Tuesday, October 23, 2007

Segami Revealed

Today is the launch of SegamiSearch.com , Version Alpha.

Its called version alpha, since this is more of a technology demo, to demonstrate the technology capabilities of the search engine. The image index is small, and mostly has images from Univ of Southern California, where the project was originally developed. Currently crawling is underway to build a much bigger index of images.

This is an exciting moment, as for the first time in web search history we see a web image search engine with image similarity search feature. Even though the algorithm is still being refined, it shows the future to where we are headed, breaking the traditional "keyword only" search.

Play around with the search engine, have fun with it, and most importantly, provide me with feedbacks and suggestions.

Sunday, October 7, 2007

"Segami Search"

SegamiSearch.com is an image search engine. I had to figure out a name which reflected that.

Naming websites is an art, these days, since most of the names are already registered. Some of the names were just registered, but no actual site existed.

No wonder, the current Web 2.0 companies, and search engines, seem to follow a trend of coming up with strange combination of letters, which had no meaning - but sounded catchy, which are purposely misspelled or from a non-english language.

I played around with a few words, their synonyms. After a lot of word combinations, came up with "Segami" which is "images" spelled backwards! And combined with "Search" was perfect as it reflected the core search paradigm, and sounded good.

Welcome!

Welcome to the official blog of SegamiSearch.com, the world wide web image search engine. Currently the site is under development, the alpha version will be launched soon.

Segami Search started as a graduate project at University of Southern California. It brings a new flavor to web image search, with the feature called image similarity search. Once images are searched using keyword, the user can narrow down on the search results by searching for similar images. This will help the user find the most relevant image, and covers the limitation posed by pure keyword based image searches.

This blog, will discuss the development of the search engine and the technology behind it.

More details soon, watch this space for updates.