Factverifier: Search Engine Based Question/Answering System to Verify Facts towards identifying and answering polar questions authoritatively by using pattern matching for web search engine

Main Article Content

Gundeep Singh Bindra
Seema Khanna, Dr. Harish Chaudhry

Abstract

“Is the Eiffel Tower 324m tall?â€, “Was Thanksgiving Day on Nov 22 last year?â€, “Was Thanksgiving Day on Nov 22 in 2007?†If you have similar facts and you are not sure about them, verify the facts with your favourite search engine. With more and more complicated assertions people make these days, it is difficult to tell which is true and which is false. In this era of information technology where the major source of our information is Wikipedia, Google, Blogs etc. our belief and dependence in them is immense. This paper aims to discuss new techniques being evolved for using search engines beyond their basic purpose. Truth is a concern for most of us in this con world. It is an examination of the various search engines on how they respond to well known facts and even factual statements. In the name of VERITAS, the Paper proposes a new method of answering yes/no questions for the purpose of verification of facts based on pattern matching, we focus on questions that the search engine already knows the answers for. For example, “How tall is the Eiffel Tower?†In order to answer such questions the search engine applies mapping pattern matching its algorithm. This paper describes the algorithm that finds the answer to the queries in the form of a yes/no. Proposed here, is a method for evaluating the algorithm and report results that are obtained for a self-made question set. Finally, there are some suggestions for possible extensions of the methods. This conceptual paper is definitely expected to contribute to future research on similar and related topics as a spin off from this study.

 

Keywords: Search Engine, QandA, YES/NO Questions, Fact Verifier, TRUE/FALSE, Information Retrieval, Question Answering System, Answer extraction

Downloads

Download data is not yet available.

Article Details

Section
Articles