Σεμινάριο του Τομέα Λογισμικού: Κύκλος διαλέξεων 2010-2011
Τρίτη 05/10/2010, 14:30 (Ναι, αυτή τη φορά θα γίνει ΤΡΙΤΗ, λόγω της αργίας της Δευτέρας 4/10/2010)
Εργαστήριο Τομέα Λογισμικού, Ισόγειο κτιρίου Α, Πανεπιστημιούπολη
Ομιλητής: Αυγερινός Αραμπατζής, Λέκτορας
Τίτλος: Modeling Score Distributions in Information Retrieval & Applications
Περίληψη:
Modeling score distributions in Information Retrieval (IR) is an
important theoretical issue which has many practical applications
in Distributed Retrieval, Fusion/Meta-search, Filtering, Topic
Detection & Tracking, and elsewhere. We review the history of
modeling score distributions, focusing on the mixture of Normal-
Exponential by investigating the theoretical as well as the empirical
evidence supporting its use. We discuss suggested conditions
which valid (from an IR perspective) binary mixture models should
satisfy. Last, we present our related on-going work on
Legal Document Retrieval, and Multi-Modal Image Retrieval.
Βιογραφικά Στοιχεία:
Avi Arampatzis graduated from the Computer Engineering
and Informatics dept., University of Patras, Greece, in 1996.
He then moved to the Netherlands where first he conducted
his PhD in Adaptive Document Filtering at University of Nijmegen,
and then worked as a researcher in various academic institutions
such as Utrecht University and University of Amsterdam,
as well as an independent consultant. Currently, he is a Lecturer
in Databases at DUTH. His interests are in the general field of
Information Systems and include: Information Retrieval
theory and models, Classification, Natural Language Processing,
Data Mining, and Databases.