There is quite some time that we have not reported any news but this does not mean that work on iServe has stopped. Far from that. In the last period we have been focussing on fixing existing parts of the infrastructure that were not as polished as one would desire but also on pursuing new directions. One such direction that we are particularly keen on at the moment concerns the identification of Web APIs.

Indeed, as you may be aware locating Web APIs is nowadays not very well supported on the Web. The best option is perhaps ProgrammableWeb but it sometimes is out of date and its search facilities, although useful, are not particularly advanced. We have recently been working towards improving this situation with a first step devoted to actually locating Web APIs out there. Although we are far from having finished the work, we already have some promising results with a new solution we have devised that has allowed us to reach an accuracy of about 80% while trying to discern between a normal Web page and one documenting a Web API. More exciting features are on the way and we shall soon provide access to some of them… until then, find at [1] some details on this research and see below the slides of the presentation we gave at the AAAI symposium last March.

[1] Pedrinaci, Carlos; Liu, Dong; Lin, Chenghua and Domingue, John (2012). Harnessing the crowds for automating the identification of Web APIs. In: AAAI Spring Symposium 2012, 26-28 March 2012, Stanford, California, USA.