13

June

Automation Personified – Online Extraction, The SJV Way

Created by Corey Beavers | June 13, 2017

SJV's Automation Team devotes considerable time and attention to the process of finding, evaluating, and automating criminal court sites across the country. With a library spanning several hundred sites and nearly two thousand counties, the process by which we expand our depth of automated coverage has been refined since we began extracting court site data from the web back in 2010. Of course, not all sites make the cut, and there are many sites, even high quality ones, that have not made it into our library with the coveted label of "Default Process"!

Our method of evaluation can be distilled down to a couple of key insights, and for a site to continue to the point of even basic automation, we're looking for these core criteria to be satisfied:

1. Scope of Data


At the very minimum, a site under consideration must make available all cases that would fall into the scope of the search. As courts across the country have seen the benefit to making their records available online and reducing the burden of supporting court retrievers or clerk involvement, this has become a less common deficiency to discover in court data. Additionally, 3rd party platforms such as Odyssey, CourtView, and Benchmark, have succeeded in helping promote better practices for recording and displaying data in a consistent, well organized manner. Over the years, this has been a criterion that was quite a gap in the early days of automation within the screening industry, but becomes less of a concern with each passing year.

2. Equivalent Coverage


In states were multiple courts must be checked and results combined to complete a check, verifying that the site being evaluated contains data from all required courts is crucial. Up until recently, an excellent example of this was our home county of Cobb County, Georgia. For years, the Superior Court has made their case data available online, but it was not until the end of 2016 that the State Court adopted a platform and released their data for public online viewing. As a rule, if a county requires more than one court to be searched to provide a complete result, SJV will only automate that county if all courts are online. Piecing together a set of results obtained partially online and partially from conventional court research presents its own unique set of challenges.

3. Sufficient Detail


Particularly of importance when evaluating whether a site is appropriate as a Full Extraction jurisdiction, a few key data points must be reliably present. The usual fields such as Name, DOB, Case Number, File Date, Charge Type, Charge Description, Disposition, Disposition Date and Sentencing, must all be clearly identifiable. Supplemental data such as amended information and warrant details must also be identified and collectable with a level of consistency equal to what a field researcher would find in court.

4. Researcher Review


Many sites where we have automation running are also sites used by SJV's Internal Researcher team. When an order fails to successfully complete through automation, or requires an extra human touch to obtain missing data, that is the team that fulfills that role. Consequently, Internal Researchers are intimately familiar with the site from a human knowledge standpoint and are a crucial part of the scoping phase as our automation developers rely on their knowledge to accurately navigate the site, collect the proper data, and simulate the same behavior that a user of the public access terminal would. In many instances, Internal Researchers will have used the site prior to automating, so by the time an Automation Analyst is preparing to build an agent, the site is generally well understood if adopted for use.


Assuming a site can pass the initial review phase described above, the process by which a site will then grow from a possible to an actual agent includes additional vetting that can either further affirm the integrity of the site or cause it to be disqualified from the list of sources we adopt as our default process.