Thomas Demeester

On the robustness of event detection evaluation

2015

Research on evaluation of IR systems has led to the insight that a robust evaluation strategy requires tests on a large number of events/queries. However, especially for event detection, the number of manually labeled events …

UGent-T2K at the 2nd DialDoc Shared Task: A Retrieval-Focused Dialog System Grounded in Multiple Documents

2022

This work presents the contribution from the Text-to-Knowledge team of Ghent University (UGent-T2K) 1 to the MultiDoc2Dial shared task on modeling dialogs grounded in multiple documents. We propose a pipeline system, comprising (1) document retrieval, …

Adversarial training for multi-context joint entity and relation extraction

2018 · Ghent University Academic Bibliography (Ghent University)

Adversarial training (AT) is a regularization method that can be used to improve the robustness of neural network methods by adding small perturbations in the training data. We show how to use AT for the …

A Self-Training Approach for Short Text Clustering

2019

Short text clustering is a challenging problem when adopting traditional bag-of-words or TF-IDF representations, since these lead to sparse vector representations for short texts. Lowdimensional continuous representations or embeddings can counter that sparseness problem: their …