Papers by Jelke Bloem
Fusion Training for Mathematical Generalization in Large Language Models (2026.acl-srw)
Copied to clipboard
| Challenge: | Existing efforts to improve reasoning efficiency have focused on switching between simple queries and complex problems. |
| Approach: | They analyze the effects of the training schedule and data ratio between thinking and non-thinking modes and construct a benchmark to test their theory. |
| Outcome: | The proposed model unifies a thinking mode and a non-thinking mode within a single model. |
Automatic Animacy Classification for Romanian Nouns (2024.lrec-main)
Copied to clipboard
| Challenge: | Animacy is a semantic property of nouns that describes the quality of the noun's referent of being alive, sentient or volitional. |
| Approach: | They propose a type-based binary classifier of Romanian nouns into the classes human/non-human using pre-trained word embeddings and animacy information derived from Romanian WordNet. |
| Outcome: | The proposed classifiers perform well on the Romanian language and in a naturalistic setting. |
Impact of Task Adapting on Transformer Models for Targeted Sentiment Analysis in Croatian Headlines (2024.lrec-main)
Copied to clipboard
| Challenge: | BERT models are often taken off-the-shelf and fine-tuned on a downstream task. |
| Approach: | They propose an extra stage of self-supervised task-adaptive pre-training to perform a task on a number of Croatian-supporting Transformer models. |
| Outcome: | The proposed approach improves performance across multilingual models but not in Croatian-dominant models. |
Expert Concept-Modeling Ground Truth Construction for Word Embeddings Evaluation in Concept-Focused Domains (2020.coling-main)
Copied to clipboard
| Challenge: | a novel method for concept-modeling ground truths is developed to evaluate the application of word embeddings in concept-focused textual domains. |
| Approach: | They propose a domain expert-controlled procedure for the construction of concept-modeling ground truths with the aim of evaluating the application of word embeddings. |
| Outcome: | The proposed method evaluates the application of word embeddings in concept-focused textual domains. |
Mapping semantic networks to Dutch word embeddings as a diagnostic tool for cognitive decline (2025.emnlp-main)
Copied to clipboard
| Challenge: | Semantic networks are abstract representations of the semantic memory system and can be used to estimate networks . |
| Approach: | They used Dutch verbal fluency data to explore the relationship between semantic networks and cognitive health. |
| Outcome: | The proposed measures predict cognitive health scores on the Mini-Mental State Examination (MMSE) while the traditional number-of-words measure was not significant, the results suggest that semantic network metrics may provide a more sensitive measure of cognitive health than traditional scoring. |
Challenging distributional models with a conceptual network of philosophical terms (2021.naacl-main)
Copied to clipboard
| Challenge: | Existing methods for analyzing philosophical data are not accurate enough to support philosophers . comparative research on concepts should follow a conceptual model approach, authors argue . |
| Approach: | They propose a ground truth for evaluation created by philosophy experts and a blueprint for using DS models in a sound methodological setup. |
| Outcome: | The proposed model does not perform well enough to directly support philosophers yet, but it yields promising directions for future work. |
SimLex-999 for Dutch (2024.lrec-main)
Copied to clipboard
| Challenge: | Word embeddings have revolutionised natural language processing by effectively representing words as dense vectors. |
| Approach: | They developed a Dutch variant of the SimLex-999 word similarity dataset by gathering similarity judgements from 235 native Dutch speakers. |
| Outcome: | The proposed model outperforms Bertje and RobBERT in terms of human similarity ratings and better represents semantic similarities between words. |
Using Collostructional Analysis to evaluate BERT’s representation of linguistic constructions (2023.findings-acl)
Copied to clipboard
| Challenge: | Collostructional analysis is a technique devised to find correlations between words and linguistic constructions in order to analyse meaning associations of these constructions. |
| Approach: | They propose to compare English BERT’s meaning representations to known constructions from the linguistics literature by predicting words that can be used in open slots of constructions and finding similar sequences using BERT's output embeddings. |
| Outcome: | The proposed methods show that English BERT represents constructional meaning to a certain extent, but does not separate instances of a construction from a near-synonymous construction that has a different form. |