Homepage > Papers |
Methods
for reports processing which reveal the
characteristics of
figurants and incidents
Igor P. Kuznetsov, (Moscow, Institute for Informatics
Problems of the Russian Academy of Sciences)
RESUME
A class of systems is considered designed for processing of reports for
the tasks of Criminal Investigation Departments. The design is based on the fact
that the reports are natural language texts abounding in abbreviations and
synonymous ways of expressing the same content.
This kind of system should be provided with the features of analysis
for ambiguity elimination
at different levels, for restoration of insufficient information and
supplying it with relevant characteristics,
for supporting the variety of terms in the course of similar incidents
and figurants retrieval.
This paper describes an approach to the development of
such systems which consists
in employing a
linguistic processor providing
lexical and morphological
analyses followed by the
subsequent simplest semantic network
formation. The further processing
is completely transformed to the level of such networks and is performed on the basis of logical-analytical
means and methods. The use of this type of networks for representation of content images of documents and for
a terminological dictionary presentation are considered as well as their use
for revealing the incidents and figurant characteristics and for the retrieval
space expansion.