I'm interested in IR and Search Engines and want to work on a theory problem in this space. Most thoughts that I get are focused towards implementation and would like to explore something not solved or can be bettered instead of implementing
Since this type of understanding is still poorly implemented in software, automated anaphora resolution is currently an area of active research within the realm of natural language processing.
These are two big problems that search engines have. Relivence is easy to explane. The search Engine is not a Neuro Interface yet, but it could be, there is your awnser, if a search engine such as google was operated by neuro-interface, with the write programing and the web cam to read the enviroment the search engine could bring back much more relivent results instead of 500 results of nonsense for the next spider.
Speed ou type the querry in, goes wizzing of across thosands of miles of fiber optic and possably up to satilites on its journey. Finaly it arives at Google Data Center, one of the largest Computer clusters in the world and a very secret location. that querry is sent around all those computers and when it comes out the other side it has gotten a bit larger because it now has the response to your origonal search term, and so it starts its return journey across thosands of miles of fiber optic, coper and those two satilites. Than you get your results. Now this only takes a between 3 seconds to 2 miniutes your but while your waiting it is chuging your bandwidth and up your broadband bill. Wouldent it be faster to lay a global network of Fiber optic cable with repeaters ? Than Broadband would run at close to the speed of light.
Thats my take on the Speed part of the Question.