A new type of Internet search tool presents a much smaller, but more relevant result, classified into categories.
A two-year-old start up company, co-founded by two Indians, created waves at the: "Demo 06" conference in Phoenix, Arizona, United States, with an innovative Net search engine.
At the Demo conference, various startup companies launched products ranging from a search engine that phonetically looks through audio and video to technology that relies on structured data to provide perfect—as well as close—matches.
A new search engine launched here from Mountain View, Calif.-based Kosmix provides categorized results instead of showing linear results.
Kosmix has the chance to significantly change the search experience, by helping consumers better handle the massive amounts of information on the Web, says Chris Shipley, executive producer of the Demo conference, an annual launch pad for emerging technologies.
It is powering leading comparison shopping and medical information services in the U.S. Kosmix Search currently provides search for three verticals: health, politics and travel.
"Kosmix" is a new type of Internet search tool which goes beyond listing millions of results for a key word, to present a much smaller, but more relevant result, classified into categories that will be more meaningful to the searcher.
For example, a search on the term "psoriasis," a common skin disease, turns up some 39,000 results classified into Basic Information: treatment, causes, symptoms, definition; Expert Information: guidelines, journals, clinical trials; Alternative medicine, fitness, diet; Medical Organizations; Men’s health; Women’s health; babies and kids as well as links to message boards and blogs.
However, the general-purpose search leaders such as Google and Yahoo have more results for the same term — approximately 6 million and 14 million — but wading through them might be a problem unless the basic search term is refined and narrowed.
Kosmix Search, which will also search weblogs, is aimed at companies that want to reach audiences with a customized search engine.
Using algorithms and technology to categorize the Web, Kosmix Search is able to provide a user using its travel engine to search for "Sydney" with results in categories such as hotel, transportation, museums and kid-friendly sites.
Kosmix was co-founded in 2004 by Anand Rajaram, an IIT-Madras Graduate, who is now Consulting Assistant Professor at Stanford University’s Computer Science Department. He and his partner Venky Harinarayan, a PhD from Stanford, sold their database technology company "Junglee" to Amazon, world’s number one book shop, in 1998 and moving over to guide their acquirer for a few years before branching out on their own again. They have since been joined by Srinivasan Seshadri, another IIT-Madras alumnus, who moved from the Indian end of Yahoo, to become Chief Technology Officer at Kosmix.
Mr. Harinarayan says that their patent-pending algorithms "deliver information that is not only more relevant, but richer and more accessible." The Kosmix site indexes over three billion web pages, but is currently restricted to three subject areas: health, travel and U.S. politics. The search engine, still in beta, can be tried out at www.kosmix.com.
Other Innovators Made Their Debut Are:
Transparensee Systems launched BestMatch, a search engine that searches structured data to offer results including the best match, even if the search parameters have no exact matches.
Targeted at e-commerce, BestMatch uses a proprietary algorithm dubbed "fuzzy search" to scour a database for results based on parameters chosen by the user. But while most structured searches will return results that will match search terms exactly, Transparensee will also show weighted results that are close to what a user is looking for.
Two companies debuted products with the ability to search rich media. Nexidia’s Media Search product allows users to find spoken word content on the Internet.
Executives at the company said Nexidia turns a search query into phonetic code and matches it against a database of audio and video files.
While other search engines provide results by tagging audio files and turning speech into text, Nexidia Media Search’s phonetic capability makes it easy for users to search through large databases of audio such as podcasts.
And since the product uses phonetic code, users can search for audio even if they don’t know how to spell the search term.
America Online showed what it has done with Truveo’s Visual Crawler Technology since acquiring the video search company in December.
Truveo’s technology was known for looking at a video’s surrounding web application in search of detailed metadata. AOL Video Search featuring Truveo will be available to AOL users this spring