Entity-Relationship Extraction from Complex Type Natural Language Text

Abstract

Entity-Relationship (ER) extraction from natural language text is an open research problem. Earlier methods are only capable of extracting ER from simple natural language text. If natural language text is of complex type, then ER extraction will be difficult for any NLP based system. Complex type natural language text usually contains simple or compound sentences with many special characters. Complex type text may also contain words which are used in different contexts. The extraction of ER from these type of natural language text is difficult. In this paper a system has been proposed that will extract ER from complex type natural language text. This system is able to handle complex type natural language text without disturbing the context of the sentence. The Stanford POS tagger is used in POS tagging step with modifications required for ER extraction. By employing proposed method for ER extraction in complex natural language text, the Database Administrator, Database Engineer, System Engineer, NLP researcher, etc. may use this ER as per their requirements in Database schema representation, SQL or No-SQL generation, etc. which can be derived from the entity relationships.