoS prediction mechanism based on the context of web-services and users calculated on the basis of the historical QoS values. The model, namely S-RAP, uses these contextual factors to predict the QoS of a web-service that would be experienced by a user who has not invoked it before. The work emphasizes on extracting and employing the relational data in the users context and web-services context both, for proposing a comprehensive model to predict the QoS values for a user. In the web-services context, the terminology of web-services relevance is introduced with a computation mechanism being used to calculate the degree of relevance between two web-services. The QoS values are predicted in both users and web-services contexts separately, which are then combined to generate the active prediction which is the final output of the S-RAP. The experimentation and result analysis have proved that S-RAP generates satisfactory results.
