Goal:
Q&A is an important knowledge data to enable many scenarios like auto question answering in bot. This Q&A miner provide a platform to 1. Extract Q&A data automatically w/ human knowledge in loop. 2. Mine sematic tags like domain, entity, relation as well as intent and condition from Q&A data. Q&A extraction contains two parts: FAQ extraction from both web pages and enterprise documents like word, and Q&A extraction to extract from crowd sourcing data like online forum. After we extract many Q&A pairs, Q&A Miner learns the semantic tags by using NER, intent taxonomy mining and recognition, conditional knowledge mining as well as question linking techniques.
Portal: http://msraml-s004:2410/ (opens in new tab)
Scope:
Techniques:
QnA Extraction provide an active framework to extract QnA pairs from web pages. This pipeline can not only extract QnA pairs by using existing model, but also provide a human in loop way for users to help to improve the performance. The unconfident pages will be selected to ask users to label QnA, and the newly labeled data will be fed into the classifier to fine tune the model to achieve better performance.
QnA Mining can learn the semantic tag from the QnA like domain, entity, intent and condition.
Application:
Digital doctor
Customer support
People
Lei Ji
Senior Researcher