Is a new type of search coming?
The signs are that search engine technology is about to rapidly evolve. HP has just purchased Autonomy, a UK-based company specializing in unstructured searching, for $10.3 billion. Autonomy CEO Mike Lynch hopes that the marriage will produce a super-search engine, combining HP’s expertise in structured textual searches with Autonomy’s know-how in unstructured search.
New Scientist suggests that HP’s Vertica analytics tool could be combined with Autonomy’s Idol platform, which helps determine the relationships between elements in unstructured information, including audio and video.
Autonomy’s search engine can trawl multimedia files, finding, for example, anything that looks like a Coca Cola bottle in a series of photographs or videos, or any instance of the word “insider trading” in recorded testimony. As the two come together, are we looking at the foundations for a new kind of search engine?
how would a search engine with such capabilities be used?
Experts such as John Battelle, author of The Search, suggest that search engines themselves will become more like middleware, front-ended by specific applications. Perhaps, then, a super-search engine might serve a calendar application in one instance, that would enable you to find a voicemail in which you arranged a particular appointment and retrieve the address.
In another instance, a search engine might support a set-top box application that looks for recipes in TV shows containing oranges. Perhaps it may also serve a niche application for law enforcement, scanning video for number plates and faces for policing purposes.
a new generation of search engines
The merging of structured text searches with unstructured searches could also lead to a new generation of search engines that find both subtle relationships in structured data, and links between structured and unstructured data. Perhaps a police officer with little more than a witness’s name could turn up addresses for that witness along with information about other cases that they had been involved in, buried in long documents. If that witness’s name also turned up in recorded testimony on digitized audio tapes, it could end up being the difference between a solved investigation and a cold case file.
A new generation of super-search engines is coming, but ironically they may become the invisible engine behind specialized applications that exploit them more effectively.
What other applications do you anticipate there will be for search engine technologies?