Digital White Papers

July 2013: Knowledge Management

publication of the International Legal Technology Association

Issue link: https://epubs.iltanet.org/i/143561

Contents of this Issue

Navigation

Page 17 of 61

THE PRICING PROFESSIONAL'S KM TOOLKIT way firms are solving this is by developing process maps of legal work that serve, among other things, as budget templates for new opportunities and then tracking work performed against the more granular tasks. While process maps are an ideal solution to the problem, they are also time-consuming to create and require a significant amount of vetting by the lawyers. Fortunately, technology can reduce the burden involved in these tasks. Unlike the past, today's clients often ask for budgets and fee estimates before engaging a law firm for a new matter. For the law firm, that often means identifying the detailed tasks that will occur and estimating the amount of time to complete them. Without process maps as a starting point, pricing professionals fall back on finding comparable matters to understand the tasks and effort that normally occur. All too often, one or more of the comparable matters were not billed using any task codes. When that occurs, the pricing professional finds himself exporting all of the time entries and narratives to Excel and manually grouping them into a task structure that lends itself to creating a budget. The process is manual, tedious and difficult to reuse in larger analyses. Bryan Cave has created an approach to solving this problem through custom software in which: Pricing is the overlap in the Venn diagram between accounting, knowledge management and information technology. •All data are stored inside SQL Server for easier reuse •Manipulation of the time entries into different tasks is performed through a website •Most important, a Bayesian classifier is used along with other algorithms to enable the computer to start suggesting task codes for each time entry A Bayesian classifier is a probabilistic pattern recognition algorithm that leverages a training library to teach it how to make educated estimates on the likelihood a particular phrase should be associated with a given outcome. For example, it is frequently used in junk email filters to identify spam. In Bryan Cave's case, it estimates the likelihood a particular narrative should be associated with a given task. Since the system requires training to improve accuracy, an analyst still needs to manually

Articles in this issue

Archives of this issue

view archives of Digital White Papers - July 2013: Knowledge Management