FACULTY OF PHYSICAL SCIENCE

This paper presents a comprehensive comparative search analysis of three prominent generative AI models: ChatGPT, Gemini, and Perplexity. By focusing on their architecture, performance, training processes, and real-world applications, we provide a detailed case study that highlights the strengths and limitations of each model in various natural language processing (NLP) tasks. The analysis covers aspects such as model scalability, accuracy, response time, and adaptability across different domains. Through systematic benchmarking and evaluation of these models on both factual and creative prompts, we explore their potential to revolutionize industries such as education, customer service, and content generation. This study also aims to inform future developments in generative AI by identifying gaps and opportunities for improvement in model design and training methodologies.

This research gives attention to the non-energy use of Lignite samples collected from Ohordua town of Edo State and Azagba-Ogwashi in Delta State. A total of six composite Lignite samples,from ten sampled points were subjected to analytical study, with a view to determining their humic content and properties for use in soil beneficiation and conditioning.Proximate and Ultimate analytical technique was used to characterize the lignite from which humic acid was extracted under alkaline condition, and precipitated using acidic medium of 6M HCl.X-Ray Diffraction and X-Ray Fluorescence analytical methods were employed to determine the mineralogical and elemental composition of the samples respectively,while Ultra-Violent visible spectroscopy was used for organic characterization of the humic acid.Proximate analyses of the Lignite show percentage composition of Ohordua and Azagba-Ogwashi Lignite respectively,Fixed Carbon (65.22 wt% and 67.16wt%),Moisture Content (26 wt% and 25.4wt%) and Ash (5.02 wt% and 4.5 wt%).Ultimate analyses shows Hydrogen (6.60 wt% and 6.22wt%), Nitrogen (1.34 wt% and 1.19wt%,) Sulphur (0.77 wt% and 0.70wt%), Oxygen (26.07 wt% and 24.73 wt%), for Ohordua and Azagba-Ogwashi respectively.The analytical results suggest the existing lignite with varying thickness is of Lignite rank. The Ultimate analyses of the extracted Humic acid show Carbon (57.5wt% and 55.8wt%)%, Hydrogen (5.1 wt% and 3.9wt%), Nitrogen (2.5 wt% and 1.1wt%), Sulphur (Nil and 0.3wt%) and Oxygen (34.9 wt% and 38.9wt%,) for Ohordua and Azagba-Ogwashi, respectively.The mineralogical composition in the lignite samples shows Kaolinite was the main clay mineral identified. Non clay minerals such as anorthite, drierite, yeelimite, spurite, belite, wollastonite, hematite, pyrite, bauxite and quartz were detected.For Ultra-Violent visible spectroscopy characterization of the Humic Acid, E2/E3 and E4/E6 absorption values for Ohordua (21.69 and 4.50) respectively, while Azagba-Ogwashi reveal (19.39 and 4.6) respectively. The humic substance has percentage aromaticity values of 33.90% and 34.06%, total acidity has 10.4% and 10.1%, COOH has 3.4% and 3.3%, Phenolic has 7.0% and 6.8% and acidity ratio has 0.49% and 0.48%, for Ohordua and Azagba-Ogwashi, respectively which depicts the Humic Acid has a high degree of Humification and a very good total acidity ratio. Results from this research indicate that the Lignite from both locations are similar in geochemistry, mineralogy and organic properties. Although,humification is high in both samples, Ohordua Lignite with lesser aromaticity and E4/E6 ratio, higher E2/E3 ratio and higher acidity ratio is more suitable for soil beneficiation and conditioning than that of Azagba-Ogwashi.

This project focuses on developing an intelligent system for heart disease prediction usingAdaptiveNeuro-Fuzzy Inference Systems (ANFIS) and Genetic Algorithm (GA). Cardiovascular diseaseisaleading cause of mortality worldwide, necessitating improved diagnostic tools. The proposedsystem leverages the strengths of neural networks and fuzzy logic to handle the complexitiesanduncertainties inherent in medical data. The ANFIS model is optimized using GAto enhancetheaccuracy of predictions. The system was tested on the UCI Cleveland heart disease dataset, achieving a satisfactory accuracy rate of 91.25%. This indicates the potential of the systemtoassistclinicians in making reliable diagnoses based on a combination of key medical attributes. Futurework may involve extending the system to diagnose other complex diseases and incorporatingadditional heuristic algorithms to further refine model performance

Corrosion is the deterioration of metal caused by chemical reactions or interactions with its
surroundings. In order to address the issue of corrosion which is a major challenge in the
metallurgical industry, there is need to introduce an inhibitor to combat this issue of corrosion. Before now, chemical inhibitors has been in use and because of the risk, toxicity and
disadvantages of these chemical inhibitors, there is a current trend that is moving away from the
use of chemical inhibitors to the use of green organic inhibitor that is sustainable and
environmentally friendly. In this research work, the inhibition of corrosion on mild steel in 1M
HCl solution was evaluated by utilizing Ficus sur extract which serves as the corrosion inhibitor. The synthesis of the ficus sur was done using the hot maceration method and the study of the
corrosion rate was done using the gravimetric weight loss method and the concentrations of
Ficus sur extracts used ranges from 0.0g (blank), 0.1g, 0.2g, 0.3g and 0.4g and the temperature
was varied at three temperatures of 25°C (298K), 48°C (321K) and 60°C (333K). From the result obtained it was shown that as inhibitor concentration increases, corrosion rate
decreases following the gravimetric weight loss method used. Also adsorption studies were
carried out using the Langmuir, Temkin and Freundlich models. From the adsorption studies
carried out although they followed the three adsorption model, these research work best fit the
Langmuir adsorption model which gives us R² = 0.9931. The mild steel surface was also characterized using Scanning Electron Microscope (SEM) which
provides a clear morphology of the metal surface.