Relation Extraction in Vietnamese Text Using Conditional Random Fields

Rathany Chan Sam, Huong Thanh Le, Thuy Thanh Nguyen, The Minh Trinh

Abstract: Relation extraction is the task of finding semantic relations between entities from text. This paper presents our approach to relation extraction for Vietnamese text using Conditional Random Field. The features used in the system are words, part-of-speech tag, entity type, type of other entities in the sentence, entity’s index and contextual information. In order to evaluate the effect of the contextual information to the system performance, different window sizes have been tested in our experiments. It shown that the system performance is affected by the window size, but it is not directly proportional to the F-score of the system. Our future work includes: (i) testing the system with a larger corpus in order to get a more accurate evaluation of the system; (ii) investigating other features used in the CRF algorithm to increase the system performance; and (iii) researching methods to extract relations outside the sentence’s scope.