Files in this item



application/pdfparse-adapt-wikipedia.pdf (96kB)
(no description provided)PDF


Title:A Wikipedia based Algorithm for Online Adaptation of a Syntactic Parser
Author(s):Kundu, Gourab
Subject(s):domain adaptation, wikifier, syntactic parsing
Abstract:Adaptation of models to a new domain is important for many natural language tasks. Because without adaptation, NLP tools trained on news domain achieve sub par results on other domains. In many practical scenarios, the identity of the domain of the test set is unknown. For this difficult but important setting, we propose a novel method of adaptation using entity disambiguation systems to Wikipedia. We get significant improvements for adapting a syntactic parser trained on news domain to biomedical domain.
Issue Date:2013-11-14
Genre:Technical Report
Publication Status:unpublished
Date Available in IDEALS:2013-11-14

This item appears in the following Collection(s)

Item Statistics