Web Virtual Assistant is a computer program which responds to a user query relating to that particular website. The virtual assistant is realized using two modules: static module and dynamic module. The static module is used for extracting the answer to the question based on the static content of the website. The dynamic module is used for querying structured information present in the website. The system works for the website which have semantic web technology implemented.
The technology used are Web Ontology Languge(OWL) and SPARQL querying language. Admin is allowed to enter intents related to database or FAQ. Each intent is created by adding number of examples and the answer structure to that question. The comparison of Naive bayes classifier and word2vec classifier used for identifying intent to the question, is made. The expert system when fed with 10 test cases, gave 80% accuracy. This is the best case scenario. Keywords: natural language processing, web scraping, chatbot, semantic web technology, web ontology language, SPARQL, Naive bayes, word2vec.
Read full project report on: Research Gate