Agents

1
iSOCO
The Semantic Web Legal Application
Dr. V. Richard Benjamins iSOCO
May 2005
2
Iuriservice SW in the legal domain

• Final user
• Consejo General del Poder Judicial
• General Council for Juridical Power
• Intelligent FAQ for recent graduated judges
• Criminal Process Law questions, Office practice
  (Non binding)
• Technologies
• NLP
• Ontology of legal domain
• Semantic distance between new questions and
  stored ones

3
Identifying the problem domain
A survey covering interviews with gt 250 judges
4
Using textual statistics method
5
Example FAQs

• Questions about data and facts
• What should be done when a detainee accused of a
  minor crime which does not imply jail is brought
  to a judge and the public defender who has been
  designed delays his arrival to the judicial
  premises?
• What should be done, given the case that a mother
  files a complaint against her son whom she wants
  to exclude from home for ill treatment and
  cruelty?
• A corpse is produced at home, at 2 o'clock a.m.
  During the legal search, millions of pesetas
  appear in a drawer in the bedroom. What should
  the judge do in this case?
• Questions related to legal culture
• In which cases is a judge on-duty allowed to send
  a detainee to jail, when he depends on another
  court?
• What must a judge on duty do when a detainee with
  several pending criminal offences in different
  courts, if they do not belong to the same
  jurisdiction?

6
Specific inputs for the ontology
Process is a fundamental notion
7
Specific inputs for the ontology
isa
Part of
Followed by
Level 1 is-a Level 2 part-of Level 3 followed
by
8
Current version of the ontology

• Covers procedural knowledge of criminal law in
  Spanish
• Three main concepts
• Process (e.g. charge)
• Object (e.g. indictment)
• Actor (public prosecutor)
• Relations
• Generalisation (is_a)
• Equivalence or similarity (synonyms/antonym)
• Actor participates in process
• Follows on Processes
• Part of (processes, objects)

• Standard process (follows)
• Formal Complaint
• Charge
• Filing a suit
• Institution of proceeding
• Oral trial / Stay a case
• Sentence
• Appeal
• Each sub process has its associated objects and
  actors

9
Legal ontology
10
The application an intelligent FAQ

• Analyze input sentence using NLP techniques and
  link terms in the sentence to the ontology.
• Calculate the semantic distance between the set
  of terms of the input sentence and the set of
  terms of the stored questions.
• Stage-based Search process

11
Technological considerations
iFAQ System
Multistage Searching Subsystem
Accuracy
Ontology graph path matching
Eficiency
Ontology domain detection
Keyword matching
Natural Language Processing
Ontology Technology
Caching subsystem
Persistence subsystem
12
Multi Stage Search System
Ontology graph path matching
Keyword/synonym matching stage
User Question
FAQ Candidates
iFAQ Search Engine
Other search engines ...
Search Factory
13
Ontology Domain Detection
Multi Stage Search System
Ontology Engine
NLP Engine
Morphological engine
Statistical Occurrence algorithm
Relevant words
Thesaurus engine
User Question
Searching System
14
Keywords/synonyms matching stage
Multi Stage Search System
15
Ontology grammatical pattern graph path matching
stage (I)
Multi Stage Search System
16
Ontology grammatical pattern graph path matching
stage (II)
Multi Stage Search System

• Minimum distance between grammatical pattern
  graph path

FAQ Question What are the actions to follow in
case a mother denounce one of her own sons ...?
User Question Can a mother accuse to her own son
...?
Ontology
Denounce
Actions
Grammatical Pattern Actions Follow mother denounc
e son ...
Grammatical Pattern Mother Accuse Son ...
Accuse
Follow
17
Semantic Distance calculation
Multi Stage Search System

• Semantic distance is based on weighted path
  length between concepts
• Path length is based on navigation from one
  concepts to another through any relation
  available
• Is-a
• Part-of
• Follows
• Actor
• Etc.

18
(No Transcript)
19
Contact Information
Intelligent Software Components, S.A.
For more information on iSOCO, please contact us
at isoco_at_isoco.com www.isoco.com
iSOCO Valencia Tel [phone] Oficina 305 C/
Prof. Beltrán Báguena 4, 46009 Valencia Spain
iSOCO Barcelona Tel [phone] C/ Alcalde
Barnils 64-68 St. Cugat del Vallès 08190
Barcelona Spain
iSOCO Madrid Tel [phone] C/Pedro de
Valdivia 10 28006 Madrid Spain
Back