I implement a Question-Answering (QA) system using NLP features and techniques for the following Question Types:
a) WHO Questions
b) WHEN Questions
c) WHERE Questions
Input: natural language question
Output: a. Exact answer phrase(s)
b. Supporting sentence(s) in Wikipedia document
c. Supporting Wikipedia document name(s)
Feature Extraction:
Implementing NLP pipeline to extract following features from the Wikipedia articles-
a. Tokenizing the text into sentence and words.
b. Lemmatize the words to extract features.
c. POS tagging the words.
d. Performing dependency parsing to parse tree-based patterns as features.
e. Named Entity Recognition on the sentence
f. Word Relations like hypernyms, holonyms, meronyms, synonyms and holonyms as features for the corpus lemmas