TechNayak

Google unveiled their universal search page for their search results, basically a page where you get more than just web pages, you get blogs, pics and videos even.  But  this is my question  - is  the concept of universal search even possible ?

Penetration of user generated data into spaces like technology and entertainment have been exponential. Every person who wants information on code and solutions looks up forums and blogs but there are certain niches which are still firmly controlled by the creators like astro physics and rocket science ( pun) . Now when we have such a divide how can we make sure that universal search is actually delivering to its full potential. The primary criteria I guess is to put the search query in its proper classified subset. By this I mean, if I search for a topic that has more material in a blog than a wiki or a full blown site, then how will the algorithm know its relevance. Ranking clearly fails in these conditions and its pretty evident.

The second biggest problem is of course, the real estate. What I may be looking for may not turn up in the first search result and with most of my observation people don’t like shifting pages to more results, they decide to change the query instead. In that minimal screen how will you fit in the right contents based on the rank, relevance, semantics and many other jargons that even I,  haven’t heard of. One thing clear though is you need to mine.

Mine data from the users, their previous search results, probably from the one place where they have disclosed something about their life, their interests, hobbies etc - Social Networks.  Even the real estate management should change , there should be a new scheme of viewing results for a query, probably a tag based scheme that can subdivide the query into many groups and subsets of relevance and then refine it on top of the tags. It requires a change, in peoples usage, which is a dreaded de-motivator , but tags started the same way and its doing pretty alright in sites like flickr and del.icio.us.

The concept and implementation of web search is still unexplored and untapped. There are a wealth of things that can be done and I have mentioned a couple above.  Semantic web could solve some major problems, but its still a long way away and more time means more space for innovation.