Output details
11 - Computer Science and Informatics
Open University
Feature LDA: a supervised topic model for automatic detection of web API documentations from the web
<16>This publication presents the first solution to successfully discover Web APIs on the open Web. It extends Latent Dirichlet Allocation with semi-supervised features leading to a superior performance over top Machine Learning classifiers. The algorithm established in this paper is a fundamental extension to iServe, our flagship services registry that was core to the EU projects SOA4All and NoTube. This work is now central to the crawler feeding the registries of two EU projects: VPH Share (Rod Hose <d.r.hose@sheffield.ac.uk>) and COMPOSE (Benjamin Mandler <MANDLER@il.ibm.com>). It is also the basis for an international collaboration with Barcelona Supercomputing Centre (Yolanda Becerra <yolandab@ac.upc.edu>).