Pronto, is a bootstrapping based relation extraction system. Given a small set of example facts (e.g. "Paris is located in France.", "Bejing is located in China") it is able to come up with a large amount of facts of that kind (e.g. a huge lists of cities and where they are located). To do so, Pronto considers the way the relations are mentioned in the text (i.e. what stands around "Paris" and "France" when these words occurr together). We set up Pronto to discover relations in this test Wikipedia between page titles and hyperlinks. When it discovers new facts, Pronto generates the questions, you will find on the bottom of the pages. The answers you give to these questions are used as confirmation of the results before they are written into the wiki and as feedback to improve output in the next iteration of the system.

If you are interested in the topic, you are invited to read the following research paper:

Sebastian Blohm, Philipp Cimiano: Using the Web to Reduce Data Sparseness in Pattern-based Information ExtractionIn Proceedings of the 11th European Conference on Principles and Practice of Knowledge Discovery in Databases (PKDD). Springer , Warsaw, Poland, September 2007 (http://www.aifb.uni-karlsruhe.de/Publikationen/showPublikation?publ_id=1507).

This work was co-funded by the X-Media project (www.x-media-project.org) sponsored by the European Commission as part of the Information Society Technologies (IST) program under EC grant number IST-FP6- 026978.