Second Opinions

We Want To Catalog Experts Data

By Timothy Anderson and Jo M. Haraf

DURING the Gold Rush, droves of citizens packed their belongings and headed west to make their fortunes. With little, if any training, they headed down to the rivers with pans to sift through the soil at the river bottom looking for the nuggets of gold that would make them rich. Some people got lucky, but many people failed.

The search for knowledge in the modern law firm is similar to the techniques used by the gold panners. With limited training, attorneys and paralegals are expected to put together complex queries to retrieve the nuggets of information that they know exist within the firm. What if there was a better way? How about a technique for panning for gold that allowed everyone to walk down to the stream and harvest as much gold as the river held? There is, and it starts with a commitment to knowledge management.

Knowledge management can be broken down into three key concepts: "expert locators," "question and answer pairs," and "data mining."

Expert locators are tools used to find people (or at least narrow down the list) with a particular expertise within the firm. Tacit, for example, has a product called KnowledgeMail that analyzes outgoing e-mail to passively determine a user's areas of expertise.

Question and answer typically refers to a user posting questions to an online community which are in turn answered by experts, presumably other members of the firm. Once answered, the Q & A pairs are available for searching. Orbital Software's Organik product functions in this arena.

Let's look at data mining.

At the outset, it is important to note that data mining systems are not cheap. Costs to implement start at $500,000 and go up from there. A small law firm considering making a commitment to data mining should first consider the size and number of their information repositories and prepare a cost/benefit analysis.

Structured vs. Unstructured

From accounting to human resources, document management systems to e-mail and voicemail, your firm is bursting with data that's ready to mine. At least 80 percent of the content attorneys and paralegals use is considered unstructured data. Unstructured data means that the contents of documents, Web pages, or text files is free-form, that is they are not specifically broken out into a fielded database such as in the case of the structured data in the firm's accounting package. Without a system that ties together structured and unstructured data, multiple searches must be performed in separate systems. Not only are you still panning for gold, but you're having to go to several different rivers to do it.

The largest and most logical information store in the firm is the document management system. In some firms, millions of documents exist on file servers located in several countries.

But what about e-mail, image libraries, voicemail, firm Intranet, or litigation support databases? Seamlessly accessing data from multiple (heterogeneous) data sources is perhaps the single most important feature of data mining products and most will support a wide range of potential information sources.

Hummingbird's Fulcrum KnowledgeServer, for example, will connect to Web sites and structured data sources in addition to the document repository.

Searching

The point of the data mining system is to access the materials in the underlying information stores. Data mining products in use today allow for a variety of methods of searching the data including categorization, keyword, and similarity searches to access data from multiple data sources.

Firms using a taxonomy can incorporate their categories and sub-categories into a product to automatically catalog documents as they arrive. Email notifications are then configured to alert users of new content in a particular category with a link back to the document. Verity's K2 Enter-prise product allows you to automatically create a taxonomy based on the existing data in your repositories. Traditional keyword searches have expanded to include single point of access to multiple information stores such as the document management system, email, and firm intranet.

In addition to keyword searching, some systems also allow natural language searching. Similarity searching allows attorneys to use a specific document or topic as an example to find other documents with the same or related concepts. A product like Autonomy Server from Autonomy uses "advanced pattern matching" technology to find related documents as well as images.

All of the complex technology in existence doesn't do you any good if you always find fools gold, so the system must deliver the correct content based on the criteria. So before you jump in the river, determine what you need, where it is stored, and how much you're willing to spend to find it. With the proper product in place, you can put the firm's data to work for you and strike it rich in no time.

Timothy Anderson is project manager, and Jo M. Haraf is chief technology officer at Morrison Foerster L.L.P. (and a member of the LTN Editorial Advisory Board).