Written by Weiru Chen, Dean Hathout, David Zheng, Tyler Yoo
_Code can be found on our _Github
IACS_ Poster can be found here:_
We would like to thank Gabriel Altay and Georg Kucsko at Kensho for their graciousness in sharing their time and resources with us throughout this project.
Finally, we thank Chris Tanner of IACS at Harvard for his invaluable guidance to our group and for his leadership throughout the Capstone experience.
Introduction
Named Entity Linking, also known as Named Entity Disambiguation (NED) is the task of uniquely identifying entities (such as individuals, locations, companies, or historical events) mentioned in text. To give a canonical example, if given the sentence “Paris is the capital of France,” we want to be able to discern if the word ‘Paris’ is referring to the French capital, some other city, Paris Hilton, or many other possibilities, shown below.
Figure 1: Wikipedia’s “Paris” Disambiguation Page
Along with Named Entity Recognition (NER) — the process of actually identifying mentions of such entities in text — NED is one of the most foundational tasks in Natural Language Processing (NLP); being able to identify the specific things a text is talking about is essential for countless NLP applications, including general text analysis, semantic search systems, building chatbots, etc.
#kensho #entity-linking #nlp