Cataphora - Investigation

Investigation

Here at Cataphora, "analysis" goes way beyond simple counting of documents or high school statistics about average numbers of email messages sent.

Modeling, not counting

Unlike some others, we don't just count, for example, how many Japanese language Microsoft Word documents your data contains. For us, analysis would mean determining how many revisions of each type of document should be expected, by whom, when, whether they would normally be translated, and how many seem to be absent. The word analysis in most dictionaries refers to modeling complex reality so as to make it more understandable. That's what we've been providing at Cataphora for almost eight years now. Based on our experience, we know that what's not there is often more important than what is. And we know that the most interesting pieces of data often require the greatest amount of context to be properly understood. For example, even a mild sense of frustration expressed by a normally calm individual is probably of far more interest than a routine fit of pique unleashed by someone who is invariably short tempered. More generally, one must first know what is normal before any piece of data can be properly viewed as being anomalous. This is the essence of what a model is, and exactly what our COBRA (Cataphora Organizational Behavior Risk Analysis) technology provides.

Patterns of behavior

Covered now by three patents, this advanced technology begins with the premise that today's data increasingly exists in very small fragments, such as the two sentence abbreviation-laden message you dash off from your BlackBerry, and that these fragments must be analyzed within a broader context in order to be correctly interpreted. No longer can one assume that helpful keywords will appear in the same fragment, making its topic easily identifiable. But if these fragments seem to be ever shrinking in size, the number of them is exploding, resulting in an increasingly detailed and complex set of electronic breadcrumbs that mark our daily activity.

These many individual electronic breadcrumb trails form the basis of our underlying multi-dimensional behavioral model, which allows the regularity of behavior to be readily computed. Most people are very much creatures of habit, especially when considered in an organizational context. By being able to establish patterns of behavior, we become able to distinguish abnormality. This is critically important because while not all abnormal behavior is necessarily bad, the vast majority of really bad behavior is quite statistically abnormal.

Our analytics algorithms and their accompanying visualizations have been used by many leading law firms for early case assessment, to vetting different case theories, to preparing for depositions and trial to creating trial exhibits. Based on graph theory, statistics, social networking theory, computational linguistics and multi-dimensional scaling technology, our analytics truly merit the use of the term.