Computing for automated web service discovery of educational services from local repositories using probabilistic matchmaking

Abstract

This research provides a new way to discover online educational services with focus on retrieving and analysing web service descriptions from WSDL documents through UDDI entries in addition to API documentation of various websites and sources. Key steps involved in the proposed algorithm are semantic enrichment, text processing, ranking, similarity computation and data retrieval. First of all, different APIs and WSDLs from UDDI entries were collected and studied. Textual refinement of web service descriptions is done using text preparation techniques such as tokenization, stop word elimination, lemmatization etc. Synonym expansion is an example of how lexical databases like WordNet can be used for semantic enrichment enhancing the contextual understanding of the service descriptions. Thereafter enhanced similarity measures are applied for determining semantic similarity between user queries and web service descriptions so that matching query intents with feature sets becomes simpler. Finally search results are ranked based on both authority (derived from PageRank algorithm) as well as semantic relevance metrics so as to suggest most trustworthy and appropriate contextually relevant services to users since. In online service discovery, the suggested technique has been found to be effective and useful through experimentation. This suggests that it can enhance the user friendliness and availability of web service ecosystems.