Multimodal retrieval systems are essential for deriving semantic insights from visual and textual data. Achieving precise alignment between image captions and text queries remains a non -trivial challenge. Conventional methods such as Long Short -Term Memory ( LSTM )-based or transformer -based encoders often struggle to balance efficiency, semantic accuracy, and interpretability. These approaches typically lack robust mechanisms for handling the fuzzy nature of human language. This stu dy proposes a lightweight and int erpretable framework that combines Bootstrapped Language Image Pretraining ( BLIP )-generated captions with embeddings using Meta -Learning Rectified Hyperbolic Secant Group Norm -Lasso -based Long Short -Term Memory networks (MeL -ReHS 2LSTM) and uses Neutrosophi c Cosine Similarity -based Fuzzy Inference System (NCS -FIS) to improve semantic retrieval. The architecture uses Parentheses Parsing Abstract Syntax Tree (PP -AST) for the symbolic structuring of captions and Dilated Skip -Connection Graph Neural Network (DSC -GNN) for the enhanced context embeddings to provide the ability to perform both neural and rule -based analysis in parallel. A custom dataset of 20 real -world video clips was created for evaluation, including a variety of scenes such as interviews, team me etings, and sports events. Empirical results show that the accuracy of top -1 is 92.0%, top-5 recall is 96.7%, Bilingual Evaluation Understudy ( BLEU ) score is 0.7521, and the average matching time is 0.58ms, which overcomes the traditional retrieval baselines. The proposed methodology provides a good trade -off between accuracy, runtime performance and interpretability. Future work will explore the extension of the model to zero -shot domains and the combination of adaptive caption refinement strategies .

[1] Kumar R., “Visual Linguistic Model and its Applications in Image Captioning,” SN Computer Science , vol. 1, pp. 1 -12, 2024. https://osf.io/preprints/osf/zy9m4_v1

[2] Lee B., Bang J., Song H., and Kang B., “Alzheimer’s Disease Recognition Using Graph Neural Network by Leveraging Image -Text Similarity from Vision Language Model,” Scientific Reports , vol. 15, pp. 1 -14, 2025. https://doi.org/10.1080/1448837X.2024.2448376

[3] Li Q., Zhao F., Zhao L., Liu M., and et al., “A System of Multimodal Image -Text Retrieval Based on Pre -Trained Models Fusion,” Concurrency and Computation: Practice and Experience , vol. 37, no. 3, pp. 1 -11, 2024. DOI: 10.1002/cpe.8345

[4] Li X., Wang J., and Yan Z., “Can Graph Neural Networks be Adequately Explained? A Survey,” ACM Computing Surveys , vol. 57, no. 5 , pp. 1 -36, 2025. https://doi.org/10.1145/3711122

[5] Liu Z., Liu J., and Ma F., “Improving Cross -Modal Alignment with Synthetic Pairs for Text -Only Image Captioning,” in Proceedings of the 38 th AAAI Conference on Artificial Intelligence , Vancouver, pp. 3864 -3872, 2024. https://ojs.aaai.org/index.php/AAAI/article/view/ 28178

[6] Long Z., Killick G., McCreadie R., and Camarasa G., “Multiway -Adapter: Adapting Multimodal Large Language Models for Scalable Image -Text Retrieval,” in Proceedings of the IEEE International Conference on Acoustics, Speech and Signal Processing , Seoul, pp. 6580 -6584, 2024. https://ieeexplore.ieee.org/document/10446792

[7] Long Z., Liang K., Aragon -Camarasa G., Henderson P., and Mccreadie R., “Zero -Shot Interactive Text -to-Image Retrieval via Diffusion - Augmented Representations,” arXiv Preprint , vol. arXiv:2501.15379, pp. 1 -11, 2025. https://arxiv.org/abs/2501.15379v1

[8] Luo B., Luo B., Wang J., Zewen W., and et al., “Graph -Based Cross -Domain Knowledge Distillation for Cross -Dataset Text -to-Image Person Retrieval,” arXiv Preprint , vol. arXiv:2501.15052, pp. 568 -576, 2025. https://arxiv.org/abs/2501.15052

[9] Lyu S., Tian Z., Ou Z., Zhu Y., and et al., “TSVC: Tripartite Learning with Semantic Variation Consistency for Robust Image -Text Retrieval,” arXiv Preprint , vol. arXiv:2501.10935, pp. 19269 - 19277, 2025. https://arxiv.org/abs/2501.10935

[10] Osman A., Shalaby M., Soliman M., and Elsayed K., “Ar -CM -ViMETA: Arabic Image Captioning Based on Concept Model and Vision -Based Multi - Encoder Transformer Architecture,” The International Arab Journal of Information Technology , vol. 21, no. 3, pp. 458 -465, 2024. https://doi.org/10.34028/iajit/21/3/9

[11] Patel V., Modi A., Mistry H., Mishra A.., and et al., “From Alt -Text to Real Context: Revolutionizing Image Captioning Using the Potential of LLM,” International Journal of Scientific Research in Computer Science, Engineering and Information Technology , vol. 11, no. 1, pp. 379 -387, 2025. DOI: 10.32628/CSEIT25111238

[12] Patil D., “Multimodal Artificial Intelligence in Industry: Integrating Text, Image, and Audio for Enhanced Applications Across Sectors,” SSRN , pp. 1 -9, 2025. DOI: 10.2139/ssrn.5057428

[13] Qian X. and Liu B., “Enhanced Image -Text Retrieval Based on CLIP with YOLOv10 and Next -ViT,” in Proceedings of the 4 th International Conference on Computer Vision, Application, and Algorithm , Chengdu, pp. 134862K -134862K -6, 2024. https://doi.org/10.1117/12.3055876

[14] Qin X., Li L., Pang G., and Hao F., “Heterogeneous Graph Fusion Network for Cross - Modal Image -Text Retrieval,” Expert Systems with Applications , vol. 249, no. 1, pp. 123842, 2024. DOI: 10.1016/j.eswa.2024.123842

[15] Tan C., Zhang W., Qi Z., Shih K., and et al., “Generating Multimodal Images with GAN: Integrating Text, Image, and Style,” arXiv Preprint , vol. arXiv:2501.02167v1, pp. 1 -8, 2025. DOI: 10.48550/arxiv.2501.02167

[16] Wang J., Zhang H., Zhong Y., Liang Y., and et al., “Advanced Multimodal Deep Learning Architecture for Image -Text Matching,” in Proceedings of the IEEE 4 th International Conference on Electronic Technology, Communication and Information , Changchun, pp. 1185 -1191, 2024. https://ieeexplore.ieee.org/document/10594167

[17] Wei S., Tan A., and Wang Y., “A Multiscale Hybrid Model for Natural and Remote Sensing Image - Text Retrieval,” in Proceedings of the 6 th International Conference on Machine Learning, Big Data and Business Intelligence , Hangzhou, pp. 41 -45, 2024. DOI: 10.1109/MLBDBI63974.2024.10823800

[18] Yang X. and Tan L., “Multimodal Knowledge Graph Construction for Intelligent Question Answering Systems,” Australian Journal of Electrical and Electronics Engineering , vol. 22, no. 4, pp. 755 -764, 2025. DOI: 10.1080/1448837x.2024.2448376

[19] Yuan Y. and Xue H., “Multimodal Information Integration and Retrieval Framework Based on Graph Neural Networks,” Preprints , vol. 202501.0405.v1, pp. 1 -8, 2025. DOI: 10.20944/preprints202501.0405.v1

[20] Zhang Q., Zhou Y., Yang Z., and Chen A., “Cross - Modal Prominent Fragments Enhancement Aligning Network for Image -Text Retrieval,” in Proceedings of the IEEE International Conference on Multimedia and Expo , Ontario, pp. 1 -6, 2024. https://ieeexplore.ieee.org/document/10687706

[21] Zheng W., Han N., Hu Y., Kang P., and et al., “Cross -Modal Image Text Retrieval Based on Cross -Optimization and Joint Strategy,” in Proceedings of the International Conference on Image, Signal Processing, and Pattern Recognition , Guangzhou, pp.131802I -131802I - 11, 2024. https://doi.org/10.1117/12.3034096