Sailing the Semantic Seas by Structural Vessels: Problems and Perspectives for Knowledge Discovery in the Legal Domain

Sailing the Semantic Seas by Structural Vessels: Problems and Perspectives for Knowledge
Discovery in the Legal Domain


G. Ajani




gianmaria.ajani@unito.it


P. Rossi





p.rouge@inwind.it





In this paper, we
consider
how
the
relationship of L
aw and AI

concerns
the
issue

of knowledge
discovery

and whether new advances in the complexity paradigm, in terms of
social
network
analysis, can be fruitful to knowledge disc
overy in the legal domain
.

Seen from an epistemological point of view,
knowledge discovery in AI theory may reproduce

the
traditional concept of truth as
adequacy between object and cognition

at different levels of world
description
.

T
he current st
ate of
legal research discusses

knowledge discovery
,
either
moulding

diverse
pragmatist, coherent theories of truth,
or
promoting a notion of knowledge construction

instead of
knowledge discovery

in the legal domain.

When se
en from an instrumental point of view,

A
I

model
s

are

a representation of information in a

specific structure, and claims for the model’s effectiveness
do stress

the interface between model
s

and

the world
,

asking for instruments in

the assessment of the represented knowledge.

Since AI models
c
an only arise from
their

employ

by

particular agents in specific
situations
, the
question of effectiveness in the representation is often reduced to the
usefulness

to some
environments or uses
.


The legal contexts are difficult to be grasped in controlled

situations, because of the interconnection
within
and beyond
national jurisdictions
among the different players (judges,
legislators
, scholars)
and
among
the different instances they produce (judicial decisions, statutes, doctrinal comments) .


T
he pictu
re of knowledge
in a social phenomenon, such as law,
can be

symbolized by a network

of
meanings and practices
, where the

notion of
knowledge
as

a matter of
exact
correspondence
between
concepts and world

has

to be discarded.


Ontological mapping in the leg
al domain is thus a difficult
assignment

where a task
-
based approach

cannot be really achieved due to

the overlapping of situations, while domain level approach
encounters the social construction of concepts
that
may diverge among legal orders in considera
tion
of their historical,

cultural and political inheritances.


Despite the theoretical uncertainty about the effective analysis instruments to be implied for, t
he
real challenge

for legal research

introduced

by

the complexity paradigm,

in terms of
socia
l network
analysis
,

seems to be
the
assess
ment of

the impact of
the production of legal discourses on legal
practices and social contexts.



The promises of social network analysis are interesting
even

for

the ontology modelling:
social
network analysis

co
uld improve
, indeed,
the epistemological completeness of
the ontology
mapping, importing a way to assess

the
effectiveness of
knowledge discovery
in the legal domain.


Consequently, we evaluate in details
both the theoretical feasibility of the matching b
etween
semantic nodes and social nodes, and
its

practicability in fields, like the multilingual legal
applications, where the accuracy of the currently implemented
tools for the needs of jurist
s

and
practitioners is

rather

inadequate
.