The Munich based company, Proximic has been the result of 5 years of research into a search methodology that is a set apart from the keyword and the meaning based approach used by other prominent search engines. Termed as “pattern proximity”, the core algorithm applies pattern matching across sets of characters and matches the profiles of web pages in real time. The result is similar to a keyword search with 200+ words, resulting in more relevant articles for users and betters ad targeting for publishers.
A quote from the Proximic site :
Proximic basically understands composition of text, not text itself. Take an example to imagine how Proximic works: how would you intuitively describe a person’s face? The fact that it has eyes, nose, and mouth are normalities. But the particular size, shape, color, and location (specifics) provide the description, or context.
The whole model does not rely on keywords and essentially provides a very powerful platform for publishers to make their content available to users and also use well targeted ads. This gels well with Proximic’s drive for a contextual content network (MarketWire), a syndication framework for content across the web.
For the users the power of the proximic engine can be tapped with a FireFox plug-in that continuously lists relevant content as you browse. While the company is not immediately targeting Google or the other biggies per se, their Ad model does seem to have a lot of potential for publishing networks that cannot curtail themselves to the keywords model.
The basic aspect of Proximic technology is that it is not trying to make sense of the queries or the search terms. This goes well with the fact that a smart chess playing program need not be intelligent to play chess well, just as a smart search engine need not understand the query as long as it can match the patterns of data real well. And Proximic also happens to have as its CTO, a former computer chess champion.
Will their technology pay-off in the long run is what really matters.