The key word approach has dominated search in the consumer web. Google is my default launch page for the web. However, within the enterprise this approach is not always sufficient. People do not always know what they are looking for and, even if they do, they do not often think in the same key words as those who put together the taxonomy or drafted the documents. There are also many more data types than the vast amounts of unstructured data found on the web. In addition, there are many more security concerns within the enterprise.

Recommind was started in 2000 to address this challenge. It sprang from academic work at MIT and UC Berkeley where researchers were developing algorithms for automatic categorization and conceptual search. Recommind developed their MindServer Enterprise Search tool from this work. It is designed to help you find the most relevant information without having to know the “correct” keywords. Last week, I spoke with Craig Carpenter, Vice President of Marketing and Business Development of Recommind about their work. MindServer provides an advanced search engine for those who need to dig deep and find the unexpected. It focuses on several market areas: highly regulated fields such as financial services, highly litigious companies, industries that do a lot of research such as pharma, and professional services such as legal and consulting.

MindServer Search utilizes its concept-based search capabilities to connect relevant information in the many forms of content within an enterprise such as document management, records management, portals, e-mail systems and many other applications and databases, whether structured or unstructured. It returns results in context for more accurate understanding, as well as increased usefulness of the information for the tasks at hand. It integrates with major document management systems such as Open Text and Interwoven. Results are ranked by relevance and search terms highlighted to reveal the useful part of a document. They can be automatically filtered by such categories as industry, author, location, date, topic, or other metadata. Mindserver can handle information security and the use of business rules. It understands user privileges, role-defined access parameters, and rule based application usage found in enterprise applications. It supports over 30 languages and over 370 document formats. Here is a sample search returns.

MindServer is not limited to documents and can also be used for expertise location. Unlike some expert location systems, it does not require people to fill out and maintain profiles. It automatically builds a profile based on the individual’s work product. The profile includes their contact information, bio, relevant documents, and other information depending on the industry. You can refine people searches by the same filters as documents and create industry specific filters. For example, law firms can add filters for bar admission, law school, practice areas, and other legal terms. Here is an example of results for People.

Here is an individual page profile for a person.

Recommind customers include the top 15 UK law firm, Addleshaw Goddard, Australian Government, Bertelsmann, BMW, DLA Piper, Eversheds, Field Fisher Waterhouse and Simmons & Simmons. Novartis, the Swiss-based pharma company, uses MindServer in their legal department of over 400 lawyers and staff spread around the world. Because of this global distribution, the legal department found it hard to share information across time zones and languages. They built a legal portal known as RADAR with MindServer as the system’s brains that delivered relevant documents, projects and expertise to users irrespective of geography, language or business unit.

Because of its conceptual based search, MindServer is especially good at finding the hard to find people and documents and the unanticipated. It works at the “long end of the tail” within the enterprise. They are now looking to expand the social side of the search results by allowing users to bookmark and rate what they find, moving further into enterprise 2.0.