eniafe festus ayetiran poster em last jd nov 2013

1
Joint International Doctoral (Ph.D) degree in Law, Science and Technology ENIAFE FESTUS AYETIRAN Doctoral candidate of Joint International Doctorate in Law, Science and Technology Contact details: e-mail.: [email protected]. Phone: +34651599932 AN INTELLIGENT HYBRID APPROACH FOR IMPROVING RECALL IN E-DISCOVERY Conclusion METHODOLOGY CONT. INTRODUCTION E-discovery - the discovery and production of electronically stored information (ESI) sought by an opposing party during litigation for the purpose of evidence to support a legal argument, is an important area that poses difficulties for lawyers, litigants and the entire court all alike. Discovering and producing required document(s) among huge volume of data created and stored electronically in various formats, in repositories is a big challenge which needs to be addressed. Poor recall and precision of production documents can have adverse effect on litigations. We aim to develop a system to address these issues with the overall goal of improving recall of responsive documents while also maintaining a high degree of precision. PROBLEM STATEMENT Query Query expansion Figure 1: Illustration of the E-dicovery System RESEARCH AIM & OBJECTIVES The problem statements have been formulated in the following research questions as follow: !How can we improve recall with a single query irrespective of the terminologies used? !How can we produce accurate query result while avoiding several queries by users in the name of relevance feedback? !How do we produce a scalable system to handle several documents usually involved in E-discovery. !How we handle the heterogeneous nature of document formats within the document collection for appropriate classification? The aim of this research to develop a scalable E-discovery tool that will achieve the following objectives: !Increase the recall of responsive documents !Retrieve responsive documents irrespective of the format !Captures the semantic of the user query and retrieve accordingly without relevance feedback METHODOLOGY www.last-jd.eu User Interface WSD System Indexing &Search System Document Collection Responsive Documents !Development of a technique for word sense disambiguation of query based on semantic relationship among query terms using a knowledge- based approach. !Query expansion using the result of word sense disambiguation for constructing term dictionary for indexing and classification. !Development of a independent document-format platform for reading and indexing documents of varying formats. !Document classification using vector space classification. This involves a lot of processes which includes document-term dictionary development, indexing, dictionary compression etc It is believed that this approach will greatly enhance the ultimate goal of recall in during E- discovery task as most classical challenges have been identified for solution

Upload: tilburg-institute-for-law-technology-and-society

Post on 11-Mar-2016

222 views

Category:

Documents


2 download

DESCRIPTION

 

TRANSCRIPT

Page 1: Eniafe festus ayetiran poster em last jd nov 2013

Joint International Doctoral (Ph.D) degree in Law, Science and Technology

ENIAFE FESTUS AYETIRAN Doctoral candidate of Joint International Doctorate in Law, Science and Technology Contact details: e-mail.: [email protected]. Phone: +34651599932

AN INTELLIGENT HYBRID APPROACH FOR IMPROVING RECALL IN E-DISCOVERY

Conclusion METHODOLOGY CONT….

INTRODUCTION E-discovery - the discovery and production of electronically stored information (ESI) sought by an opposing party during litigation for the purpose of evidence to support a legal argument, is an important area that poses difficulties for lawyers, litigants and the entire court all alike. Discovering and producing required document(s) among huge volume of d a t a c r e a t e d a n d s t o r e d electronically in various formats, in repositories is a big challenge which needs to be addressed. Poor recall a n d p r e c i s i o n o f p ro d u c t i o n documents can have adverse effect on litigations. We aim to develop a system to address these issues with the overall goal of improving recall of responsive documents while also maintaining a high degree of precision.

PROBLEM STATEMENT Query Query expansion Figure 1: Illustration of the E-dicovery System

RESEARCH AIM & OBJECTIVES

The problem statements have been formulated in the following research questions as follow: ! How can we improve recall with a single query irrespective of the terminologies used? ! How can we produce accurate query result while avoiding several queries by users in the name of relevance feedback? ! How do we produce a scalable system to handle several documents usually involved in E-discovery. ! How we handle the heterogeneous nature of document formats within the document collection for appropriate classification?

The aim of this research to develop a scalable E-discovery tool that will achieve the following objectives: ! Increase the recall of responsive documents ! Retr ieve responsive documents irrespective of the format ! Captures the semantic of the user query and retrieve accordingly without relevance feedback

METHODOLOGY

www.last-jd.eu

User Interface

WSD System

Indexing &Search System

Document Collection

Responsive Documents

! Development of a technique for word sense disambiguation of query based on semantic relationship among query terms using a knowledge-based approach. ! Query expansion using the result of word sense disambiguation for constructing term dictionary for indexing and classification. ! Development of a independent document-format platform for reading and indexing documents of varying formats. ! Document classification using vector space classification. This involves a lot of processes which includes document-term dictionary development, indexing, dictionary compression etc

It is believed that this approach wi l l g reat l y enhance the ultimate goal of recall in during E-discovery task as most classical challenges have been identified for solution