Natural Language Interface System for Querying Wikidata

Nowadays, information has a huge impact in our daily life; hence its extraction from the web is a need for everyone. However, due to lack of knowledge of query languages, an end-user cannot extract information from knowledge bases as they should. Asking question in Natural Language from knowledge bases enable end-user to extract data from such knowledge sources in a more efficient and fast way. Wikidata is one of the biggest knowledge bases which extract data in a triple format from Wikipedia. Therefore this research project aims at designing and implementing Natural Language Interface system for querying Wikidata. To achieve this objective, a Natural Language Interface was designed using Wikidata in python for processing the input question of user and give some answer to user. Firstly, preprocessing of input query was done, which involved tokenization, Part of Speech (POS) Tagging, Stemming and Lemmatization, Name Entity Recognition (NER), Chunking, Synonym check for words taken using WordNet. Then, Question Classification and SPARQL were performed to minimize the search area, show the type of output, and display the answer to the user respectively. The results of the experiment showed that processing is much efficient based on response time.