SortFix Powered Search Blog

In the past few years, but especially in the last year, we have started to feel the “breeze” of change in the search engine sector — and more precisely the human-search engine interface and means of communication.

Search engines have been around for more than a decade. In that time, they have transformed our lives by turning us into information hunters and gatherers. I remember as a teenager I tried to learn about radio devices. I asked around; even went to the public library. All I could find were two measly books, one that was too technical and complicated for a beginner like myself; the other oversimplified. In those days, I had a 28 K modem, but it was useless for finding information. That was the end of my information seeking. Then the search engine era dawned. I started using Hotbot, AltaVista, and Ask Jeeves, until ultimately relying on Google.

As the Internet has evolved, and the amount of information has grown and diversified, sites like Wikipedia have cropped up that contain vast amounts of information on almost any subject, in fact, it is just a high information knowledge base. Millions of smaller, more sites are devoted solely to a certain subjects.

So today, learning about radio devices — or any other subject — is just a click away. But when you want to discover more, get exact statistic or data, solve a specific problem or explore “sub-subjects,” that’s when things get more complicated.

Plain Old Text

The search engine companies (Google, Yahoo!, Microsoft [Bing]) have improved coverage power and results-ranking algorithms to enable indexing billions of sites and pages, but until recently the search engine-user interface remained the same: a plain text-based search box that delivers tens, hundreds, and millions of results. The user is left baffled by the results because he doesn’t have the power to sort or to receive the exact relevant results, unless he does another search.

So we need efficient tools to filter the results.

Bring in the Clusters Bring in the clustering engines.

These tools like Vivisimo/Clusty and Carot2 attempt to categorize or “cluster” the results. Each cluster would describe a sub-subject, permitting the user to drill down into exact information to solve the problem.

These clustering engines have been around for quite some time but were outside the main search engine industry. Only search enthusiasts were aware of clustering engines and included these sites as part of their search process.

But in the last year the leading search engines began to embrace the clustering paradigm.

Yahoo! introduced Yahoo Assist!

Yahoo Assist

Google recently added Google Search Options, which displays/suggests related searches:

Google Search Options

Last but not least, Bing features the Decision Engine, its name for related searches or clustered search results.

Bing Clusters

Change Is in the Air

So why do I say there is a “breeze” and not “the winds” of change? Clustering engines are definitely a step in the right direction, but they are not a foolproof way to help the user describe his individual and exact search needs. Related searches are analyzed in various ways — by search history logs, computerized online/offline clustering algorithms — or a combination of these techniques. Therefore, related search clusters are statistically optimized to match a broad range of users. They are not inherently optimized to individual searches.

And the user interface of the leading search engines is just more of the same: basic hyperlinks that do not make use of RIA (Rich Internet Application) offered by new technologies such as Flash/Flex or Silverlight. The user cannot navigate between different categories or related searches and can easily become lost in the maze of possibilities.

Beyond the Clusters: A Truly New Way to Sort

SortFix’s technology enables the user to describe individual search needs in a precise way. The unique, powerful search environment presents the only search engine that can “handle” detailed, optimized, and truly individual search queries.

By combining a rich SUI (Search User Interface) with powerful semantic algorithms, the user can easily create, modify, and monitor his search query. The algorithms provide additional and optimal terms based on the available information and relevant to specific searches.

We are now developing a semantic analysis of these individual search queries. We’re sure this feature will bring about a “storm of change” to the world of searching.

17 Responses to “What’s New? Clustering Engines”