Hello Everyone I am building chat-bot for information based portal which provides articles on six domains mainly Education, Health, E-Governance, Energy, Social-Welfare, and Agriculture. I want bot to reply to user query based on intent and entity recognized by chat-bot. There can be multiple entities in query. What would be the best approach to build such type of chat-bot. Please provide your inputs. Here is Link the to the website for which i am building chat-bot. https://vikaspedia.in/InDG
Hello @pankaj676. My suggestion would be to start with a list of questions you would like your assistant to answer. Then those question can be converted into the intents and from there you can start generating some NLU and stories training data.
My suggestions though would be to start small and pick a few the most important questions you would like your assistant to answer and build things up as the time goes.
@Juste thank you for reply. Basically what i want is to improve search in our website. Like if we searching for educate then it should show the results for education as well and also for misspelled words like eduacate.Do you think chat-bot can help in this.
Hey @pankaj676. If you are only interested in the search, I would challenge that you need a chatbot for such a task, because to achieve what you are looking for you could utilise some fuzzy matching and some other techniques to improve the search and word matching.
A chatbot could provide your users a nice conversational interface if instead of searching they would like to ask questions about specific topics. Combined with the power of handling follow-up questions it could become an alternative way of how your users find the information on your website.
@Juste I want users to search for information by asking questions as well as to improve simple word matching search. Would love to know how would assistant know the context of question instead of full text search using slots and without using slots. Is it possible to search using assistant without using slots.