1 of 19

The innards of BAILII:�opening internal interfaces

Roger Burton West, BAILII

2 of 19

Objectives

“Ignorance of the law is no excuse”
Free public access to primary materials
Not competing with commercial services
Public-domain material
No large-scale added value

3 of 19

Deep linking

Part of original vision for WWW
Actively discouraged by many sites
Users must run own searches
BAILII encourages direct links

4 of 19

Stable URLs

BAILII maintains fixed URLs
Some people are still using URLs from 2002

5 of 19

Coverage

Not comprehensive, but good post-1998
Higher courts only
Full-text
Open formats
OpenLAW: significant historical cases

6 of 19

Searching

SINO search engine
Boolean capability
Not domain-specific
OpenLAW indices

7 of 19

Automated citation markup

HTML version only
Markup on demand
Regexp and database
Not 100% accurate

8 of 19

Automated citation markup

HTML version only
Markup on demand
Regexp and database
Not 100% accurate

/^(.*?)

(\[?\d+\]?

(?:\s+

(?:

[-A-Z_\/\d]*\d[-A-Z_\/\d]*|

\d+$[A-Z][_\sA-Za-z]+$|

$?[A-Z][A-Za-z']+$?|

[A-Z.]+

)

){2,}

)

(.*)$/sx

9 of 19

find_citations.cgi

Same code as for in-line citation markup
Marks up submitted HTML documents
Usable from web form
Also usable programmatically:�curl -o out.html -F infile=@in.html http://www.bailii.org/cgi-bin/find_citations.cgi
Or from any HTTP POST

10 of 19

check_citation.cgi

Given citation, lists URL
Also lists alternative citations
If no match as entered, checks stripped form

11 of 19

Citation database

VNCs loaded automatically
Other citations entered manually
Forward maps: citation to URL
Reverse maps: URL to citation

12 of 19

Paragraph tags

Wherever official paragraph numbers exist
<A NAME=''para1''>

13 of 19

Stored searches

IAT lists by country
Can be generated for any key words
Full search functionality is available

14 of 19

Users

Academics (especially students)
Practising lawyers
Paralegal and equivalent
Laymen

15 of 19

Small but fast

Two permanent staff
Rapid uploading of new material
Largely automated (custom Perl)
All free software

16 of 19

Cross-LII linking

Protocol agreed in November 2004
BAILII: 128,000
CanLII: 2,145
other LIIs: <100

17 of 19

Dispersed servers

BAILII hardware is fully redundant
But dependent on a single network connection
Dispersed servers can serve as local caches
...and take load when main servers are unavailable
DNS-based load balancer under development