Design and Evaluation of a Retrieval-Augmented Generation System for Placement Data Querying in Academic Institutions

SmartSpace, a browser-native, totally open-source platform that significantly democratizes professional-grade 3D interior design, was introduced in this paper. SmartSpace removes the long-standing gap between conceptual planning and spatial visualization by combining a topologically sound half-edge data model with real-time bidirectional synchronization between an intuitive 2D floor plan editor and a high-fidelity 3D renderer. Without the need for installation, plugins, or costly licenses, the system enables precise wall thickness and apertures, achieves interactive frame rates (>55 FPS) on commodity hardware, and offers constraint-aware item placement. Its practicality and performance advantages over current web-based alternatives are confirmed by empirical results and real-world deployments in e-commerce and real-estate staging. The following important extensions will be the focus of future work: ·        Conflict-free replicated data types (CRDTs) allow for real-time collaborative editing to facilitate concurrent multi-user design sessions. ·        On gadgets like Meta Quest and Apple Vision Pro, native WebXR integration allows for AR furniture placement and immersive VR walkthroughs. ·        NURBS or Bézier edge representations are used to support curved and sloped walls in order to increase architectural expressiveness. ·        Integration of style transfer and layout optimization powered by machine learning for automated design support. ·        Improved compatibility for professional downstream workflows through standard export formats (IFC, glTF with BIM information). SmartSpace demonstrates that advanced spatial computing can—and should—be publicly available in all browsers, reducing obstacles to excellent interior design for academics, professionals, and homeowners alike.    You may find the full source code at https://github.com/amaankasu/PlacifyAI-Final

[1] P. Lewis et al., “Retrieval-Augmented Generation for Knowledge-Intensive NLP Tasks,” NeurIPS, 2020. [2] V. Karpukhin et al., “Dense Passage Retrieval for Open-Domain Question Answering,” EMNLP, 2020. [3] G. Izacard and É. Grave, “Leveraging Passage Retrieval with Generative Models for Open-Domain QA,” EACL, 2021. [4] N. Thakur et al., “BEIR: A Heterogeneous Benchmark for IR Models,” NeurIPS, 2021. [5] V. K. C. Vemulapalli, “Enterprise Generative AI Chatbot Architecture,” 2023. [6] R. Perumallapalli, “Conversational AI for Customer Support,” 2022. [7] P. S. R. P. Muntala, “AI-Powered Chatbots in Enterprise Applications,” 2021. [8] E. Asfoura et al., “Chatbot Conversational Systems for Enterprise Platforms,” Procedia CS, 2020. [9] D. Völker and M. Oberhauser, “Agentic AI in Enterprise Systems,” IEEE Access, 2023. [10] S. Hosseini and H. Seilani, “Agentic AI: A Systematic Review,” ACM CSUR, 2023.