Author (s) :

---

1). Dr. Manvi Breja, The NorthCap University, Gurugram, Haryana, India

Abstract :

---

Answering non-factoid questions such as why and how type questions require exhaustive knowledge of contextualization and common-sense reasoning. Traditional neural techniques help in finding semantic similarity but ignore the causal context, which is handled by incorporating symbolic techniques for interpreting the concepts at reasoning level. The paper proposes context-aware neural-symbolic reasoning system which integrates the transformer-based context representation with several traditional causal, syntactic and semantic features to improve the performance of answer re-ranking module. The datasets with the evaluation metrics that are utilized for each module of framework are also discussed in detail. Further the framework is evaluated against standard baseline models such as BM25, DPR, Cross-Encoder and HGN on benchmark datasets HotPotQA, ELI5, BioASQ and Consumer Health QA. Integrating symbolic causal features with neural encoders helps in increasing the performance of ranking metrics as compared to basic models.

---

Author (s) :

---

1). Meet K. Vaghela, Darshan Institute of Engineering & Technology, Rajkot, Gujarat, India
2). Ketan Abhani, Darshan Institute of Engineering & Technology, Darshan University, Rajkot, Gujarat, India

Abstract :

---

This study investigates the influence of Nano-Silica (NS) and Graphene Oxide (GO) on the physical and mechanical properties of M50 grade concrete. Sixteen different concrete mixes were prepared, including a control mix, individual NS and GO mixes, and hybrid mixes using both materials. Workability tests (slump and compaction factor), strength tests (compressive, split tensile, and flexural), and non-destructive tests (rebound hammer and ultrasonic pulse velocity) were conducted. Results indicate that the addition of NS and GO reduces workability due to increased fineness, but significantly improves mechanical strength and surface hardness. The hybrid mix containing 1.0% Nano-Silica and 0.06% Graphene Oxide exhibited the highest strength improvement, showing approximately 30% enhancement compared to the control mix. The ultrasonic pulse velocity results classified this mix under excellent-quality concrete. The study concludes that the combined use of Nano-Silica and Graphene Oxide significantly improves the strength, durability, and quality of concrete, making it suitable for high-performance structural applications.

---

Author (s) :

---

1). M. N. Nwoga , Enugu State University of Science and Technology, Enugu, Enugu State, Nigeria
2). Charles Chinwuba Ike, Enugu State University of Science and Technology, Enugu, Enugu State, Nigeria
3). Ugwu Juliet Nneka, Enugu State University of Science and Technology, Enugu, Enugu State, Nigeria

Abstract :

---

In an effort to utilize agro waste products such like rice husk ash (RHA) and palm kernel shell ash (PKSA) to improve the geotechnical behaviour of soil. While also, reduce cost and minimized the lengthy procedures common in geotechnical laboratory tests. This study focused on developing a predictive model for optimizing the influence of RHA and PKSA on ESUT Agbani soil California Bearing Ratio (CBR). Tests revealed that the natural soil exhibited a CBR value of 6%. Scheffe’s model was employed to achieve optimization. Optimal proportions were determined to be 5.30: 0.26: 0.18: 0.28 for soil, RHA, PKSA and water correspondingly, produced an unsoaked CBR of 27%, equivalent to the output Y12. Formulated model estimated unsoaked CBR at 5 mm penetration for soil treated with RHA and PKSA was expressed as: ?CBR at 5mm?_ ((unsoaked)) =?17e?_1+?21e?_2+?13e?_3+?10e?_4+?32e?_1 e_2+?24e?_1 e_3+?38e?_1 e_4+?4e?_2 e_3+?10e?_2 e_4+?22e?_3 e_4. Model performance was verified by F-test and t-test. The computed F-statistic of 1.8845 was below the critical F-value of 3.1789, and the t-statistic of –0.5849 was lower than the critical t-value of 2.1199. These outcomes confirm that the model is valid at the 95% confidence level. It demonstrated no meaningful difference between the predicted and experimental CBR values. Accordingly, the model was considered adequacy and the null hypothesis was retained. Model validation was carried out using Microsoft Excel 2016.

---