Enhancing Hotel Recommendations with AI: LLM-Based Review Summarization and Query-Driven Insights

Abstract: The increasing number of data a booking platform such as Booking.com and AirBnB offers make it challenging for interested parties to browse through the available accommodations and analyze reviews in an efficient way. Efforts have been made from the booking platform providers to utilize recommender systems in an effort to enable the user to filter the results by factors such as stars, amenities, cost but most valuable insights can be provided by the unstructured text-based reviews. Going through these reviews one-by-one requires a substantial amount of time to be devoted while a respectable percentage of the reviews won't provide to the user what they are actually looking for. This research publication explores how LLMs can enhance short rental apartments recommendations by summarizing and mining key insights from user reviews. The web application presented in this paper, named "instaGuide", automates the procedure of isolating the text-based user reviews from a property on the Booking.com platform, synthesizing the summary of the reviews, and enabling the user to query specific aspects of the property in an effort to gain feedback on their personal questions/criteria. During the development of the instaGuide tool, numerous LLM models were evaluated based on accuracy, cost, and response quality. The results suggest that the LLM-powered summarization reduces significantly the amount of time the users need to devote on their search for the right short rental apartment, improving the overall decision-making procedure.

• The paper introduces instaGuide, an AI application that uses LLMs to summarize and query hotel reviews in real time.
• It compares web scraping and third-party API methods for review retrieval, highlighting trade-offs in speed and legal compliance.
• Evaluation shows Gemini 1.5 Flash achieves rapid 3-second summaries, significantly enhancing user decision-making in hotel selection.

Enhancing Hotel Recommendations with AI: LLM-Based Review Summarization and Query-Driven Insights

The paper "Enhancing Hotel Recommendations with AI: LLM-Based Review Summarization and Query-Driven Insights" explores the development of an intelligent web application, "instaGuide," designed to enhance the user experience of finding and selecting short rental apartments on platforms such as Booking.com through AI-powered review summarization and query-based retrieval.

Introduction and Motivation

The proliferation of user-generated reviews on booking platforms like Booking.com and AirBnB complicates decision-making for travelers while also posing challenges for accommodation providers to utilize feedback effectively. Traditional recommender systems utilize various filtering methods; however, they often fail to extract meaningful insights from the vast amount of unstructured text-based reviews available.

The paper introduces "instaGuide," which leverages LLMs to automatically summarize reviews and support interactive user inquiries about specific aspects of properties. This approach aims to significantly streamline the decision-making process by condensing essential information and providing personalized insights.

Methodology

Review Retrieval

To gather reviews, two methods were tested: web scraping and third-party APIs. While web scraping offered faster retrieval times (approximately 5 seconds for 200 reviews) and no cost, it presented legal challenges due to the terms of service of platforms like Booking.com. Third-party APIs, such as those from Arel Ventures and Caprolok, provided compliant yet slower and costlier alternatives.

LLM Integration

A variety of LLMs, including GPT-4, GPT-4o, Claude 3.5 Sonnet, and Gemini 1.5 Flash, were tested for summarizing reviews. The choice involved balancing factors such as cost, response speed, and the capability to handle large context windows. Gemini 1.5 Flash emerged as the preferred model for its superior speed and zero cost under certain constraints, making it ideal for the academic scope of instaGuide.

Application Development

Developed in Python using the Django framework, instaGuide facilitates review summarization and query responses via a user-friendly interface. Docker was employed to ensure a consistent and portable development environment, streamlining deployment across different platforms.

Evaluation and Results

The system was evaluated based on the efficiency of review retrieval, the responsiveness of LLMs, and user feedback. Gemini 1.5 Flash provided the fastest response times, with impressive summarization and query results in just 3 seconds each. Claude 3.5 Sonnet was noted for its structured, concise responses. User feedback was overwhelmingly positive, highlighting significant time savings in the rental selection process.

Societal and Ethical Considerations

The instaGuide application underscores the need for ethical deployment of AI technologies, particularly considering the potential impacts on job markets and decision-making processes. While web scraping presents legal challenges, collaboration with platforms to incorporate such technologies could transform industry standards. Bias and AI over-reliance remain areas for ongoing research and consideration.

Conclusion

The research demonstrates that the integration of LLMs with RAG technologies significantly enhances the processing and personalization of review insights, providing practical improvements to user experience on rental platforms. While challenges persist, particularly around legal compliance and potential biases, the study paves the way for broader adoption of AI-driven enhancements across similar domains.

The source code for instaGuide is available on GitHub, supporting further exploration and development in this promising area of AI application.