Towards Selecting and Recommending Online Software Services by Evaluating External Attributes

Selecting an online software service for a given set of requirements can be based on the quality of results (i.e., relative ranking of the services) and associated recommendations (i.e., applicability of the suggested services). Prevalent approaches for product-based selection (e.g., the ones used by Amazon) and recommendations, such as Content-based Filtering (CBF) and Collaborative Filtering (CLF) do not typically consider information about products beyond primitive attribute-value pairs. Compared to a tangible physical product, a reusable and updatable software service cannot be effectively described using only a set of strict attribute-value pairs or using a sparse matrix of user-product relationship. This is because a software service has various programmatic, functional, and non-functional properties which potentially could also be dynamic in nature. Due to these challenges, it is not sufficient to apply product-based ranking and recommendation techniques to software services available from a marketplace. This research proposes an approach for better selection and recommendation of software services that enhances both CBF and CLF algorithms, using external reviews.