Welcome to SDU Institutional Repository
Communities
Select a community to browse its collections.
Recent Submissions
Investigating Knowledge Graphs for Context-Aware Search in Financial Transactional AI Agents
(SDU University, 2026) Baisenkul N.
This thesis presents an empirical evaluation of a hybrid knowledge-graph and largelanguage-model (KG–LLM) architecture for financial transactional compliance reasoning. The architecture was proposed in a companion conference paper [1] that established the design and demonstrated it on a small worked example; the contribution of this thesis is the quantitative, ablation-based, robustness-aware empirical study that completes the proposal. A deterministic synthetic dataset of 500 transactions across eight compliance-rule scenarios plus a clean negative class is generated under two complementary disciplines: a crossrule labeling pass that mirrors how real anti-money-laundering pipelines assign overall compliance, and an explicit per-channel audit of base-context fields for label-correlated content (which uncovered, and fixed, a substantial description-field scenario-leakage problem in the dataset’s earlier versions; the empirical impact of the leakage and the corrected results are reported throughout). A four-arm ablation matrix isolates the contribution of structured KG relations by varying only the grounding mode supplied to the language model. A parametric fifth arm sweeps per-transaction KG coverage from 100 % down to 0 % to characterise the architecture’s degradation under realistic incomplete-graph conditions. On the corrected, description-neutral dataset, the KG-augmented arm achieves accuracy 0.891 with perfect precision (1.000, zero wrongly-cleared violations across 1500 predictions), recall 0.691, and F1 0.817 across three independent repeats. Both no-KG baselines—a vanilla LLM and a rule-text-grounded LLM—collapse to majority-class-default behaviour at F1 of approximately 0.01, so the architecture roughly fifty-fold lifts F1 relative to the baselines while never wrongly clearing a violation. The KG-augmented arm’s accuracy advantage degrades smoothly with KG coverage and disappears at coverage below approximately 20 % (the dropout sweep crosses the no-KG baseline at p ∗ ≈ 0.81). The methodological contributions of the work include the rule-name neutrality discipline—a previously-undocumented prompt-design pitfall whereby loaded rule identifiers override structured per-rule relations in language-model reasoning, identified during the project’s first benchmark and described in detail as finding F-001 in the project record—and the description-channel audit protocol (F-007) that the thesis argues should be a default hygiene step in any synthetic-data pipeline feeding a language model.
IMPROVING HIGH SCHOOL STUDENTS UNDERSTANDING OF PHYSICS CONCEPTS THROUGH FLIPPED CLASSROOM
(SDU University, 2026) Yeldos N.
This study explores the flipped classroom approach that transforms traditional teaching by teaching learning content using technology outside the classroom. The attendance time is used for interactive, student-centered activities such as discussions and problem solving. The aim of the study is to investigate the effects of a modified flipped classroom model on students' conceptual understanding and motivation. Physics is often considered a difficult subject for secondary school students and is often associated with low conceptual understanding and limited participation in traditional frontal teaching. The flipped classroom model, which holds face-to-face lessons outside the classroom and uses classroom time for active learning, has proven to be a promising alternative for overcoming these challenges. This study investigated the impact of the flipped classroom approach on the conceptual understanding of physics in eighth grade. A quasi-experimental design was used with 41 students divided into an experimental group (flipped classroom) and a control group (traditional teaching). Both groups completed tests with 30 multiplechoice questions to capture conceptual understanding before and after the flipped classroom. In addition, a questionnaire was used to measure the motivation and commitment of the students. The results showed that although both groups improved, the experimental group achieved significantly higher test results in the post-test compared to the control group. A one-factor analysis of variance (ANOVA) revealed a statistically significant difference between the groups (p < 0.05). The experimental group also reported a higher level of commitment, motivation and self-confidence in learning physical concepts. These results suggest that the flipped classroom approach is an effective method for improving students' conceptual understanding and engagement in the subject of physics. By combining pre-lesson preparation with active, student-centered learning during the lesson, this model offers a viable alternative to traditional teaching methods.
The Role of Laboratory Work in University Chemistry
(SDU University, 2026) Mukhangali N.
The present research seeks to explore the effect of laboratory classes in the learning process at university level in terms of improving conceptual understanding among Kazakhstani chemistry students as well as the attitudes of their university professors about the effects of such classes on their performance. Thus, a mixed methods research strategy was utilized that consisted of the application of a quasiexperiment based on the pre- and post-test design with two groups (N=75 third year students: N1=38 and N2=37, respectively) and a set of semi-structured interviews conducted with 17 university professors. For validation purposes, the research instruments were assessed by a panel of experts using the Content Validity Index (CVI) and Kendall’s Coefficient of Concordance (W). In all four post-tests, the experimental group significantly outperformed the control group, with the effect size gradually increasing from η² = 0.388 to η² = 0.535, thus proving that the conceptual advantages gained in the course of laboratory instruction have an accumulative effect. Four key themes have been identified through content analysis of the interviews – the positive effect of laboratory instruction on conceptual development, the difference in performance of laboratory and theoretical students, the difficulties associated with laboratory instruction and its positive evaluation.
Enhancing students' functional literacy through authentic problems in equations and inequalities
(SDU University, 2026) Mataibayeva A.
This study examines the effect of authentic problems involving equations and inequalities on Grade 6 students’ mathematical literacy. A quasi-experimental quantitative design was employed with 26 students: 13 in the experimental group and 13 in the control group. Both groups completed a pre-test and a post-test. The experimental group learned through authentic real-world problems involving equations and inequalities, while the control group received traditional procedural instruction. Since the sample size was small and visual inspection showed some deviations from normality, non-parametric tests were used. The results showed that both groups improved after instruction. However, the Mann-Whitney U test indicated that the experimental group had significantly higher gain scores than the control group. In addition, the large effect size for the gain score comparison showed that this difference was practically meaningful. Therefore, authentic problems involving equations and inequalities can be considered a promising approach for supporting Grade 6 students’ mathematical literacy.
The Effects of PhET Interactive Simulation Activities on Secondary School Students’ Physics Achievement
(SDU University, 2026) Zhumakhan S.
PhET Interactive Simulations refer to digital learning technologies extensively used to educate students regarding science theories using a highly engaging approach. In this research, the effects of the use of PhET interactive simulation-based teaching on the academic performance of secondary school students' in physics were investigated. A quasi-experimental research design was adopted with an experimental and a control group for each of the three participating schools. Ninety-six participants from the eighth grade were sampled for this study. The instrument was found to be reliable (Cronbach’s α = 0.87). Descriptive statistics, Shapiro-Wilk, independent samples t-test, and ANCOVA were conducted to analyze data. The findings revealed that the mean score in the post-test was significantly higher for the experimental group than that of the control group (EG: M = 10.40, SD = 2.75; CG: M = 7.35, SD = 3.06). The independent samples t-test also showed that there was a significant difference between the post-test scores (t = -5.12, p < .001, Cohen's d = -1.045), indicating large effect size due to the use of PhET simulationsbased teaching approach. Similarly, ANCOVA results revealed a significant influence of the intervention (F = 23.9, p < .001). Based on the above analysis, the findings suggest that the use of PhET interactive simulation-based learning is more effective than traditional teaching method.