PDF version of this document
Neural networks are not quite so feasible as they are too many parameters (hence dimensions) to consider. Each word is a dimension of its own, but it is worth understanding how search engines overcome that same problem.
The need for caching is apparent, yet in our case it does not involve indexing. The cache is in some sense the 'brain' of the engine. Apart from knowledge indexing, this cache may be needed for providing references along with an answer to a user's query (which should ideally be labelled a ``question'').
To test the idea at a very small scale, experimental engine can be made available for schestowitz.com (in case of University refusal) and a little search bar added to the main page.