Nowadays, there are so many AI applications of Question and Answer System which could solve some problems in daily life.
In most situation, when AI answers your questions, it can be defined in three styles:
- IRQA (Information Retrieval Question Answering), a kind of normal form, can answer questions similar to others which are in its database. In this form, the AI returns to people the highest similarity answer but not creates a new one.
- KBQA (Knowledge Base Question Answering): AI splits your question and links to its knowledge base, searches the keywords and recombines the answer. This kind of AI is based on the huge data and a precise algorithm, so it can be used in wide areas.
- MRCQA (Machine Reading Comprehension Question Answering) is a pattern which relies on templates. It just likes KBQA, but the distinction is that when AI splits question into keywords, it seeks templates in database rather than combines on answer by itself.
These different solutions are using in our world and providing the smart AI like Siri.
Even though we can build simple AI with QAS, there is still a large of problems waiting for us to solve.
References
《企业计算中的机器人智能问答》 王昊奋 奚 宁 周 扬
《从问答系统看知识智能》 赵 军