Mining the Meaning of Web Services

Web Services are becoming a routine mechanism for accessing online resources. However finding them and subsequently using them is difficult. They have very different approaches to implementation. They need constant monitoring for uptimes and interface changes. Most importantly of all, trying to figure out what a service does and how to use it without direct negotiation with the original author/provider is much harder than it seems it should be. The BioCatalogue (http://www.biocatalogue.org) provides a common interface for registering, finding, and monitoring bio-web services. It is a joint development by the University of Manchester and the European Bioinformatics Institute, and currently has over 1200 services registered.

A serious problem is curating those services – describing them well enough to actually make them useful and keeping those descriptions up to date. Doing this manually is best but expensive. Crowd-sourcing the effort – the community contribute up to date descriptions and comments – is challenging. We need smarter automated or semi-automated tools to figure out what services do and we need to mine the evolving profiles and content of the BioCatalogue, and its myExperiment.org sister web site of computational workflows that use the services - to better recommend services.
This PhD focuses on smart web service curation using a mix of techniques, from text mining and profile analysis to social collaboration.