Stuart Shulman
OUTLINE POWER
POINT
Technological Innovations and Tools to Enable Public Participation and
Manage Public Comments
Jamie Callan - Carnegie Mellon University, Eduard Hovy - USC/Information Sciences
Institute Stuart Shulman - Drake University, Stephen Zavestoski - University
of San Francisco
Technology: An Overview
Our objectives:
Help agencies handle the information flood
Help produce more defensible regulations
Language Technologies
Computers can sometimes do useful things with human language
The systems that do these tasks are not perfect
They enable people to work with far more information
E.g., carpenters and power saws
Simple Statistical Techniques
Recognize words and phrases
Count how often things occur
In general or as compared to a reference model
(e.g., general English)
Duplicate Detection Solutions
Duplicate detection algorithms
Generate summary counts and identify the reference
copy
Automatic Stakeholder ID
Useful to know which stakeholder groups are represented in the comments
"As a family farmer, I believe…."
"As a long-time consumer of organic foods, I
am horrified…"
Clustering By Opinion
For each (sub)topic
Group together all Yes, No & In-between
Extract reasons/motivations/authorities using
characteristic phrases
Summary of Modules
Search
Duplicate detection
Near-duplicate detections
Frequent concepts, as drill-down points
Finding all substantively different material
Stakeholder identification
Relationship organization
Linking comments to proposed regulations
Clustering by opinion
Summarization
Acknowledgement: This research is supported by exploratory grants from the National
Science Foundation, EIA-0089892 SGER Citizen Agenda-Setting in the Regulatory
Process: Electronic Collection and Synthesis of Public Commentary and EIA 0327979,
0328175, 0328914 & 0328618 SGER Collaborative: A Testbed for eRulemaking Data.
Any opinions, findings and conclusions or recommendations expressed in this
material are those of the authors and do not necessarily reflect those of the
National Science Foundation.