Yet Another Framework for Tweet Entity Linking (YAFTEL)
Authors:
- Hyoil HanHHHyoil Han
- Pattamon ViriyothaiPVPattamon Viriyothai
- SeungJin LimSLSeungJin Lim
- Dominik LameterDLDominik Lameter
- Benjamin MussellBMBenjamin Mussell
- 1 March 2019
- conference paper
- Published by Institute of Electrical and Electronics Engineers (IEEE)
- p. 258-263
Abstract
Social media data has become an important source of information. Linking a named entity mentioned in tweets with a knowledge base is called Named Entity Linking. By linking a tweet with a knowledge base, it clarifies the implication of the named entity. The drawback of tweet entity linking is that Twitter limits a status to 280 characters in a tweet and this fact leaves a tweet with less semantics compared with a lengthy document. The proposed framework 'Yet Another Framework for Tweet Entity Linking (YAFTEL)' incorporates the degree of direct references between candidate entities into the traditional approach adopted by the KAURI system. In our approach, two candidate entities get a higher score for entity mapping when at least one of them references the other. Our experiment shows that YAFTEL maps entity mentions to Wikipedia entities more accurately than KAURI when candidate entities reference mutually or in one-way.Keywords
This publication has 8 references indexed in Scilit:
- Recency-based candidate selection for efficient entity linking2017
- Entity Disambiguation with Linkless Knowledge Bases2016
- Microblog Entity Linking with Social Temporal Context2015
- Linking named entities in Tweets with knowledge base via user interest modeling2013
- Discovering and disambiguating named entities in text2013
- LIEGE:2012
- LINDEN2012
- DBpedia: A Nucleus for a Web of Open Data2007