Institution profile

Charles University

Academic institutioneurope · cz
Official website
Research library332linked papers
Opportunities0open roles
Selected work

Representative Papers

LLM-ABBA: Understanding time series via symbolic approximation

Nov 27, 2024arXiv.org

This work addresses the challenge of effectively leveraging semantic information in time series for large language models (LLMs). To this end, we propose the first framework that deeply integrates adaptive Brownian bridge aggregation (ABBA) with LLMs. Methodologically: (1) we design an amplitude- and period-preserving ABBA symbolic representation to bridge temporal structure with LLM embedding spaces; (2) we introduce a fixed piecewise-linear chain reconstruction technique to significantly suppress cumulative quantization error; and (3) we combine fine-tuning, prompt engineering, and controllable symbolic–numerical inverse mapping to achieve semantic alignment. Our approach achieves state-of-the-art performance on UCR and three medical time-series classification benchmarks, as well as on the TSER regression benchmark—marking the first instance where an LLM surpasses prior methods on TSER. Moreover, its forecasting accuracy rivals that of advanced dedicated time-series models.

2 citations1 influentialRead paper

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

Approximation of Spanning Tree Congestion using Hereditary Bisection

Oct 01, 2024

This paper studies the NP-hard Steiner Tree Congestion (STC) optimization problem: constructing a spanning tree that minimizes the maximum number of times any original graph path traverses a single tree edge. For sparse graphs with $m = O(n log n)$ edges and bounded maximum degree $Delta$, we establish the first tight lower bound relating STC to the hereditary bisection width $hb(G)$: $mathrm{STC}(G) geq Omega(hb(G)/Delta)$. Leveraging this bound, we design an $O(Delta cdot log^{3/2} n)$-approximation algorithm. Our method integrates hereditary bisection width analysis, recursive graph partitioning, and degree-constrained lower-bound derivation. For graphs with $Delta leq mathrm{polylog}(n)$, the algorithm achieves an $O(log^{3/2} n)$ approximation ratio—breaking the classical $O(n)$ barrier—and constitutes the first STC approximation algorithm whose performance guarantee is sublinear in $Delta$. This yields significantly improved theoretical guarantees and constructive techniques for low-degree sparse graphs.

1 citations1 influentialRead paper

Frontier Space-Time Algorithms Using Only Full Memory

Feb 24, 2026

This work investigates the efficient solution of classic computational problems—including directed s-t connectivity, edit distance, longest common subsequence, and discrete Fréchet distance—under the stringent constraint of using only O(log n) working space. By leveraging the catalytic computation model, the study designs polynomial-time algorithms that utilize a read-only catalytic space of size n / 2^{Θ(√log n)} to achieve, within O(log n) working space, time-space trade-offs approaching those of optimal non-catalytic algorithms. The key contribution lies in the first matching of state-of-the-art time-space bounds of non-catalytic algorithms within the catalytic framework, while simultaneously reducing the number of random bits required by randomized algorithms to O(log n), thereby attaining near-optimal runtime efficiency comparable to settings without space restrictions.

1 citationsRead paper

Strategies for Span Labeling with Large Language Models

Jan 23, 2026

Large language models lack explicit mechanisms to refer to specific spans in the input text, leading to inconsistent performance with existing span annotation prompting strategies. This work systematically examines three categories of approaches: input tagging, numerical indexing, and content matching, and proposes LogitMatch—a novel constrained decoding method that enforces alignment between model outputs and valid input spans in logit space to address the inconsistency inherent in content matching. Experiments across four diverse tasks demonstrate that LogitMatch significantly outperforms existing content matching methods and, in certain settings, surpasses other strategies, while also confirming that input tagging remains a robust baseline.

1 citationsRead paper
Recent publications

Latest Papers

Free and projective LCD codes

Sep 14, 2026

研究解决了非可分解环上LCD码的自由性和射影性问题,通过推广Frobenius环特征到非交换环境,并证明群码是LCD当且仅当由中心幂等元生成。

0 citationsRead paper