Öz
Large Language Models (LLMs) are powerful deep learning models capable of understanding, interpreting, and generating natural language with high accuracy. These models significantly facilitate the processing of large-scale and complex data, providing substantial ease and efficiency in various natural language processing (NLP) tasks, including text classification, sentiment analysis, contextual understanding, and automatic content generation. This research was conducted to evaluate the sentiment analysis and contextual understanding capabilities of LLMs in the NLP domain. The study contributes to the literature by providing an integrated evaluation that assesses LLMs not only for their generative and classification capabilities but also for their ability to maintain and predict semantic integrity. A balanced dataset consisting of five emotion categories was used to test classification and fill-in-the-blank tasks with ChatGPT 5.3, Claude Sonnet 4.6, and Gemini 3.1 Pro models using the zero-shot method. In the classification task, model performance was evaluated using accuracy, precision, recall, and F1-score metrics. The results revealed that the Claude Sonnet 4.6 model demonstrated superior performance by achieving a 99.52% accuracy score. In the fill-in-the-blank task, the semantic similarity between the predicted words, the original words, and the completed sentences were measured using SBERT and cosine similarity. In this task, Gemini 3.1 Pro achieved the highest similarity performance with scores of 0.85 for word similarity and 0.94 for sentence similarity. The findings indicate that the examined LLMs generally exhibited high success in emotion classification and contextual word prediction tasks. Particularly, Sonnet 4.6 performed stronger in classification, while Gemini 3.1 Pro showed greater strength in semantic fill-in-the-blank tasks. These results highlight the potential of LLMs in understanding and completing emotion-bearing texts in everyday language, thereby underscoring their importance in NLP research.
Anahtar Kelimeler: Large Language Models, Natural Language Processing, Classification, Semantic Similarity Analysis, Sentiment Classification
Öz
In this study, a higher-dimensional flat Friedmann–Robertson–Walker (FRW) universe model filled with pressureless matter and Rényi holographic dark energy is investigated within the framework of self-creation cosmology. By adopting a specific parametrization of the Hubble parameter, exact analytical solutions of the field equations are obtained, and the cosmological dynamics of the model are analyzed. The results show that the model consistently describes an expanding universe with a dynamically evolving dark energy component. The equation of state parameter ω evolves toward negative values at late cosmic times and remains within the quintessence region, providing a consistent description of late-time accelerated expansion without entering the phantom domain. Furthermore, the self-creation scalar field Φ is found to be constant, implying that the field equations effectively reduce to a form equivalent to General Relativity with a modified coupling term. Thus, the model does not yield an independent dynamical self-creation solution but instead corresponds to the General Relativity limit of the theory. Despite this reduction, the inclusion of Rényi holographic dark energy leads to a physically viable cosmological scenario.
Anahtar Kelimeler: Self Creation Cosmology, FRW Universe, Higher Dimensional Cosmology, Rényi Holographic Dark Energy, Hubble Parameter Parametrization
Öz
Salinity is a major constraint on plant growth, although moderate salinity may enhance antioxidant-related metabolites in microgreens. This study evaluated species-specific physiological and biochemical responses of four Brassicaceae microgreens, radish (Raphanus sativus L.), broccoli (Brassica oleracea var. italica), red cabbage (Brassica oleracea var. capitata f. rubra), and rocket (Eruca sativa Mill.), exposed to 0, 50, 100, and 150 mM NaCl. Increasing salinity progressively inhibited growth, with the greatest reductions at 150 mM NaCl. Shoot length decreased by 35.6% in radish and 35.9% in broccoli, while stronger reductions occurred in red cabbage and rocket, reaching 43.3% and 43.8%, respectively. Salinity also disrupted ion homeostasis by increasing Na⁺ accumulation and the Na⁺/K⁺ ratio, particularly in rocket, where this ratio reached 1.33 at 150 mM NaCl. Severe salinity reduced total chlorophyll and Fv/Fm, indicating impaired photosynthetic performance. Despite this decline, radish and broccoli maintained higher pigment content and PSII efficiency. Moderate salinity, especially 100 mM NaCl, stimulated antioxidant-related metabolism, with radish showing the highest phenolic content, flavonoid content, and antioxidant activity. These findings suggest that moderate salinity may improve functional biochemical quality, whereas severe salinity limits growth.
Anahtar Kelimeler: Microgreens, Brassicaceae, Ion Homeostasis, Chlorophyll Fluorescence, Phenolic Compounds, Flavonoids, Antioxidant Activity, Salinity Tolerance
Öz
Rice fields are complex agroecosystems in which soil bacterial communities play essential roles in nutrient cycling, organic matter decomposition, soil fertility, and plant health. Herbicides are widely used in rice production to control weeds and enhance crop productivity; however, their effects extend beyond target plants and influence non-target soil microorganisms. This narrative review synthesises current knowledge on how herbicides influence bacterial populations in rice ecosystems. Evidence indicates that herbicide application often causes short-term reductions in bacterial abundance, microbial biomass, and soil enzymatic activities, particularly during the first few weeks after application. Nevertheless, many bacterial communities exhibit partial or complete recovery as herbicide residues degrade over time. Herbicides can alter bacterial community composition by suppressing sensitive functional groups, including nitrifying and nitrogen-fixing bacteria, while promoting herbicide-degrading taxa such as certain Bacillus and Actinobacteria species. These shifts may affect critical ecosystem processes, especially nitrogen cycling, nutrient availability, and soil multifunctionality. Long-term or excessive herbicide exposure can reduce microbial diversity, simplify ecological networks, and weaken soil resilience, potentially compromising sustainable rice production. The review also highlights how rice establishment systems, herbicide dosage, and environmental conditions influence bacterial responses. Emerging alternatives, including bioorganic fertilizers, microbial inoculants, integrated weed management, and microbiome-based remediation approaches, show promise for minimizing adverse effects on soil bacteria while maintaining effective weed control.
Anahtar Kelimeler: Herbicides, Rice Fields, Soil Bacteria, Microbial Diversity, Sustainable Weed Management
English