108495 research outputs found
Sort by
More than power, luxury and sieges : an analysis of interpretative strategies for representing female perspectives in three European castles and palaces
Get PDFThis study is set to analyse how female voices are incorporated into the narratives of European castles and palaces, and how different interpretative tools are used to reflect female agency. Through three case studies, examining the permanent exhibitions of Leeds Castle in England, Stirling Castle in Scotland and Wilanów Palace in Poland, the research aims to analyse the impact of various interpretative media, the narrative choices made and the influence of space on narration. The research seeks to enlighten current practices, recurring concerns, areas for improvement, and seemingly effective strategies with the goal of encouraging heritage sites to broaden their focus and incorporate women’s voices into their narratives. It also presents methods that may help diversify historical interpretations while enhancing inclusivity. Through display analysis, this study highlights the potential of enhanced female representation to provide more diverse perspectives on the past and to challenge prior gender structures.https://www.ester.ee/record=b6019671*es
Question-parsing with Abstract Meaning Representation enhanced by adding small datasets
No full textAbstract Meaning Representation (AMR) is a graph-based formalism for representing meaning in sentences. As the annotation is quite complex, few annotated corpora exist. The most well-known and widely-used corpora are LDC’s AMR 3.0 and the datasets available on the new AMR website. Models trained on the LDC corpora work fine on texts with similar genre and style: sentences extracted from news articles, Wikipedia articles. However, other types of texts, in particular questions, are less well processed by models trained on this data. We analyse how adding few sentence-type specific annotations can steer the model to improve parsing in the case of questions in English
The effect of owning a smart device on free play of 4–6 year olds
Get PDFKäesoleva uurimistöö eesmärk oli uurida hüpoteesi, et nutiseadmete liigne kasutamine mõjutab loovust negatiivselt (Aru & Rozgonjuk, 2022). Tööga sooviti uurida nutiseadmete ja loovuse seost lastel vabamängu kontekstis. Töö eesmärgiks oli teada saada, kuidas nutiseadme omamine mõjutab 4–6-aastaste laste motivatsiooni osaleda vabamängus. Selle uurimiseks viidi läbi küsitlus lapsevanemate seas. Lastele loodi ülesanne, kus tuli järjestada esitatud kaardid meeldivuse järjekorras, kaartide peal olid erinevad tegevused, muuhulgas mitmed mängimisega seotud tegevused ja nutiseadmes olemine. Valim koosnes 18-st 4–6-aastasest lapsest ja nende vanematest, kellest 8 omas isiklikku nutiseadet ning 10 ei omanud isiklikku nutiseadet. Hoolimata väikesest valimist leiti, et isiklikku nutiseadet omavad lapsed osalevad vähem vabamängus. Lisaks selgus, et nutiseadmeid eelistavad lapsed eelistavad vähem vabamängu, raamatute lugemist ja õppimist. Antud tulemused viitavad, et antud teema vajab edasist ja täiendavat uurimist ning nutiseadme isiklik omamine varases eas võib mõjutada laste vabamängu
LAG-MMLU: Benchmarking Frontier LLM Understanding in Latvian and Giriama
Get PDFThis paper evaluates the language understanding capabilities of various large language models (LLMs) through an analysis of 112 translated and human-edited questions from the Multitask Language Understanding (MMLU) dataset, focusing specifically on two underrepresented languages: Latvian and Giriama. The study compares the performance of six state-of-the-art (SOTA) models, with OpenAI's o1-preview model demonstrating superior performance across all languages, significantly outperforming non-proprietary models in Latvian and all other models in Giriama. Human editing of automated translations from English to Latvian yielded only a small, statistically insignificant improvement in performance estimates, suggesting that machine-translated benchmarks may be sufficient for comparing model performance in languages with established digital resources like Latvian. However, automated translation to Giriama proved infeasible, and model performance in Giriama remained poor, highlighting the persistent challenges LLMs face with low-resource languages. These findings underscore the need for more comprehensive datasets and improved machine translation capabilities for underrepresented languages, while emphasizing the importance of localized benchmarks and human evaluation in addressing cultural and contextual limitations in AI models
The impact of reading activities and children's book preferences at Sillamäe kindergarten Päikseke on children's interest in reading and their Estonian language development
No full texthttps://www.ester.ee/record=b573855
Inimkeskse koolitusmeetodi väljatöötamine gümnaasiumiõpilaste harimiseks sotsiaalse manipuleerimise tehnikate kohta
Get PDFData security plays a vital role in our society. We use different tools for communication, such as social networks, emails, and phone messages. In the security of personal data, an important role play is technology, which collects and secures user data and the human who owns it. If technology helps to secure data, then the human role in the system is to hold access to their data and not give it to another person. From different papers, it could be found that the “user is the weakest link in the security chain”. This happens because of various psychological manipulations that attackers use to receive sensitive data from users or, in other words, Social Engineering. To prevent such situations, people must be taught how attackers could receive their data through such manipulations and how to not fall into an attacker's trap by creating human-centric cybersecurity training. The current solutions lack a human-centered approach and platform tailored to high school students. Therefore, this research provides information about weak social engineering spots among high school students. Using knowledge about high school students' weak social engineering skills, this research presents a game-based training program using a one-platform solution to train high school students against current social engineering techniques with which they have problems. The efficiency of the training and platform is evaluated by the results of the first and second questionnaires to provide results of changes in the social engineering knowledge and skills of high school students.Andmeturbel on mängib meie ühiskonnas väga oluline rolli. Me kasutame suhtlemiseks erinevaid vahendeid, nagu sotsiaalvõrgustikud, e-kirjad ja telefonisõnumid. Isikuandmete turvalisuses mängib olulist rolli tehnoloogia, mis kogub ja kaitseb kasutajate andmeid, ning inimene, kellele need andmed kuuluvad. Kui tehnoloogia aitab kaitsta andmeid, siis inimese ülesanne süsteemis on hoida juurdepääs oma andmetele ega anda seda teisele isikule. Erinevatest artiklitest võib leida, et "kasutaja on turvaahela nõrgim lüli". See juhtub erinevate psühholoogiliste manipulatsioonide tõttu, mida ründajad kasutavad tundliku teabe saamiseks kasutajatelt ehk sotsiaalse manipuleerimise kaudu. Selliste olukordade ennetamiseks tuleb inimesi õpetada, kuidas ründajad võivad nende andmeid saada selliste manipulatsioonide kaudu ja kuidas mitte sattuda ründaja lõksu, luues inimesekeskset küberturvalisuse koolitust. Praegused lahendused ei keskendu piisavalt inimeste vajadustele ega paku keskkonda, mis oleks kohandatud gümnaasiumiõpilastele. Seetõttu see uuring annab teavet gümnaasiumiõpilaste nõrkade sotsiaalse manipuleerimise kohtade kohta. Kasutades teadmisi gümnaasiumiõpilaste nõrkadest sotsiaalse manipuleerimise oskustest, see uuring esitab mängupõhise koolitusprogrammi, mis kasutab ühtset platvormilahendust gümnaasiumiõpilaste koolitamiseks tänapäevaste sotsiaalse manipuleerimise tehnikate vastu, millega gümnaasiumiõpilastel on probleeme. Koolituse ja platvormi tõhusust hinnatakse esimese ja teise küsimustiku tulemuste abil, et näidata gümnaasiumiõpilaste sotsiaalse manipuleerimise teadmiste ja oskuste muutusi
Lugusid Prinarovje piirkonnast kui osa narvalaste ajaloolisest mälust: kogemused Narva koolide kakskeelsete õpilaste kultuurilist identiteeti toetavate õppematerjalide loomisel
Get PDFhttps://www.ester.ee/record=b574022
Efficient Scientific Full Text Classification: The Case of EICAT Impact Assessments
No full textThis study explores strategies for efficiently classifying scientific full texts using both small, BERT-based models and local large language models like Llama-3.1 8B. We focus on developing methods for selecting subsets of input sentences to reduce input size while simultaneously enhancing classification performance. To this end, we compile a novel dataset consisting of full-text scientific papers from the field of invasion biology, specifically addressing the impacts of invasive species. These papers are aligned with publicly available impact assessments created by researchers for the International Union for Conservation of Nature (IUCN). Through extensive experimentation, we demonstrate that various sources like human evidence annotations, LLM-generated annotations or explainability scores can be used to train sentence selection models that improve the performance of both encoder- and decoder-based language models while optimizing efficiency through the reduction in input length, leading to improved results even if compared to models like ModernBERT that are able to handle the complete text as input. Additionally, we find that repeated sampling of shorter inputs proves to be a very effective strategy that, at a slightly increased cost, can further improve classification performance
Interactive effects of leaf pathogens and plant mycorrhizal type on plant diversity–productivity relationships
Get PDFDiversity–productivity relationships can differ between forests dominated by different mycorrhizal types and be modulated by specialist and generalist pathogens. However, little is known about how these factors interact to modulate biodiversity effects. We addressed this knowledge gap with a 2-year experiment combining the manipulation of plant richness (one, two, four, eight species) and mycorrhizal tree type (arbuscular mycorrhizal [AM] tree-dominated; ecto-mycorrhizal [ECM] tree-dominated) with fungicide application for leaf pathogens (added or control). Biodiversity effects were quantified for community productivity and its two components (shoots and roots). We observed nonlinear diversity–productivity relationships, with the productivity of ECM tree-dominated communities increasing at low to intermediate diversity and declining at the highest species richness. Foliar fungicide application reduced positive complementarity effects and increased productivity in both ECM tree monocultures as well as eight-species mixtures. This finding suggests that the dilution effects of specialized pathogens may dominate at low diversity, while the spillover effects of generalist pathogens may become dominant at high diversity, resulting in unimodal diversity–productivity relationships. In AM tree-dominated communities, aboveground productivity strongly increased in response to leaf pathogen suppression in eight-species mixtures, and the release from leaf pathogens benefited most of the species that were most productive in fungicide-treated monocultures. This agrees with the prediction that spillover effects of generalist pathogens in diverse plant communities could differentially suppress highly productive species due to the trade-off between growth and defense. In addition, positive biodiversity effects on root production were significantly stronger in AM tree- than ECM tree-dominated communities. Our results demonstrate that relationships between plant diversity and productivity can be nonlinear due to the combined effects of specialized and generalized plant–fungal interactions, depend on plant mycorrhizal type, and differ between aboveground and belowground compartments
Interpretable Machine Learning for Societal Language Identification: Modeling English and German Influences on Portuguese Heritage Language
Get PDFThis study leverages interpretable machine learning to investigate how different societal languages (SLs) influence the written production of Portuguese heritage language (HL) learners. Using a corpus of learner texts from adolescents in Germany and the UK, we systematically control for topic and proficiency level to isolate the cross-linguistic effects that each SL may exert on the HL. We automatically extract a wide range of linguistic complexity measures, including lexical, morphological, syntactic, discursive, and grammatical measures, and apply clustering-based undersampling to ensure balanced and representative data. Utilizing an explainable boosting machine, a class of inherently interpretable machine learning models, our approach identifies predictive patterns that discriminate between English- and German-influenced HL texts. The findings highlight distinct lexical and morphosyntactic patterns associated with each SL, with some patterns in the HL mirroring the structures of the SL. These results support the role of the SL in characterizing HL output. Beyond offering empirical evidence of cross-linguistic influence, this work demonstrates how interpretable machine learning can serve as an empirical test bed for language acquisition research