1. Similarity Analysis on Government Regulations Gloria Lau, Kincho Law, Gio Wiederhold {glau, law}@stanford.edu, gio@db.stanford.edu Stanford University  ABSTRACT Government regulations are semi-structured text documents that are often voluminous, heavily cross-referenced between provisions and even ambiguous. Multiple sources of regulations lead to difficulties in both understanding and complying with all applicable codes. In this work, we propose a framework for regulation management and similarity analysis. An online repository for legal documents is created with the help of text mining tool, and users can access regulatory documents either through the natural hierarchy of provisions or from a taxonomy based on concepts generated by knowledge engineers. Our similarity analysis core identifies relevant provisions and brings them to the user’s attention. The similarity analysis is performed by utilizing both the structure and referencing of regulations to provide a better comparison between provisions. Preliminary results show that our system reveals hidden similarities that are not apparent between provisions based on node content comparisons.  COMPLEXITY OF REGULATIONS: CONFLICTING PROVISIONS ADA Accessibility Guidelines 4.7.2: Slope Slopes of curb ramps shall comply with 4.8.2. The slope shall be measured as shown in Fig. 11. Transitions from ramps to walks, gutters, or streets shall be flush and free of abrupt changes. Maximum slopes of adjoining gutters, road surface immediately adjacent to the curb ramp, or accessible route shall not exceed 1:20. California Building Code 1127B.5.5: Beveled lip The lower end of each curb ramp shall have a ½ inch (13mm) lip beveled at 45 degrees as a detectable way-finding edge for persons with visual impairments.  DATA CONSOLIDATION AND CATEGORIZATION Concepts noun phrases extracted from documents, e.g.access aisle, toilet stall Definitions designated section in regulation to define terminologies used clear unobstructed Measurements defines most conflicts in accessibility regulations References explicit linking to other provisions Glossary Terms similar to definitions, but from references or handbooks in the field Author-Prescribed Indices similar to concepts, but from references or handbooks in the field  FEATURE EXTRACTION  XML REGULATION STRUCTURE Original Section 4.6.3 from the Uniform Federal Accessibility Standards (UFAS) 4.6 Parking and Passenger Loading Zones 4.6.1 Minimum Numbers. … … 4.6.3 Parking Spaces Parking spaces for disabled people shall be at least 96 in (2440 mm) wide and shall have an adjacent access aisle 60 in (1525 mm) wide minimum … and shall comply with 4.3 … Exception: If accessible parking spaces for vans... Refined Section 4.6.3 in XML Format … … Parking spaces for disabled people shall be at least 96 in (2440 mm) wide... If accessible parking spaces for vans designed for handicapped persons... …  SIMILARITY ANALYSIS  APPLICATIONS: REGULATION COMPARISONS & E-RULEMAKING UFAS 4.13 Doors 4.13.1 General … 4.13.9 Door Hardware Handles, pulls, latches, locks, and other operating devices on accessible doors … … British Standards 8300 12.5.4 Doors 12.5.4.1 Clear Widths of Door Openings 12.5.4.2 Door Furniture Door handles on hinged and sliding doors in accessible bedrooms should be … ADAAG Rights-Of-Way Draft 1105.4.1 Length Where signal timing is inadequate for full crossing of all traffic lanes or where the crossing is not signalized … Public Comment Deborah Wood, October 29, 2002 … This often means walk lights that are so short in duration that by the time a person who is blind realizes they have the light, the light has changed … ADAAG Rights-Of-Way Draft No relevant section identified Public Comment Donna Ring, September 6, 2002 If you become blind, no amount of electronics on your body or in the environment will make you safe and give back to you your freedom of movement. You have to learn modern blindness skills from a good teacher … Comparisons between American and British Regulations E-rulemaking: Comparisons between drafted rules and public comments Base Score feature matching f o = (  i = features f i ) / # features i Neighbor Inclusion psc(A) = parent, siblings and children of node A  = discount factor of update Set S = f o (A,psc(B)) U f o (psc(A),B) N = sizeof(S)  GT =  s>fo(A,B) (s – f o (A,B)), s  S if (N  0) f psc (A,B) = f o (A,B) +   (  GT /N) else f psc (A,B) = f o (A,B) Reference Distribution ref(A) = out-references from node A ref(A) replaces psc(A) from the above