Your Bloglet Update from "excited utterances"

"excited utterances - http://excitedutterances.blogspot.com/

What are the guys at Thomson's Research & Development Department up to now? Thomson research scientists Jack G. Conrad and Dr. Khalid Al-Kofahi (along with University of Minnesota computer scientists, Ying Zhao and George Karypis) published their latest technical paper, Effective Document Clustering for Large Heterogeneous Law Firm Collections. Here is an excerpt from the paper's introduction: Many firms currently deploy systems that offer their practitioners both full-text search capabilities and the ability to browse through a general legal classification system such as KeySearch, which is based on common legal practice or research areas (e.g., bankruptcy, intellectual property, torts & personal injury). Useful as such a broad-ranging taxonomy is, however, there are numerous instances when it is rendered incomplete if not irrelevant. For example, some firms offer fewer legal practice areas, but among those provided, they offer them at substantial depth. A broad-based, but shallow taxonomy would thus be inadequate for such expertise. Other examples include those firms that are now practicing areas of law that may not be well covered by an existing taxonomy (e.g., elder law, law involving abortion, gay marriage law, or law involving anti-terrorism). Still other firms find such taxonomies only partially relevant to their portfolio of practice areas. At least in the context of automatic text categorization, the strength of these firms? KM applications will depend upon the relevance of their taxonomies and the availability of exemplar documents for each category. So clearly the incompatibilities outlined above are problematic. In the absence of a relevant taxonomy and corresponding training data, an effective and intelligent document clustering application could prove to be useful. For a list of other publications authored by the members of the R&D Department at Thomson Legal & Regulatory, click here. (8/8/2005 11:01:17 AM)
"