By Rachel Chalmers What the web needs is another search engine, according to a start-up based in Los Angeles’ Hollywood Hills. Oingo Inc’s technology is being promoted as a meaning-based search engine. That means the software performs contextual and other analysis on web sites in order to place them in its meaning-space or a large […]
By Rachel Chalmers
What the web needs is another search engine, according to a start-up based in Los Angeles’ Hollywood Hills. Oingo Inc’s technology is being promoted as a meaning-based search engine. That means the software performs contextual and other analysis on web sites in order to place them in its meaning-space or a large database (lexicon) containing around 250,000 meanings. The lexicon is partly automated and partly human-approved. Proximity in the meaning-space, in theory, implies a relationship between different pieces of information. That should mean the search engine can distinguish between different meanings of a single word, as well as identifying relationships between different words with similar meanings.
Executives claim that Oingo can distinguish between, for example, different uses of the word bull – as in the Chicago Bulls, the bull market and an adult male head of cattle. A given phrase is assigned a probability of meaning – apparently, most people who search for Bulls are looking for the NBA team, not the bovine. If it works, Oingo could combine the best features of both a search engine and a directory or index. But Oingo’s real strength is said to be its ability to produce results where no actual word- matching occurs. A search on Linux consulting for example, might turn up a positive match with a Unix contracting agency.
Oingo’s analysis is processor-intensive, making it far slower than its text-based rivals. A typical search takes about five times as long as its Google equivalent, but executives say Oingo is still so fast, no one will notice the delay. The index is currently based on the million or so web sites listed in Netscape’s Open Directory, but the plan is eventually to extend Oingo to the entire web. A patent is pending on the meaning-based search technology. The company is planning an official launch at Internet World in New York on October 7, 1999.
The company started from scratch about a year ago. It was founded by Gilad Elbaz and Adam Weissman, who met at Caltech. The founders received a half million dollar boost from seed capital firm Spot Ideas, and now boasts a head count of 13 to 14 people, depending on whether a recent offer has been accepted. The idea is not so much to build Oingo into a portal – although its context-sensitive results should in theory be more attractive to web advertisers than the less relevant results produced by text- based search engines. Instead, Oingo executives want to license their technology to other portals. In a year, we expect all the major portals to be offering meaning-based search, founder Elbaz explains, we want to be the ones who are providing that search.