Issue link: https://epubs.iltanet.org/i/7694
ILTA White Paper The Business of Law 16 techniques must be combined and properly applied as an effective response to searches for unstructured data. These techniques include the following: • Automatic Query Guidance As the user enters search terms, the system can look for related terms and searches that may be of interest to the user to help produce a more accurate search. For example, typing "Einstein" might turn up a more appropriate search such as "Albert Einstein." Other relationships provided through a firm-supplied thesaurus and through automated analysis by the search engine might turn up other interesting relationships such as "Enrico Fermi." • Relevancy Ranking It is quite common for the conflicts analyst to be forced to search on several variations of a name using wildcard characters and various spellings to be thorough about finding a conflict. Today, these techniques can be automated, and the results can be automatically scored as to how close they are to the original query without the user having to guess at every variation that can occur within the text. For example, using relevancy ranking, the system can determine that an exact match on "Albert Einstein" is highly relevant, and "Albert went to Einstein Bagels for coffee" may be irrelevant. Relevance can be further tuned to suit the process, such as prioritizing results from active matters as higher in relevance than matters closed more than five years ago. • Entity Extraction: Finding Names of People and Companies By employing various methods of detecting the context of word usage, technology can automatically differentiate proper nouns. For example, it may be important to weight references to "General Electric" or "GE" as much higher in relevance than a document that simply contains the words "general" and "electric." • Automatic Categorization Not all content is equal. A flat search to find documents or data that contain a simple keyword or wildcard pattern may turn up many thousands of hits. Today's technology is capable of automatically determining the context of a document within the overall repository and can be taught about the firm's unique business processes through simple "training" and learning techniques. Once this process is applied, certain categories can be weighted higher than others. For example, the firm may assign a high degree of relevance to e-mail messages referencing "discussion about potential representation" or "do not represent," as this type of dialogue is critical to the intake process but is difficult to determine through traditional keyword searches. Without automated categorization, determining the differences among various types of content depends on the application of structured metadata to each document, which is both laborious and typically incomplete. These technologies, and many others, when carefully applied to the conflict of interest process, increase the effectiveness of the conflicts analyst and attorney evaluating new business. The speed at which conflicts are resolved and new business is approved is ever more important. Advances in meaning-based computing reduce risk by increasing the scope of due diligence without overwhelming the process. iLTa