Institution profile

Poznan University of Technology

Academic institutioneurope · pl
Official website
Research library109linked papers
Opportunities0open roles
Selected work

Representative Papers

Polish-ASTE: Aspect-Sentiment Triplet Extraction Datasets for Polish

Feb 27, 2025International Conference on Language Resources and Evaluation

Aspect-Sentiment-Opinion Triplet Extraction (ASTE) lacks annotated resources for Slavic languages, particularly Polish, which has no publicly available dataset. Method: We introduce the first Polish ASTE dataset, covering two domains—hotels and e-commerce—and strictly adhering to the standard English ASTE format to ensure cross-lingual comparability. The dataset is manually annotated with fine-grained sentiment structures and released under a CC-BY-NC license. Contribution/Results: Using this resource, we conduct the first systematic evaluation of two mainstream ASTE paradigms and two Polish large language models, revealing critical performance bottlenecks of existing methods on Slavic languages. This work fills a key gap in low-resource, fine-grained sentiment analysis for Slavic languages and establishes a benchmark dataset and empirical foundation for future multilingual ASTE research and model development.

2 citationsRead paper

Keyframe-based Dense Mapping with the Graph of View-Dependent Local Maps

May 01, 2020IEEE International Conference on Robotics and Automation

This work addresses the challenge of efficiently constructing accurate and globally consistent dense 3D maps by proposing a keyframe-based RGB-D dense mapping system. The method introduces a viewpoint-dependent 2D structure to store Normal Distributions Transform (NDT) cells, better aligning with the observation characteristics and uncertainty inherent in RGB-D sensors. Local NDT maps are maintained through keyframes and integrated with pose graph optimization and loop closure detection to achieve global consistency. The system further supports fusion and filtering of local maps to produce a complete environmental model. Experimental results demonstrate that the proposed approach outperforms Octomap and NDT-OM in both mapping accuracy and completeness, making it well-suited for high-quality dense 3D reconstruction in real-world scenarios.

2 citationsRead paper

An interpretable prototype parts-based neural network for medical tabular data

Mar 05, 2026

This work addresses the challenge of limited clinical trust in machine learning models for medical tabular data due to poor interpretability. It introduces an intrinsically interpretable neural network by adapting the concept of prototype parts—originally developed in computer vision—to the domain of medical tabular data. The model discretizes inputs through trainable feature patches, learns prototype parts grounded in clinical semantics, and performs case-based comparisons between patient features and these prototypes in a latent space, yielding transparent predictions articulated in clinically meaningful language. Evaluated across multiple medical benchmark datasets, the approach achieves classification performance on par with state-of-the-art models while generating human-readable explanations that align with clinical reasoning.

1 citationsRead paper
Recent publications

Latest Papers