As the title suggests, I am in need of finding / creating a system which is capable of retrieving relevant information for a query, by searching a set of given documents. From my own research, I have only found libraries who perform index based searching, that can point out which documents match the query, but they cannot extract relevant information. In short, a user will ask a question, and the system will try to find an answer to that question, by searching a given set of documents.
Can anyone point me towards a method of achieving this, or a library that might help me?
To give custom advice to your use case these infos would be helpful:
Generally this task seems to require tagging and semantic parsing of the texts. I like the nltk http://www.nltk.org/ best.