GET PLACE GO: An AI-Powered Vibe Based Location Recommendation System Using NLP, RAG and Agentic AI

Abstract

Conventional location recommendation systems such as, Google Maps and Zomato are mainly depend on ratings, proximity, and keywords, which lead to inefficiency, when considering the subjective intent of the user, like the user\'s mood, the ambiance of the place, and the contextual preference. Therefore, the user has to wade through numerous results and is more likely to experience information overload. To solve this problem, this paper proposes an intelligent AI-powered personalized recommendation system, GET PLACE GO, to make vibe-based location recommendations and optimize day trips efficiently. The new system will uses technologies like NLP, RAG and Agentic AI to showcase and guide user interactions with these tech and make sure to enhance intelligent decision-making processes. These system also kind of uses dynamic data collection from platforms such as Google Maps, Zomato and Reddit. Sentiment and semantic analysis techniques are applied to get qualitative attributes like ambiance, among others.

Introduction

Digital platforms such as Google Maps and Zomato have transformed how people discover cafes, restaurants, and travel destinations. However, existing recommendation systems primarily rely on ratings and popularity metrics and often fail to understand subjective factors such as mood, atmosphere, purpose, or "vibe." Users frequently search using natural language expressions like "romantic café" or "calm workspace," which traditional systems struggle to interpret, resulting in irrelevant recommendations and information overload.

Recent advances in Artificial Intelligence (AI), particularly in Natural Language Processing (NLP), Large Language Models (LLMs), and Retrieval-Augmented Generation (RAG), enable systems to better understand user intent, context, and emotions while incorporating real-time external information to improve recommendation accuracy.

The literature review highlights several approaches:

• Chat-based recommendation systems improve personalization through conversational interactions but often lack real-time data integration.
• RAG systems combine retrieval and generation to provide accurate, up-to-date recommendations.
• Sentiment analysis extracts opinions from reviews but typically focuses on basic positive/negative sentiment rather than nuanced vibes.
• Agentic AI systems perform planning, reasoning, and multi-step decision-making but require significant computational resources.
• Context-aware systems improve recommendations by considering factors such as time, location, and user preferences.
• LLM-based recommenders can understand complex natural language queries without extensive training data.
• Knowledge-augmented systems enhance recommendation accuracy by combining structured knowledge with LLM capabilities.
• Embedding-based NLP approaches capture deeper semantic relationships and are effective for vibe-based recommendations.

The comparative analysis shows that existing systems each address specific aspects of recommendation but fail to provide a comprehensive solution. Common limitations include poor understanding of mood and vibe, lack of real-time adaptability, insufficient explainability, limited personalization, and weak integration between different AI components.

The identified research gap is the absence of an integrated recommendation framework that combines conversational AI, sentiment analysis, context awareness, real-time data retrieval, and multi-step reasoning. Current systems often operate independently, lack transparency, struggle with complex natural language understanding, and cannot effectively support advanced tasks such as itinerary planning.

To address these challenges, the proposed methodology introduces a multi-stage AI pipeline:

1. Data Collection – Gather structured and unstructured data from platforms such as Google Maps, Zomato, and Goibibo, including reviews, ratings, prices, and location information.
2. Data Preprocessing – Clean and normalize textual data using NLP techniques such as tokenization, lemmatization, and stop-word removal.
3. Feature Extraction – Use transformer-based models such as BERT and SBERT to generate semantic embeddings that capture contextual meaning and support advanced recommendation and retrieval processes.

Conclusion

These paper studies according to us how today’s AI-based recommendation systems are improving but also where older methods are not working as well.As per my study on past mechanism mostly depend on ratings,user feedback and simple keyword matching.These methods are useful,but they don’t always understand what as a user actually wants.User’s most of the time makes choices depending upon their mood,feelings,and overall experience.It also make difficult the to analyze these factors in cleay way and also quite challenging. Generally what happen is how newer technologies are helping to solve this problem. AI tools along with methods such as Natural Language Processing (NLP), Retrieval-Augmented Generation (RAG), Generative AI, and Agentic AI, are making recommendation systems more creative,flexible,shaper as well as smarter. These technologies help in understanding systems and also user intent better in meaningful way and adjust responses based on context,suggestions are more precisive manner and way,which improves personalization [3][5][16]. Many systems are not able to use real-time data properly, and some do not explain how they generate recommendations. Because of this,users may not trust the system and also do not agree with the results in practical real-time situations[6][7]. Most of the time generally,together these modern AI approaches can headed to the better recommendation mechanism in the future. The aim is must not only be to improve accuracy but also to make sure about the framework that feel more natural, easier to understand by human, and also closer to how humans so that it can actually make decisions.

References

[1] Y.Hou, S.Zhang, and X.Zhao,“Large Language Models as Zero-Shot Recommenders.” Proc. ECIR 2024. [2] OpenAI,“GPT-4 Technical Report.” arXiv:2303.08774, 2023. [3] J.Baek et al.,“Knowledge-Augmented LLMs for Context-Aware Recommendation.” Proc. WWW Conference 2024. [4] J.Xie et al.“TravelPlanner- A Benchmark for Planning with Language Agent.” arXiv:2402.01622 2024. [5] N.Shinn et al.,“Reflexion: Language Agents with Verbal Reinforcement Learning.” NeurIPS 2024. [6] W.Chen et al.,“Sentiment-Aware Recommendation Systems Using Transformers.” IEEE Access 2023. [7] R.Patel et al.,“Context-Aware Intelligent Recommendation Systems Using Hybrid AI.” IEEE Transactions on AI, 2024. [8] S.Nandi et al.,“Sankalp: Explainable AI-Based Recommendation System.” IEEE Access 2025. [9] S. Yao et al.,“ReAct: Synergizing Reasoning and Acting in LLMs.” arXiv 2023. [10] J. Xie et al.,“Planning with Large Language Models for Real-World Tasks,” arXiv 2024. [11] H. K. Naing et al., “AI-Driven Recommendation Using LLMs.” IEEE Conference, 2025. [12] A. Kumar and S. Sharma, “AI-Based Recommendation Systems Using NLP.” IJACSA 2023. [13] M. Matero et al.,“Opinion Mining from Social Platforms for Recommendation.” IEEE/ACM 2023. [14] K. Verbert et al.,“Natural Language Interfaces for Next-Gen Recommender Systems.”Springer 2023. [15] L.Friedman, S. Ahuja, D. Allen, T. Tan, H. Sidahmed, C. Long, J. Xie, G.Schubiner,A.Patel, H. Lara, et al., \"Leveraging Large Language Models in Conversational Recommender Systems,\" arXiv:2305.07961, 2023. [16] Y.Gao,Y.Xiong, X.Gao, K. Jia, J. Pan,Y.Bi,Y.Dai,J. Sun, M. Wang,and H.Wang,\"Retrieval-Augmented Generation for Large Language Models: A Survey,\" arXiv:2312.10997, 2023 (revised Mar. 2024)

Copyright

Copyright © 2026 Yash Gatkal, Om Karande, Moreshwar Vanmare, Dhanashree Phalke, Smita Shedbale. This is an open access article distributed under the Creative Commons Attribution License, which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited.