In the first half of our discussion on rolling collections, we outlined an overview of the typical collection and review process. We identified one of the common challenges of legal review, namely, what to do when some document review has taken place and more data must be added to the collection of documents. Should a separate project be created for the newly collected data or can the newly collected data simply be added to the ongoing review project? If the latter, how do you ensure a legally defensible review has taken place?
In Part 1, we ended with a discussion on how to determine if two data sets can be merged into one review project based on analysis of the statistical fingerprints of the data sets. Now we turn to the more pressing question: How do you determine if the existing training set will work effectively on the newly collected data.
