To Google or not to Google?

That is the question!

Shall we suffer the endless calls of "just get Google!"?
Or take arms against a sea of "search stinks" comments, and by iterating, end them?

Google quietly entered the Enterprise Search market in early 2002. Analyst firms dismissed it as a tool for sophisticated environments, assigning it to the realms of inexpensive, simple, DIY search. The product line has grown since then, and has gone on the offensive in terms of marketing. New capabilities, notably the ability to integrate with Google Desktop Search for Enterprise, are renewing calls to "just get Google." Is it right for your organization? Perhaps.

This call comprised presentations by Seth Earley and two guest speakers, Avi Rappoport of Search Tools Consulting, and Christine Connors, who works for Raytheon but gave her own independent presentation today.

Shall We Google? Search Challenges and Issues
Seth Earley, Earley & Associates, Inc.

Seth began by explaining that the appropriateness of a search engine such as Google depends on the content, and that not all content is the same. Content can be created in structured or unstructured contexts, and its value can vary depending on audience, context or process. Some content is extremely nuanced and requires more precise access, such as according to audience, task, solution, etc. Imprecise terms are not suitable for precise access. Search can be based on inherent structure and content of a document or on information applied to that content.

Google is known for it ranking and relevance features. However, what can be leveraged in the Internet world cannot necessarily be leveraged internally within an organization’s content. Seth covered topics of tagging, ranking and relevance, relative value, indexing factors, and the types of content that requires structure. He presented several graphical schemes of how to deal with content. Finally with marked up screenshots of the interface of the Google OneBox, Seth demonstrated how Google uses keywords and thus can leverage taxonomy values. But then one still has to do all the work of creating the taxonomy.

Choosing an Enterprise Search Engine
Avi Rappoport

In choosing an enterprise search engine, Avi covered comparisons of Web and enterprise searching, elements of enterprise search engines, best practices, types of search engines, and the testing and evaluation of searches.

Avi began with defining enterprise search in comparison with Web search, presenting both the similarities and differences. She then explained how search and taxonomy complement each other, rather than compete. She listed elements of enterprise search engines, including its tools, indexing, query, ranking and interface, and then provided some best practices for search.

The presentation also included a checklist for choosing a search engine. Avi advised against building one’s own, unless there have unique requirements. Also as part of the search engine selection process, one should conduct an information needs analysis and a content survey, for which Avi listed some of the questions to be asked.
Avi described the types of search engines available: software only, search appliances, and remote search services. She then suggested how to test and compare different candidate search engines and listed factors to consider when evaluating the searches. Finally, Avi discussed the specifics of the Google search appliance and concluded by saying one should choose what works best for one’s situation.

To Google or Not to Google?
Christine Connors

Christine’s entire presentation focused on the Google enterprise search products. She covered the models, pricing, and methods of extending the Google Search Appliance.

There are two models of Google’s OneBox enterprise search product: the lower-end Google Mini and the higher-end Google Search Appliance (GSA). Christine provided information on the pricing, customer support, number of documents that can be searched, and purchasing options for each.

Christine explained the crawling process, showing a screenshot and providing a link to a presentation of a tested algorithm she had done before. She also explained the front-ends and showed two front end interface screenshots. You can have multiple front ends, although only one index.

The presentation included explanations and screenshots of the features of Collections, KeyMatch, Synonyms, and Metadata. With Collections you can focus the search on a subset of the objects. With KeyMatch, you can manually insert links at the top of results pages. For Metadata, you can import a taxonomy, but there is no support in creating it.

In conclusion, Christine said that the Google enterprise search products are very easy to manage, very stable, but do not do taxonomies as the Taxonomy Community of Practice knows them.