Google Search Appliance Gets New Algorithm

    October 20, 2009
    Chris Crum

Google has introduced some new features for the Google Search Appliance (GSA). There are about ten of them, but the one the company is really touting is one that allows the GSA to learn the habits of searchers and use that information to fine-tune its information retrieval process to increase relevancy.

The feature is called the "Self-Learning Scorer. "Self-Learning Scorer analyzes employee clicks and behavior to automatically fine-tune and improve its built-in relevance," explains Google Enterprise Search Product Manager Cyrus Mistry. "For instance, if most users click on the fourth result for a given query, the GSA recognizes that and automatically boosts its placement – without any intervention from an administrator. Along with our existing and intuitive biasing features and Ranking Framework, this provides ease of relevance tuning."

The feature is discussed in the following clip:

"Most enterprise search engines do three things: crawl, index and serve," says Mistry. "With the new Self-learning Scorer, we’re adding a new step to the mix: analysis. As the GSA continually serves up results, it’s also learning to dynamically improve – automatically. This gives the GSA new self-improving intelligence, and adds a new step to the enterprise search cycle."

Self-Learning Scorer

Other new features include a new log-in feature and expansion of connectivity to Sharepoint and Lotus Notes. The new features are discussed in this video:

A couple months ago Google also related a couple other enterprise search features. They launched side-by-side search comparison and new connectors for the GSA in Google Enterprise Labs.