Monday, November 5, 2007

Improvements to the Interwoven Document Management System

Some representatives of Interwoven gave my group a look at the new searching tools that are coming out in their 8.3 release in December. These upcoming changes look like they will transform their document management system into a powerful knowledge management tool.

We use Interwoven Worksite as our document management system, with the Desksite client application and Filesite Outlook integration. We are lawyers. We write lots of agreements, memos and briefs. So, the document management system has always been our largest knowledge repository.

It has worked well to store and manage the document, but it has never worked well at retrieving the knowledge from the system. As I mentioned in earlier posts, a document management system typically is great on the management side and less on the knowledge side. The key role of the document management system is controlling the drafting of documents. It makes it easier to identify particular documents, fetch them back from the system, and edit them. A document management system works great to recall particular sets of documents back from the system using profiled metadata, like client and matter designations.

The document management system usually comes up short in searching for a precedent or research. Interwoven is particularly hampered because the search engine in the product seemed to be underpowered for the multi-million number of documents in our system. The other problem was the scattering of information across the profile fields for a document. For example, I am looking for a purchase and sale agreement used by ABC Corp. to buy the Blackacre Shopping Center. The key terms in the search "purchase and sale agreement", "ABC Corp." and "Blackacre Shopping Center" may be in any number of metadata fields: the client name field, the matter name field, the document name field or the full text of the document itself. You would have to run multiple searches to deal with the multiple combinations of which fields may contain the search criteria.

Interwoven is planning to improve its performance in these types of searches with its new 8.3 release. The big change is adding in the Vivisimo Velocity search engine under the hood. This appears to give searches in the system a big boost in speed and responsiveness. This 8.3 release is separate from the Universal Search web-based product but uses some of the same hardware and indexing technology.

The first great change is Interwoven’s goal to give the user a single textbox to search. (Just like Google.) In giving the user one box to search, Interwoven combines that across all (or at least many) of the metadata fields. This new version should allow you to find my earlier example of the "purchase and sale agreement used by ABC Corp. to buy the Blackacre Shopping Center" just by typing into a single search box. The search will query the various metadata fields instead of the user having to guess which field may contain the information you need.

The next great change is that search results now come back based on relevancy, which is a huge boost to the usability of Interwoven for a research type of search. In its current version, Interwoven presents documents in a grid, sorted by one of the columns: title, last edit, date, document number, etc. That is useless when you are trying to find information on a subject. You want the deepest treatment presented first on the list, not the documents starting with the letter “A.”

We have been pouring our documents into the document management system for decades. We have been looking for better ways to better pull that information out and utilize it. This new release may finally allow us to release that.

1 comment:

  1. Interesting article. I have found most of the DM/KM systems out there do fall short in search capabilities. They all seem to perform just fine in the beginning, but as you begin to bog down the database, it just gets slower and slower. This is definitely the case for full text searching. It usually pays to setup index fields to narrow down the body of knowledge, and then do full text searching from those results.

    Can I ask what you use to capture your paper documents?

    I have some articles on Document Management and Planning at


Note: Only a member of this blog may post a comment.