Enterprises Widen the Search Net

Computerworld - Intranet portals can offer a single point of access to all of a company's documents, but employees may still waste time trying to find what they need among those thousands -- or millions -- of documents. By interpreting user queries, searching across different data stores and presenting the most relevant results, enterprise search engines (ESE) simplify the task of finding the right information in the right context, regardless of the source.
Search-and-index technologies are built into many applications, but they let you find data only within the associated application and can't hunt through the many data repositories an organization might have. ESEs typically search both structured and unstructured information in many different formats and across all the devices on the corporate network.
The sheer volume of documents at Anadarko Petroleum Co. drove Bob Downing, manager of business systems, to deploy an ESE. "Our engineers, geologists and geophysicists were spending about 50% of their time searching for information among over 2 million documents on our intranet," he says. The Woodlands, Texas-based oil company installed Vienna, Va.-based Convera Corp.'s RetrievalWare to index and search its content repository. Anadarko's ESE indexes all of the company's SQL databases, its intranet, Windows file-server directory structures and everything stored in its content management system from Pleasanton, Calif.-based Documentum Inc., as well as several online news sources. Downing estimates that the engineering group now saves 78,000 hours annually by using the new search functions.
Bag of Tricks
ESEs use a mix of technologies to find the right result, and the products continue to improve. "A lot of the technology has been around for a while, but we are now starting to see practical applications . . . in the business realm," says Laura Ramos, an analyst at Cambridge, Mass.-based Giga Information Group Inc.
ESEs have the same core elements as Internet search engines: They spider enterprise content (locate documents by following hypertext links), index it and present it in response to user queries. However, the indexes can be quite large. Washington-based law firm Dickstein Shapiro Morin & Oshinsky LLP has nearly 1,000 employees and several million documents: 2 million managed with Fulcrum KnowledgeServer from Hummingbird Ltd. in Toronto and millions more that it accesses using specialized litigation software. The firm's database includes files saved as Word and WordPerfect documents, as well as PDFs. The firm also has content in its proprietary CRM database and in documents in still more formats received from clients and litigation opponents. Since the attorneys need to do word-proximity searches, the index required