The first application to be featured on the Labs website is EvidenceFinder, an application that surfaces “facts” based on a user’s search terms from full text articles.
The user is presented with a list of questions based on the search terms. Selecting a question refines the search results, and displays the specific evidence from the full text pertinent to that question. All evidential sentences from a given article can be also be browsed via the “Extracts from full text” tab. EvidenceFinder is built by NaCTeM, University of Manchester.
A good example of the power of this tool can be seen by searching for ‘malaria’ and selecting the “What causes Malaria?” question.
UKPMC Labs also includes Whatizit, built by the EBI. Whatizit is used in the main UKPMC web service to identify biological terms of interest.
There is now a link from the UKPMC homepage to Labs. We welcome Feedback on Labs-featured applications. Please tell us what you think!
Evidence finder needs work, it is not ready for public release. You need to at least work with parsed sentences and cluster results. "What interacts with BRCA1?" gives a huge list of citations and not summary. The literature contains many "XXX interacts with BRCA1", "XXX binds…", "XXX co-immunoprecipates with…" or "BRCA1 interacts with YYY" sentences. You need to have a depth of parsing and rearrangement sufficient that you can recognize these as similar and cluster them.