how phrasefinder works
Attorneys identify search terms that can be used to identify relevant data, but they are doing so in-context. Then, due to the limitations of currently available search technologies, these very same search terms are executed out-of-context, yielding a great number of false-positive results.
PhraseFinder collects five words on either side of the search term and calls this a "phrase," then keeps track of how many unique phrases occur in the indexed universe of data.
PhraseFinder will count how many times each unique phrase occurs and display those results on screen.
The result is a contextually rich view of each of the unique contexts in which we see our search term in the indexed universe of data.
the risks of false positives and false negatives
There is never a 'perfect' set of keywords, and up until now, keyword construction has involved a trade-off between being overly-inclusive and being overly-exclusive.
- false positives: Even the most well-crafted search terms yield false positives because they are missing a crucial component - context. This issue is compounded exponentially when a corporation or entity relies heavily on standard language.
- false negatives: As risky as false positives are, false negatives present more risk. Here, relevant documents and data are neither identified nor collected. The worst case scenario is that the key documents, or even the proverbial “smoking gun,” is not identified.
Imagine Merck trying to identify relevant documents that contain the search term Vioxx. As the name of Merck’s controversial painkiller, Vioxx will yield an amazing amount of false positives because it will exist in standard language as well as within relevant and unique documents that attorneys should be focused on.
Since eliminating Vioxx from the list of search terms would likely result in the unacceptable exclusion of relevant data, the search will yield false positives that result in over-collection, increased review costs and wasted time and resources. The ability to search using the context surrounding search terms such as Vioxx can eliminate false positives.
PhraseFinder allows users to make decisions about search phrases using contextual intelligence. Using our Vioxx example, users can “ignore” any phrase within the data universe that contains the word Vioxx that is not in the correct context. Ignored phrases can include standard language phrases that appear repeatedly within the data universe. At the same time, PhraseFinder allows users to “promote” or highlight phrases where the word Vioxx appears in the relevant context.
See for yourself. Schedule a demo.