Context aware Named Entity Recognition with Pooling
Shreyam Shah, Jyoti Ramteke · 2021
Natural Language Processing is concerned with processing and analyzing natural language with the help of a computer, i.e., any language that is spoken or written by us. Natural language processing is one of the fundamental elements of AI. Named Entity Recognition is a standard Natural Language Processing problem which deals with information extraction. NER involves the automatic scanning through unstructured text to locate entities.Many of the Named Entity Recognition systems available are not context aware. Many a times a named entity appears in a text and it is used in that context in some other context. For eg. the word Washington may be either a person or location. The actual named entity depends upon the context in which the word Washington is used. It is possible that the NER might not recognize it correctly since none of it saves the context anywhere. This paper revolves around the exploration about making the algorithm more accurate by making use of different pooling mechanisms. The pooling mechanisms can be used along with the BiLSTM + CRF architecture to get the entire sentence context.