סמינר מחלקתי - ניר אופק

Abstract

The unprecedented popularity and evolution of communication platforms has led to an 

abundance of user-generated content across the Web. Mining and aggregating this information 

across opinion-rich resources help to address important questions and provide organizations with 

information vital for decision making. Opinion mining—the computational detection and study 

of opinions and viewpoints underlying a text span—enables organizations to provide significant 

insight into opinions without having to directly survey populations, a time-consuming and 

expensive task. However, opinion mining is considered a challenging task because text, despite 

being perfectly suitable for human consumption, is largely unintelligible for machines, 

particularly so for short user-generated content. This talk will highlight the challenge of opinion 

mining in short text spans, and suggest novel statistical and machine learning methods we 

developed in addressing these challenges. Additionally, it will discuss how such methods help 

organizations to identify issues that bother their users.

Bio

Nir Ofek wrote his PhD on granular opinion mining at the Department of Information System Engineering, Ben-

Gurion University, Israel under the supervision of Lior Rokach. He received there his M.Sc. and B.Sc. degrees 

under the accelerated M.Sc. program for outstanding students, specializing in AI search. His research focuses 

on machine learning in mining sequential data.