https://hdl.handle.net/20.500.14178/909 Thu, 21 May 2026 22:36:48 GMT 2026-05-21T22:36:48Z https://hdl.handle.net/20.500.14178/3809 Automatic Suggestions Help Extending Eventive Ontology: A Case Study on SynSemClass Straková, Jana; Fučíková, Eva; Urešová, Zdeňka; Hajič, Jan Piperidis, Stelios; Bel, Núria; Henk van den, Heuvel; Ide, Nancy; Toral, Antonio Despite substantial recent progress in many areas of NLP, semantic tasks remain particularly challenging. One such task is the creation (extension, or annotation) of semantic ontologies. In this work, we present a case study on the eventive SynSemClass ontology, focusing on the challenges of semantic annotation - that is extending the ontology with new lexical units and/or new concepts - both with and without automatic support. We consider two strategies for generating annotation suggestions: (i) a knowledge-driven approach based on a small, carefully curated corpus of verbal valency frames, and (ii) a corpus-driven approach using lemma-based suggestions from a large raw text collection, disregarding semantic homonymy. Our findings show that ontology annotation is inherently difficult, and that automatic annotations statistically significantly reduce this difficulty both in terms of inter-annotator agreement and when compared with gold expert annotations. We discuss the implications for semantic resource creation and extension, as well as the limits of automation in ontology annotation.; I přes podstatný pokrok v mnoha oblastech NLP zůstávají sémantické úkoly obzvláště náročné. Jedním z takových úkolů je vytvoření (rozšíření, nebo anotace) sémantických ontologií. V této práci předkládáme případovou studii o eventivní ontologii SynSemClass, která se zaměřuje na výzvy sémantické anotace - tedy rozšíření ontologie o nové lexikální jednotky a/nebo nové koncepty - s automatickou podporou i bez ní. Uvažujeme o dvou strategiích pro generování anotačních návrhů: (i) znalostní přístup založený na malém, pečlivě připraveném korpusu verbálních valenčních rámců a (ii) přístup řízený korpusem využívající lemma-based návrhy z velké neanotované textové kolekce, bez ohledu na sémantickou homonymii. Naše zjištění ukazují, že ontologická anotace je ze své podstaty obtížná a že automatické anotace statisticky významně snižují tuto obtížnost jak z hlediska mezianotační shody, tak ve srovnání se zlatými anotacemi. Diskutujeme o důsledcích pro tvorbu a rozšíření sémantických zdrojů, stejně jako o limitech automatizace v ontologické anotaci. Thu, 01 Jan 2026 00:00:00 GMT https://hdl.handle.net/20.500.14178/3809 2026-01-01T00:00:00Z https://hdl.handle.net/20.500.14178/3801 Tailoring in-situ TiC-reinforced laser 3D printed β Ti-Nb alloy with graphite additions for biomedical implants Gonsalves, Vinicius R. M.; Sangali, Marcio; Rodrigues, Joao F. Q.; Spinelli, Jose E.; Minárik, Peter; Veselý, Jozef; Caram, Rubens; Afonso, Conrado R. M. beta-type titanium alloys, renowned for their low elastic moduli, combined with titanium carbide (TiC) reinforcements, have emerged as promising candidates for load-bearing and wear-resistant bioimplant components. However, the fabrication of beta-type titanium matrix composites (beta-TMCs) remains challenging, particularly in achieving a refined microstructure with a controlled reinforcement distribution. This investigation introduces an innovative approach using laser-based powder bed fusion (PBF-LB/M) for the in-situ fabrication of TiC-reinforced beta-TMCs. Powder mixtures composed of beta Ti-42Nb alloy and graphite (particle size similar to 3 mu m) were processed with graphite additions of 0.5, 1.0, 1.5, and 2.0 vol%, resulting in TiC fractions of approximately 2, 4, 6, and 8 vol%, respectively, in the PBF-LB/M as-built samples. The formation of TiC nanoparticles within the beta-Ti matrix was confirmed by microstructural characterization, including TEM/STEM with EDS and ASTAR analyses. While lower graphite levels led to the formation of rod-shaped TiC precipitates ranging from tens to hundreds of nanometers, higher graphite content produced nearly spherical TiC particles, consistent with a hypereutectic solidification path where TiC forms as a primary phase. The increasing TiC fraction enhanced hardness; specifically, the Ti-42Nb+2.0C composite (similar to 239 HV) demonstrated a 13 % increase compared to the unreinforced Ti-42Nb alloy (similar to 211 HV). The elastic modulus also increased with TiC content, reaching similar to 72 GPa compared to similar to 58 GPa for the alloy. Notably, all elastic modulus values were significantly lower than that of the Ti-6Al-4V alloy (similar to 110 GPa). These findings establish an alternative additive manufacturing route for the in-situ fabrication of beta-TMCs with a refined microstructure and enhanced mechanical properties, offering new possibilities for load-bearing implants. Thu, 01 Jan 2026 00:00:00 GMT https://hdl.handle.net/20.500.14178/3801 2026-01-01T00:00:00Z https://hdl.handle.net/20.500.14178/3800 High strength and low elastic modulus α/β cp-Ti/Ti-33Nb-33Zr multilayered in-situ heterostructure produced by accumulative roll bonding Afonso, Conrado Ramos Moreira; Lourenco, Mariana Luna; Pacheco, Gabriel Yoshika Ramos; Minárik, Peter; Veselý, Jozef; Magalhaes, Danielle Cristina Camilo Developing and improving new alloys, primarily Ti-based systems, have been investigated to replace, repair, or reconstruct lost or damaged tissues while balancing relative cost, mechanical properties, corrosion resistance, and biological response. In this context, the main objective of this study was to evaluate the feasibility of producing heterostructured materials (HM) by accumulative roll bonding (ARB) based on two Ti alloys: alpha-Ti (grade 2) and multi-principal element alloy (MEA) (3-Ti Ti-33Nb-33Zr (wt.%), with a combination of high strength and low elastic modulus for implant applications. The main results indicated that producing such heterostructured sheets with up to 7 ARB cycles at 500 degrees C is feasible. A continuous increase in hardness and a reduction in the elastic modulus were simultaneously observed with the increasing ARB cycles. In both cases, the rule-of-mixtures (considering 1:1 proportion) was not followed, resulting in higher-than-expected hardness and lower-than-expected elastic modulus. Microstructural characterization revealed significant grain refinement and a remarkable increase in dislocation density, which were identified as the primary drivers for the enhanced hardness. The reduction in elastic modulus was attributed to nanograined structure and interdiffusion at the interfaces, as confirmed by TEM analysis. The process successfully produced multilayered composite sheets with a high hardness of over 300 HV and a favorably reduced elastic modulus in the range of 65-89 GPa. Based on these findings, ARB is a very promising technique for producing Ti-based HM for biomedical purposes. Thu, 01 Jan 2026 00:00:00 GMT https://hdl.handle.net/20.500.14178/3800 2026-01-01T00:00:00Z https://hdl.handle.net/20.500.14178/3746 Smashcima: Full-Page Handwritten Music Document Synthesizer Mayer, Jiří; Pecina, Pavel; Hajič, Jan de Luca, Elsa Despite massive progress made in Optical Music Recognition (OMR) with deep learning, data scarcity remains an issue, especially for manuscripts. Synthetic data has been shown to alleviate this issue, but no tool for rendering a handwritten page from structured encoding such as MusicXML exists. This paper introduces Smashcima, a framework for the creation of synthetic handwritten full-page music images. It accepts MusicXML files and produces images with full information on their glyphs, segmentation masks, keypoints, notation graph, and semantics. It is compatible with the MuNG format and so can also be used to produce synthetic training data object detection and graph models. It can synthesize images of all levels of complexity of music notation, including pianoform music. Smashcima thus greatly increases the value of dataset acquisition, as it can expand a small manually annotated dataset to the scale of arbitrary available MusicXML data, thereby alleviating manuscript data scarcity for OMR.; Navzdory obrovskému pokroku v oblasti rozpoznávání notopisu (OMR) pomocí hlubokého učení zůstává nedostatek dat problémem, zejména u rukopisů. Je známo, že syntetická data tento problém zmírňují, ale neexistuje žádný nástroj pro vykreslení ručně psané stránky ze strukturovaného kódování jako například MusicXML. Tento článek představuje software Smashcima, framework pro vytváření syntetických ručně psaných celostránkových hudebních obrázků. Přijímá soubory MusicXML a vytváří obrázky s úplnými informacemi o jejich glyfech, segmentačních maskách, klíčových bodech, notovém grafu a sémantice. Je kompatibilní s formátem MuNG, a proto jej lze také použít k vytváření syntetických trénovacích dat pro detekci objektů a grafových modelů. Umí syntetizovat obrázky všech úrovní složitosti hudební notace, včetně klavírních. Smashcima tak výrazně zvyšuje hodnotu získávání datových sad, protože dokáže rozšířit malou ručně anotovanou datovou sadu na libovolný rozsah dostupných dat MusicXML, čímž zmírňuje nedostatek ruk Wed, 01 Jan 2025 00:00:00 GMT https://hdl.handle.net/20.500.14178/3746 2025-01-01T00:00:00Z