Marking up narrative text in court documents using the JXDM

["Rolly Chambers" <rlchambers@smithcurrie.com>]

Rex, et al.

This is a long message. 

A while back I experimented with marking up a few simple court documents
using the JXDM. I wanted to see if the JXDM XML vocabulary could be used to
markup narrative text information typically appearing in printed form court
documents as well as the data items that have been its primary focus. I
figured if the JXDM provided reasonably adequate markup tags for the
narrative text in court documents, then by using an xslt/xslfo style sheet,
it would be possible to publish/display an XML court document instance
marked up using the JXDM. A single XML court document instance thus could
contain both marked up narrative text and marked up data elements and would
satisfy both uses -- publication/display for people and data
interchange/automated processing for machines. The same XML instance that a
machine "reads" would be the same one that a person reads.

Court documents are a mixture of both structured data and narrative text. I
continue to be concerned that placing narrative text in a style sheet and
placing only the structured data elements in a xml instance separates the
legally meaningful narrative text information, such as an order or a waiver,
from the associated data elements, such as a person's name and identifying
info, the date, the case, the court, etc. Such a separation no doubt makes
sense from a data-centric perspective, but it is troubling from a
document-centric point of view. 

Assembling the contents of a human readable document from different XML
sources and outputting them in a single display document for human
consumption is technically viable. However, this approach raises issues in
terms of information storage (how and where would all the different XML
sources be stored), information integrity (how would the integrity of all
the different sources and the output document be secured), and reliance on
applications to correctly associate the appropriate narrative text with the
appropriate data items (how would the human readable document be generated
correctly if applications for associating the narrative text with the data
elements become obsolete or unavailable). Placing the data elements and the
narrative text into a single XML instance alleviates these concerns. 

Based on my initial experiment, I believe the JXDM XML vocabulary can be
used to markup up narrative text as well as data elements in simple
form-type court documents. The JXDM semantics work reasonably well for
narrative text components as well as for data items in such court documents.
Attached are five "proof of concept" XML instances (two subpoenas, two
warrants, and a show cause order) marked up using the JXDM. These examples
illustrate one way that JXDM elements can be used to mark up all the
information contained in simple paper court documents -- both the data
elements and the meaningful, legally significant narrative text. 

In general, I followed the process for creating JXDM document schema for
IEP's. First I created a subset of JXDM elements and attributes that I
wanted using the Subset Generation Tool. Then I created an extension schema
for the document. As I experimented, I went back to modify the subset from
time to time. I did not focus on the data modeling step for creating an IEP
because that step is more important for identifying data elements in
designing document schema for data interchange than for identifying text
elements. The data modeling step is important -- I just didn't need to focus
on it for the "proof of concept" experiment. I was more concerned with the
narrative text parts of a court document and the JXDM elements available for
marking up narrative text. If you are interested, I can provide the document
schema I came up with.

To make sure I included appropriate narrative text in the example documents,
I worked from .pdf versions of paper form documents from Maricopa County,
Arizona. John Messing had sent them to me a few months ago and they were
simply handy for purposes of the experiment. The Arizona court document
forms are substantially similar to those used in North Carolina and probably
to the forms used in Georgia and other states as well. 

The main thing I discovered is that there are JXDM elements that can serve
as reasonably adequate containers to describe the narrative text in
form-type court documents such as subpoenas, warrants, etc. This means that
all the information in form court documents, both data elements and
narrative text, can be marked up using the JXDM and combined in a single XML
instance. The data elements can be exchanged and processed by machines, and
the same contents can be displayed or printed for human consumption. My
results are similar in many ways to the sample protective orders that Dr.
Leff and his students created a few months ago. The principal difference is
that those examples place the narrative text inside XML comments, which can
create issues for some XML parsers and for using xslt/xslfo style sheets to
display the information, while I have placed it in JXDM elements.

One implication is that during the data modeling step in creating an IEP,
someone familiar with the narrative text in various form court documents
should be involved to make sure that narrative text types and elements from
the JXDM are included in the data model along with the appropriate data
elements. This assumes, of course, that the goal is to create a document
model for XML instances that will be displayed and published for human
consumption as well as exchanged and processed by machines.

There were some quirks. The JXDM does not permit the sequence or order of
elements to be changed. Thus the better approach is to structure the
information in the documents such that the narrative text comes first (e.g.
"YOU ARE HEREBY SUMMONED to appear before this Court to answer the Direct
Complaint at") followed by the appropriate data elements containing the
designated date, time, location, etc. A few times I needed to slightly alter
the sequence of the information in the printed forms to better match the
sequence/order of elements in the JXDM. The variances, however, were not
material.

If you want more details, let me know.

Regards,

Rolly Chambers

AZ_sample_subpoena_49.xml

AZ_sample_criminal_subpoena_11.xml

AZ_show_cause_order_48.xml

AZ_arrest_warrant_51.xml

AZ_arrest_warrant_52.xml