LEX: Language Extraction Engine
LEX, is a thematic language extraction NLP engine that uses rule-based and machine learning techniques to identify and extract thematic language from documents. The goal of LEX is to highlight sentences within documents pertaining to specific type of language. The engine uses two sources of input to determine the extractions: a lexicon of specific terminology as well as a word-vector model. The lexicon contains rules constructed by a domain expert; while the word embedding model is used to find similar terms to those contained in the lexicon rules. LEX provides a way for different GUI dashboards to be generated which allows users to further browse the document corpus and interact with the results. It also highlights the extracted sentences within the original document. Thus, after homing in on relevant sentences, the user can go directly to the sentence within the document and read it in context.
Knarig is a Senior Associate Knowledge Engineer at the Federal Reserve Bank of New York. She conducts research in Natural Language Processing, Machine Learning and Semantic Technologies related to bank problems. Prior to her time at the Fed, she was a professor at Columbia and Hofstra Universities and a research scientist at Bell Labs. She holds a PhD in Computer Science from Columbia University.