Recently Bill Slawski from SEO By the Sea wrote on a patent application filed by Yahoo on Techniques for Searching future events. It brings forth a new dimension to consider in making search more timely and relevant to users.
The patent is on making search engines acknowledge the temporal information in articles. This implies interpreting the word usages that refer to chronological information and classifying articles on the basis of the timeline that the content deals with.
With the torrential increase in content creation online and the computing resources of scale that are being implemented to process this content, there is room for a slew of new developments in making search solve a lot more than queries. Case in point being the massive application of Apache Hadoop at Yahoo. Combine that with temporal analysis of information and you get a host of applications for industries that are immensely temporally sensitive ( e.g. Financial markets).
With massively scalable computing resources and no dearth in content online, the question that remains is how to leverage these to make the web more receptive to human ways of information processing. I am referring to a sort of Mechanical Turk arrangement that seeks to put the massive computational resources at the disposal of users in a manner that makes it easy to make the best of human and computational resources.
We already have a large number of sites generating social media, what is needed it perhaps a means to have machines learn and observe from the way ratings are given to content. It would result in an ever improving system where machines can scale to better ranking of content in the web while humans cover a small subset of the content online.