Abstract : Text annotation is the procedure of identifying the semantically dominant words of a text segment and attaching them with conceptual content information in their context. In this paper, we propose novel methods for automatic annotation of text fragments with entities of Wikipedia, the largest knowledge base online, a process commonly known as Wikification aiming at resolving the semantics of synonymous and polysemous terms accurately. The cornerstone of our contribution is a novel iterative Wikification approach, converging at optimal annotations while balancing high accuracy with performance. Our first two methods can be fine-tuned through a machine-learning technique over large homogenous data sets. Our experimental evaluation resulted in remarkable improvement over state-of-the-art Wikification approaches.
https://hal.inria.fr/hal-01391352 Contributor : Hal IfipConnect in order to contact the contributor Submitted on : Thursday, November 3, 2016 - 11:05:40 AM Last modification on : Thursday, March 5, 2020 - 5:41:17 PM Long-term archiving on: : Saturday, February 4, 2017 - 1:03:53 PM
Christos Makris, Michael Angelos Simos. Novel Techniques for Text Annotation with Wikipedia Entities. 10th IFIP International Conference on Artificial Intelligence Applications and Innovations (AIAI), Sep 2014, Rhodes, Greece. pp.508-518, ⟨10.1007/978-3-662-44654-6_50⟩. ⟨hal-01391352⟩