Pascal Challenge will provide a rigorous evaluation of various machine learning techniques for extracting implicit relations from documents.
Given a standardised corpus of annotated and pre-processed documents, the participants will be expected to perform a number of tasks; each examining a different aspect of the learning process.
The basic task will be to learn the textual patterns necessary to extract the annotated information. In addition, subtasks will look at the effect of limiting the availability of training data, the ability to select the most appropriate training data (i.e. active learning) and the use of unannotated data to aid learning.
The formal evaluation will take place from Monday November 15th to Monday 22nd November.