My University Final Year project involves in developing a Question Answering System that allows users to ask questions based on online material, for example Wikipedia. The idea is to traverse all the sentences and find the most applicable and accurate answer to a specific question.
I understand that this requires Natural Language Processing and I COMPLETED that part. I have developed it so that it stores all sentences in a database with all its Lexical Information and Entity extraction. My question is, what's the best algorithm to traverse through a list of sentences and find an accurate/near-accurate answer for a specific question the user asks?
For example: When was Microsoft founded?
And referring to Wikipedia's Microsoft page, the system should specify the answer: April 4, 1975
For example: Who is the creator of Marvel comics?
And referring to Wikipedia's Marvel Comics page, the system should specify the answer: Martin Goodman
This is a discussion so please point me in the right direction, good lads! Just a proper algorithm would do, I shall take care of the coding afterwards :)