EVEnt Retrieval Based on semantically Enriched Structures for interactive user Tasks (EVERBEST) is a research project funded by the Unity Through Knowledge Fund under the Crossing Borders Grant (1B). In a nutshell in EVERBEST we focus on researching event-focused information needs of general public and professionals (journalists and financial analysts) and developing event-based representations of news content and generalization operators that are optimal for satisfying these information needs. Why? Because we believe that efficient search and recommendation over events from news demand technology far beyond traditional keyword-based text retrieval and recommendation methodology. EVERBEST is a collaborative project between the project leader, Text Analysis and Knowledge Engineering Lab (TakeLab), Faculty of Electrical Engineering and Computing, University of Zagreb and partners: School of Computer Science, University of Nottingham, AI Lab, Jožef Stefan Institute, and Institute for Natural Language Processing, University of Stuttgart.

Analysis of event-based information-seeking tasks

What is an event? Is it the same thing for a journalist as for a stock broker? What aspects of events are relevant to users? EVERBEST aims to provide answers to these questions!

Read More

Modelling events and information needs

In EVEREBEST we aim to identify optimal event representations for particular information-seeking tasks as well as generalization operators to model underlying information needs.

Read More

Event-based retrieval and recommendation engine

The end results of the EVERBEST project will be the event-based retrieval and recommendation engine built on top of EventRegistry, an existing news-feeding platform.

Read More

Project News

Paper accepted for LAW 2017!

Our paper, entitled “Two Layers of Annotation for Representing Event Mentions in News Stories“, has been accepted at the 11th Linguistic Annotation Workshop (LAW 2017), the annual workshop for the ACL Special Interest Group on Annotation (SIGANN), co-located with the European Chapter of the Association for Computational Linguistics (EACL), 3-7 April, 2017, in Valencia, Spain.

Paper accepted for CIKM 2016!

Our paper entitled “Detecting and Ranking Conceptual Links between Texts Using an External Knowledge Base” has been accepted for publication at the 25th ACM International Conference on Information and Knowledge Management (CIKM 2016) that will be held in Indianapolis in October. In this work we proposed a novel algorithm for identifying conceptual relatedness between news stories based on an external knowledge base.

Exploratory user study completed

Researchers from TakeLab, University of Zagreb and University of Nottingham have successfuly completed the first set of interviews as part of the initial exploratory study on user perception of news and events. The study included interviews with seven general news consumers and 12 information workers (journalists and business analysts).