How do I make a database for a question answering system?

How do I make a database for a question answering system?

Generally, QA system knowledgebases are based on question-answer pairs. FAQs are good examples of question-answer pairs; and, CRM question answering system knowledgebases are often based on FAQs. The difference between a knowledgebase and a database, and between knowledge and data, is metadata enrichment, or annotation. In this case, answers could be considered a kind of metadata enrichment for questions.

Question-answer pairs can also be used to train machine learning. Similarly, chat logs – another form of question-answer pairs – can also be fed into machine learning. In this way, human collective intelligence can be used to train AI; in fact, the new Facebook Mwill use such Mechanical Turk style training to ramp itself up.