UIU Digital Institutional Repository System

About UIU Dspace:

DSpace is a digital service that collects, preserves, and distributes digital material. Repositories are important tools for preserving an organization's legacy; they facilitate digital preservation and scholarly communication.

Photo by @inspiredimages

Communities in DSpace

Select a community to browse its collections.

Now showing 1 - 5 of 10

Recent Submissions

News Recommendation using Advanced LLM Embedding Models
(UIU, 2026-08-12)
Ahmed, Fahim
Online news recommendation is a subset of recommendation systems used by online news portals to filter news articles according to user interest. News recommendation methods generally have three main components: news encoder, user encoder, and click predictor. After the success and popularity of BERT, many methods were developed that incorporate fine-tuned BERT for feature extraction. However, there have been further advancements in the field of LLM based embedding models, generalized embedding models, and decoder-based embedding models, that have not been studied yet for news recommendation systems. Most news recommendation studies consider only ranking a candidate set of news items for each user. In this study we compared this with other news recommendation tasks as well: user-news classification; all user classification for given news; candidate news classification for user; and candidate news ranking for user. We also studied multilingual news recommendation and the inclusion of a fake news classification component. In this study we investigated the effect of using five BERT-based models, five large decoder-based models, and one proprietary model as news encoders for news recommendation. We also studied the effect of different model inputs (title, category, and abstract) on news and user representation, as well as the effect of four user encoders:average pooling, attention pooling, dense layer + attention pooling, multi-head self-attention + attention pooling. Inclusion of the fake news classification component improved results for news recommendation tasks 1, 2 and 3, but reduced performance for task 4. For the first task, multilingual-e5-large-instruct with title, category, subcategory, and abstract input and dense layer + attention pooling for user encoder achieved the highest accuracy of 0.9149; for the second task, text-embedding-3-large with title input and multi-head self-attention + attention pooling for user encoder achieved the highest accuracy of 0.8902; for the third task, multilingual-e5-large-instruct with title as input and multi-head self-attention + attention pooling for user encoder achieved the highest accuracy of 0.8641; and for the fourth task, mxbai-embed-large-v1 with title, category and abstract as input and multi-head self-attention + attention pooling for user encoder achieved the highest AUC of 0.6648. Our findings indicate that frozen generalized embeddings can provide competitive news representations without embedding-model fine-tuning, although performance depends strongly on the recommendation task and user encoder.
A Novel Evaluation Strategy for Domain-Specific Legal QA: Application to Bangladeshi Law with RAG
(UIU, 2026-08-08)
Retrieval-Augmented Generation, or RAG, has been identified as a promising approach for question answering tasks, particularly for high-risk fields such as law, for which accuracy, interpretability, and contextuality are essential. Current evaluation approaches for RAG models are general natural language evaluation scores, which do not embody specific legal reasoning attributes such as temporal validity, jurisdictional validity, procedural adequacy, or reconciliation of numerical constraints. This paper introduces a legal-aware evaluation framework for RAG models, and we build an AI legal assistant for Bangladeshi laws. We introduce a novel evaluation framework with seven complementary metrics: (1) Gold Answer baseline correctness, (2) BERTScore semantic similarity, (3) Numeric Accuracy, (4) Temporal Validity, (5) LLM Judge reasoning quality, (6) Legal Validity statute checking, and (7) Robustness. Evaluation results show our legal-aware framework is successful and measures key unexplored failing dimensions for quality evaluation, giving a CLR score of 0.824 on 20 test queries under the Baseline configuration, with perfect scores on Temporal Validity (1.00) and Legal Validity (1.00), and strong BERTScore semantic similarity (0.861), and improved Numerical Accuracy following word-form normalization of Bangladeshi legal quantity expressions, confirming both system reliability and the framework’s ability to surface domain-specific failure modes invisible to standard NLP evaluation.
Performance Analysis of Alliance Finance PLC
(United International University, 2026-08)
Ahmed, Tahmid
This report presents a comprehensive performance analysis of Alliance Finance PLC (formerly Lankan Alliance Finance Limited) over the five-year period from 2021 to 2025. The analysis employs the Aggregate Return on Equity (ROE) framework to decompose the company's shareholder returns into their underlying operational and financial drivers, providing insights into profitability, efficiency, liquidity, and solvency. The study is based entirely on secondary data sourced from the audited annual reports of Alliance Finance PLC for the fiscal years 2021 through 2025. The methodology includes a detailed ratio analysis across five dimensions: market ratios, profitability ratios, efficiency ratios, liquidity ratios, and solvency ratios. This is followed by a systematic decomposition of ROE using both the traditional two-component framework (ROA × Equity Multiplier) and the expanded five-component DuPont framework (Tax Burden × Interest Burden × Operating Margin × Asset Turnover × Equity Multiplier). The analysis reveals that Alliance Finance PLC's ROE has been highly volatile over the five-year period, ranging from 1.95% in 2023 to a peak of 6.26% in 2024. The primary drivers of ROE were operating margin and asset turnover, while the equity multiplier showed a consistent upward trend from 2.72 in 2021 to 4.68 in 2025, indicating increasing reliance on financial leverage. The company's growth rate consistently exceeded its sustainable growth rate across all years, suggesting that growth has been largely financed through external debt rather than internally generated funds. Key concerns include a declining interest coverage ratio from 1.83 to 0.83 and rising financial leverage, which together indicate increasing financial risk. The findings suggest that the path to improved and sustainable shareholder returns lies in optimizing funding costs, diversifying revenue sources beyond interest income, maintaining credit quality to minimize provisions, and balancing financial leverage with operational efficiency.
Project Report On Liquidity Management in Foreign Commercial Banks of Bangladesh
(2026-04-05)
Rima, Noor
This project report mainly focuses on the methods of liquidity management used by foreign commercial banks in Bangladesh. Every foreign commercial bank in Bangladesh that operates efficiently abides by the liquidity ratio regulations. At first, it introduces theories and principles for managing liquidity in foreign commercial banks of Bangladesh. Additionally, it contains the key prerequisites for the mechanism of the liquidity management process. When granting loans, foreign commercial banks need to exercise extreme caution. This report’s main goal is to determine whether foreign commercial banks are adhering to the exact liquidity standards. Using financial ratios and liquidity indicators, secondary sources of data are gathered and assessed. The parameters are used to analyze the liquidity condition. According to the analysis, the banking industry has been maintaining more liquidity since 2014 than is required by law. As for that reason, banks are not under any liquidity pressure. Every bank should have a current contingency financing technique to fulfill the economic necessity in a liquidity crisis condition. In this report, an effort has been taken to assess the proficiency in liquidity management of foreign commercial banks of Bangladesh.
INTERNSHIP REPORT ON INTERNAL CONTROL SYSTEM OF UNITED HEALTHCARE SERVICES LIMITED
(2026-04-05)
Rumi, Muslima
United Healthcare Services Limited is one of the leading healthcare service providers in Bangladesh and a sister concern of United Group. This internship report focuses on the Internal Control System of United Healthcare Services Limited. The report is based on practical experience gained during the internship period in the finance and accounts department. The main objective of this report is to analyze the effectiveness of the internal control system practiced within the organization, particularly in financial transactions, cash collection, vendor payments, and accounting procedures. The study highlights how internal controls help ensure accuracy, prevent fraud, maintain proper documentation, and support decision-making. During the internship, various activities were observed and performed, including recording transactions, handling different types of collections (cash, card, bKash, and online), preparing accounting entries, and assisting in loan calculations. These practical tasks helped in understanding how internal control mechanisms are applied in daily operations. The report also identifies some challenges faced by the organization, such as financial constraints, high operational costs, and reliance on external financing. Despite these challenges, UHSL maintains structured policies, supervision, and approval processes to strengthen internal control. Finally, the report concludes that an effective internal control system is essential for ensuring transparency, efficiency, and long-term sustainability. Some recommendations are provided to further improve the control system, including better automation, monitoring, and staff training.