Technology
IT Discovery addresses the challenges of searching electronic documents, particularly email and its attachments. Those challenges fall into two categories: the volume of data is too great and keyword search is inadequate. When doing an internal investigation or discovery in litigation, it is extremely difficult to find emails that are relevant to the investigation. Even among a small set of emails, there is typically an enormous volume of email to go through, most of it irrelevant. IT Discovery allows the investigator to sort through the large volume of emails and identify those that are relevant to the investigation. IT Discovery's technology is not based on word search. It is based on the identification of social networks from email conversations, and recognition of word usage patterns in those conversations.
Corpus Reduction: Automatic Culling
As the investigator goes through an initial set of emails, IT Discovery allows the reviewer to identify emails that are potentially interesting, but more importantly those that are not interesting. As relevant and irrelevant documents are identified, IT Discovery uses machine learning technology to automatically classify the rest, hiding from view emails that are most likely not germane and highlighting those that are. As the investigator continues to review emails, identifying them as relevant or irrelevant, the system learns to distinguish relevant from non-relevant emails simply from their use of IT Discovery.
Topic Imputation - Finding What You Want Using Topics
Assume that the uninteresting emails are set aside, what ought one do with the typically large data set that remains? IT Discovery begins with a simple fact: emails are not undifferentiated discreet documents but nearly always a series of conversations about some topic. An enterprise's corpus of email consists of many sets of conversations among different people within and outside the enterprise. Unlike clustering methods, IT Discovery uses knowledge about these conversations to help generate topics.
The definition of topics and the identification of emails in particular topics is far more informative than what can be accomplished with word search alone or with clustering or with "concept searching" techniques. IT Discovery derives topics from sets of words and phrases that relate to a salient subject discussed in the emails, such as "tax audit" or "corporate restructuring". The derivation of topics is informed by an understanding of the social network in the enterprise. Phrases may be placed in the same topic not only because they frequently appear together within messages, but also because they are often written or received by the same person who is frequently involved in conversations about this topic. If you look at your inbox inside an enterprise, it defines your place and your role in that enterprise. This is what social networks bring to topics: technical conversations take place among technical people, compliance issues among compliance people, and so forth.
Once the system has separated the email corpus into topics, one can choose a person and ask what topics he authored, or choose a topic and ask which people wrote about it. By focusing on the interaction between topic and author, IT Discovery can much more accurately find the relevant communications from large volumes of communication.
Intuitive Interface
IT Discovery places the entire corpus and all functionality on a single page, with no extraneous query builders or different "views," and no need to go elsewhere for different functionality. A single search box with a single set of results is replaced by a more dynamic and intelligent "global" view of the corpus, all the while charting the results on an interactive timeline. The normal tagging, saving of searches, assignment to reviewers, flagging of events and so forth are incorporated into this single view.
Integration
IT Discovery is indifferent to the form in which email is produced and therefore, proprietary formats such as ".pst," ".msg" and ".nsf" aren't a challenge. IT Discovery can also export emails in various formats.