Speaker: Tommaso Caselli
Seminar title: Why did the building collapse?: Ask for the Storylines!
Abstract:
This talk will present the principles and the pending issues related to the creation of storylines from large amounts of textual data. A storyline is best described as coherently ordered structure of events and participants, reconciling data from multiple sources, about a specific topic. We introduce the notion of explanatory relations (i.e. plot links) to organise the flow of information in terms of narrative structures (plot structures), thus trying to minimise human intervention. The talk will be divided in two parts: in the first one, the notion of storyline will be described together with a first version of the computational model which informs its representation and implementation. In the second part, issues related to the actual realisation of the model will be presented through experimental results.
Speaker’s Bio:
Tommaso Caselli is a Senior Researcher (Postdoctoral Fellow) at the Computational Lexicology & Terminology Lab at the Vrije Universiteit Amsterdam. He received his PhD in Computational Linguistics on temporal processing of texts from the University of Pisa in 2009. He has been involved in research in NLP since 2006. His main research areas are in discourse processing, temporal relations, and (event) sentiment analysis. He is currently working on the SPINOZA-NWO Project Understanding Language by Machines – Stories and world views as a key to understanding language (ULM-3), which aims at developing computational models and NLP tools for storyline extraction from news. He took part to the organisation of semantic evaluation campaigns in NLP for English and Italian (SemEval 2010 TempEval-2; EVALITA 2014 EVENTI, SemEval 2015 CLIPEval, EVALITA 2016 FactA). His work has been presented at EMNLP, COLING, LREC, EACL and elsewhere.
Location:
Sala Direzione, Primo Piano, Edificio Ovest
Website:
