by Ron Friedmann, 1999
Written for Jnana Technologies Corporation
ABSTRACT: Why in-house counsel should consider creating expert systems to provide answers instead of merely information.
Business has become more complex and competitive. E-commerce, new technology, new alliances, and new distribution channels affect many industries. Turnaround times continue to accelerate. At the same time, regulation has become more burdensome. All these trends increase the demand for guidance from legal experts. Yet few Law Departments have the luxury of adding headcount in the face of these seismic shifts and the growth of the businesses they serve. In short, they must do more with less.
In response, many departments have built legal Web pages on the corporate Intranet. Their goal is to provide explanations in an easy-to-use computer system so that clients can help themselves. Legal Web pages help, but perhaps not as much as initially hoped. This paper briefly examines corporate Intranets and their expected evolution from providing information to answering questions.
Web Beginnings: Distribute Document
The advent of the Web was exciting. For users, it offered a single, easy-to-use interface. For information technology professionals, it offered standards-based software that was superior to proprietary systems. Regardless of software and hardware differences, it became possible to distribute information in a format that everyone could view.
The ability to distribute information widely and in a uniform format was a significant development. Law Departments now had a way to share important documents within the department and, even more usefully, with clients. Progressive departments created Web pages that provided guidance on a range of topics.
At first, just putting documents on the Web sufficed. Web sites were small enough initially that users could quickly find what they needed. All that was needed was an alphabetic list or simple index to find the right document.
As document collections grew, users needed more help. Simple lists evolved into menus of increasing detail where each menu item was a clickable link. Users could navigate through a set of menus, recognizing potentially promising paths. With some luck and a few clicks, they found a “good hit,” that is, a useful document that answered their question.
As tools became available to convert word processing files to HTML (web format), document collections grew rapidly. However, the menu approach to navigating through these growing document collections has proven quite limited. It takes too much time — and too much know-how — to add new links for each additional document. The response was to add full-text search capabilities to the menu approach.
Some hoped that full text searching would allow users to find whatever they needed. And staff would not have to spend time indexing documents or creating links. Just put a document in the database, automatically “re-index” it, and every user could access the document.
Anyone who has searched the Web knows, however, that full-text searching is not easy. Searches tend either to miss many relevant documents or, if they include most relevant ones, also include far too many irrelevant ones. A seminal study 15 years ago found that even professional searchers typically found only about 20% of the relevant documents in a large collection.
The limitations of full-text searching are now widely recognized. Web sites, librarians, online information services, and software makers use several strategies to try to overcome these problems. Some examples:
- Hire skilled professionals to index and describe documents with “controlled” key words. Users search these “concept” descriptions, increasing the likelihood of finding relevant documents and excluding irrelevant one.
- Establish a rigorous process limiting which documents go into Web systems in an effort to avoid hit lists that are too long.
- Create “natural language front-ends” so users do not have to learn arcane search commands.
- Use “smart” search engines that improve the “hit list” returned to users.
- Present search results visually (i.e., spatial maps) so that users can locate relevant documents in a large collection.
- Use “collaborative” filtering or “link analysis” software that determines which documents are relevant.
An Evolutionary Dead End: Why Do We Want to Burden Clients?
Despite these improvements, searches are still too hard. And efforts to improve searching overlook a fundamental problem. The whole approach of offering document collections rests on two assumptions:
- Static documents provide effective answers to legal problems.
- Users with pressing legal problems are willing to find and read documents to try to extract such answers.
We should question these assumptions. Even the most sophisticated human indexing schemes, combined with the most sophisticated technology, still put a big burden on users. The user must execute a search, review a hit list, select documents from the list, read at least a few, determine if any are relevant, and, if some are relevant, figure out how the document(s) apply to the problem at hand.
In the legal context, how useful is it to just offer documents? Even motivated and skilled clients have only so much patience for searching, selecting, and reading. If the goal is to help clients help themselves, thereby lightening the load of the Law Department, is there a better way? We are now on the cusp of the next evolution of the Web. One that moves us from distributing information to answering questions.
A fundamental shift in Web technology allows us to re-frame the whole problem of how best to help clients. The Web started as technology to deliver information. Various technical developments (e.g., Java and Active Server Pages) allow the Web to deliver applications as well as data. This means it is possible to manipulate, analyze, and act on data as well as just distribute it. The move from information delivery to application delivery enables a whole new approach to addressing the Law Department mission.
Jnana Technologies Corp. has built software that allows Law Departments to create applications that provide answers and, where they cannot provide answers, intelligently interview clients to collect the information needed to assess a situation. The Jnana Knowledge System is not content-specific. Law Departments can develop applications about any topic. Applications can dispense advice, collect information from users, and track actions taken.
Creating “client self-service” applications may not pay for every problem. This approach works best for high volume, moderate complexity problems. These are typically the problems that consume much of the time spent advising clients.
The benefits of answering questions instead of just distributing information are substantial. (See the case studies at the end of this document.) Frequently, a Jnana application answers the client’s question with no further action by the Law Department. The client can get the answer day or night, in any time zone, without the need to contact a hard-to-reach lawyer. As more clients answer questions on their own, the Law Department can re-deploy scarce resources to higher value activities. And the Law Department can get a report of every use of the system for purposes of review or audit trails.
Even where a Jnana application cannot answer the question definitively, the benefits are significant. Many lawyers spend a lot of time interviewing clients to obtain basic information about a situation. A Jnana application performs an intelligent interview that collects all or most of the information a lawyer needs to make a decision. Instead of a series of phone calls or e-mail messages stretching over days if not weeks, all the information can be collected in a single session, at a time convenient for the client. When he or she completes a session, the system automatically informs the Law Department by e-mail.
Deploying the web to help clients and reduce the load on the Law Department makes sense. Unfortunately, the first phases of Web technology did not fully succeed in meeting this goal. Approaches based on extracting information from static web documents, while useful, place too heavy a burden on clients. Law Departments can benefit from new technology that delivers not just information, but also answers and advice. By creating a Virtual Legal Advisor on a corporate Intranet, Law Departments can control headcount, free-up department time for higher value added advice, and maintain a better audit trail.
Overview of Jnana Technology
Jnana Technologies Corporation provides software that quickly and easily captures knowledge and re-uses it to solve specific problems. Jnana applications work in a browser, so sharing knowledge and providing advice within an organization or across the Web is easy.
The key to deploying Jnana self-service solutions is rapid knowledge capture. No programming skill is necessary. A visual “drag and drop” authoring tool makes it easy to create and maintain applications. With a day or two of training, whoever has know-how — product specialists, financial analysts, service representatives, and lawyers among others — can learn to create a Web-based self-service application.
Jnana software is built on Internet standards and developed entirely in Java for optimal Web performance. The user interface works in any standard browser such as Netscape or Internet Explorer. Applications integrate with most e-mail and database systems.
Advertising Copy Advisor
Advertisements must comply with numerous laws and regulations. Competitor comparisons, sale prices, and claims of novelty or recency are just a few examples that can cause legal problems.
Heavy advertisers face a dilemma. Releasing ads without legal review is a risky proposition. But reviewing every ad is expensive and slows down business operations.
Blake Dawson Waldron Lawyers, an AMLAW 50 Asia/Pacific law firm with headquarters in Sydney, helped a large, multi-national packaged goods maker solve the dilemma. Working with its client, BDW used Jnana software to build an advertising copy compliance advisor. (Australian and American advertising law is similar.)
When brand managers create new ad copy, they open a web browser and use the Virtual Ad Advisor. This application asks a series of questions and analyzes the advertising copy. The user learns of tentative legal conclusions immediately and, based on these, has an opportunity to modify the copy. At the end of a review “session”, the system automatically sends a streamlined report to in-house counsel.
The system prioritizes multiple reviews so that in-house counsel can work on the urgent ones first. The streamlined report usually provides all of the information counsel needs to reach a final conclusion concerning the ad copy.
The use of this Web Advisor has several benefits. First, by collecting the relevant information and reaching tentative conclusions, much less lawyer time is required. BDW’s client was able to avoid hiring another in-house lawyer. Second, the Web Advisor is always available, so business people do not have to find a lawyer at his or her desk. And third, the Web Advisor, while less creative than a human, is very consistent. The consistent analysis, combined with a complete audit trail, is useful in the event of a legal claim against the company.
When designing new products and or modifying existing ones, companies must be careful not to infringe others’ patents. Catching an infringing design late in the product cycle can mean costly re-design or, worse, patent litigation.
Companies face a dilemma. Pushing forward with new designs without clearance is risky. But reviewing all new designs is costly and wastes valuable time in moving products to market.
Patent counsel at General Electric resolved this dilemma by working closely with Jnana to create a Virtual Patent Advisor (“VPA”). Engineers or R&D staff runs the VPA from a desktop browser. The VPA asks the inventor a series of questions, much as a patent lawyer would. After collecting initial information, the VPA helps the inventor conduct an intelligent search of an Internet database of patents from around the world.
The VPA interacts with the inventor to determine whether or not patent lawyers at GE need to review the matter immediately. If review is required, the VPA creates a streamlined report for patent counsel. The report includes answers from the virtual interview, tentative conclusions, and the annotated patents, all in a single, easy to use electronic file. Moreover, the system prioritizes the clearance reviews based on business need, seriousness of legal issues, and timing issues.
The VPA has several benefits. First, it saves time. “What used to take a month or more is now reduced to less than an hour” reports Robert Lampe, Senior Patent Counsel at General Electric Power Systems. Second, products can move to market faster because less elapsed time is required for clearances. Third, the VPA lets patent counsel focus on higher value work, letting the lawyers spend more time on the truly difficult clearances. And fourth, it encourages early and frequent inquiries by inventors because it is easy to use and can be consulted without human legal assistance.