Text mining

Wikileaks publiceert 92.000 documenten en je moet in één dag de essentie en relevantie eruit halen. Wat nu? En hoe ontdek je patronen in documenten en toespraken? Leer ‘text minen’ en je werk wordt zoveel simpeler! Dé relevatie op onderzoeksgebied van de laatste maanden, en dus nu op e VVOJ-conferentie.
Text analytics or text mining is the branch of data mining that deals specifically with discovering patterns, information and knowledge in textual data. Text, written language, can be seen as a strong encryption method of information – it takes human intelligence (e.g. knowledge of a particular language and world knowledge) and effort to understand what information a text conveys. Yet, increasinly advanced methods are available from the fields of language technology and information retrieval that aim to alleviate the effort and help people find what they are looking for. In this contribution I briefly review the state of the art of these fields, and critically review the capacity of developed tools to support investigative journalism. I focus on three technologies: (1) recommender systems, of which the goal is to have information find the searcher, rather than the other way around; (2) event detection, to discover (significant) events in potentially vast amounts of free text;
(3) social network mining through text, to discover relations (e.g. friend or foe) between people.
Met John Bones, Antal van den Bosch
Datum: 20 november 2010
Tijd: 10:45-12:00 uur
* LET OP: Voertaal Engels