On the Social Aspects of Personalized Ranking for Web Services

Abstract

Ranking is an important step once automated discovery in Web Services is enabled; it allows for automated selection of the best matched service, out of the discovered ones. However, automated selection of the best matched service is not as simple as it may look like. Different service consumers may have different preferences to select the service providers, which may even depend upon their past interactions. Various approaches have been proposed that allow ranking of services based on different functional and non-functional aspects. However, we believe that the selection of services based on the analysis of the past interactions of service consumers or their social-network could be another effective way to rank the services for the benefit of service consumers. In this paper, we present a community-aware personalized approach for recommending and ranking Web Services for a service consumer. It is based on analysis of historical interactions among service consumers and service providers. We perform analysis and mining on the log information of service consumers and service providers, model their past interactions as social network, apply standard social-network analysis techniques, and use this information in ranking Web Services.

Handling incomplete data using semantic logging based Social Network Analysis Hexagon for effective application monitoring and management

“…Social network between Functional Properties FP and Non Functional Properties NFP can be denoted as SN (FP, NFP). In this paper, we will take two triangles of the hexagon and compute any missing values using the other information available, which is an extension of our earlier proposed solution in [15]. This computation also reveals hidden and non-obvious correlations among different elements of logs.…”

“…The Semantic Web Services based solutions that we discussed made promising contributions to formalize web services descriptions but do not attempt to formalize and utilize log and event descriptions. None of the approaches, except our previous related work [15], attempted to handle missing values and incomplete data which is very crucial for monitoring and managing applications. Having missing values and incomplete data in execution is a common problem due to invalid inputs from users, faults and errors that may occur during application execution.…”

“…The hexagon has several triangles between different elements of logs. We have extended and used our technique that we initially proposed in [15] to compute missing values, find out incomplete data and reveal hidden and non-obvious correlations between different elements of logs with possible problems in application execution. The Social Network Analysis Hexagon gives a generic model to represent our solution for using these connections between the elements and compute any missing values and incomplete data.…”

“…Another relevant approach is our own previous work on social aspects of personalized ranking for Web Services [15]. It is based on using light-weight semantics for modeling interactions of Service Consumers (SC), Service Providers (SP) and Non Functional Properties (NFP).…”

“…It based on our previous work on social aspects of personalized ranking for Web Services [15]. Our proposed Semantic Logging approach attempts to formalize the information from application execution but faces hindrance when some of the data is found missing from the application.…”