* Omgili: Oh My God, I Love It
A couple of issues ago I discussed a technique for encoding CAPTCHA images in HTML. This technique was devised by Ran Geva who also created the subject of today’s newsletter: Omgili.
Omgili is a vertical search engine that specializes in indexing discussions in user forums – the name is, apparently, an acronym for “Oh My God, I Love It.”
The edge that Omgili has is that the content of user forums is generally well structured so the Omgili spider can focus on the content and structure of discussion threads which, in turn, increases the relevance of topical searches.
According to a user forum posting on Slyck Forums, Geva explains: “Currently it indexes from a list of about 1,500 forums … The crawler has a special algorithm that analyzes the forum as a discussion and not as a regular Web page (like any other search engine). It can determine the topic, title and replies and give different weight to each.”
I have written about vertical search engines in previous newsletters and at least one company I can think of took my analysis to heart. While that is all well and good, what is interesting about Omgili’s approach is that unlike most vertical search engines they mine a restricted domain as opposed to the more common practice of aggregating domain specific content from multiple sources. You could think of the latter as a process of synthesis rather than one of abstraction.
That isn’t to say that synthesis doesn’t have value in defining domains of interest, but as a way of creating relevancy it is much more expensive and usually requires content translation augmented by human interaction, which makes the process expensive and error prone. Omgili avoids these issues by concentrating on well-structured content.
You can submit your own user forum for inclusion, and Omgili claims that its spiders, “recognize the popular board software such as phpBB, VBulletin, YaBB, SMF, Invision, Snitz and FuseTalk. In addition, we utilize a sophisticated algorithm that recognizes most of the custom made boards out there. Even if your forum isn’t on the above list, chances are that Omgili will still be able to parse and index its content.”
When you use Omgili you’ll find that its results are very dependent upon search terms – the more content specific the better the results.
Omgili is currently free. This service could go far – it is a cool idea that could turn out to have great value.
