A Better Way to Search?

Like most people, I spend lots of time searching for information to help me get my job done each day. For example, I search through my email messages to find room information for meetings, instructions from colleagues, and remind myself about messages I sent earlier.

When I'm starting a new project, my best research comes from Web pages I've found through searches. When I get stuck on a problem, an intense search usually yields the answers I need to keep working.

This is sometimes a brain-racking exercise, since most search engines require keywords to provide the best results. The trick is coming up with the right set of keywords, and using boolean operators to refine the results.

Sometimes I wish search engines could work the way I think. For example, instead of having to form a query like "solaris 10 + install + laptop - server," I'd rather ask the question "How do install Solaris 10 on my laptop?"

The solution might be close, at least for finding answers to Solaris 10 questions. The folks behind the Solaris 10 Knowledge Base pilot have developed search capabilities based on natural language queries. You enter a phrase or question just like you would ask a friend or colleague, and behind the scenes the natural language processing technology locates the documents that are most likely to provide the answer.

When I posed this question, I was pleased with the results:

Not too shabby! The first document returned is a How to Guide for new users wanting to Solaris 10 on x64 or x86 systems. Even if the results were not what I expected, I could easily narrow my search by selecting a category (such as Install/Boot) or a document type (such as documentation). It's definitely a new way to search, and I'm having to unlearn some old "boolean" habits and trust my instincts.

Intrigued? Try the Solaris 10 Knowledge Base pilot and let us know what you think.