ImmoAgent

With its capability of processing multimodal input data(image, text and table), ImmoAgent seeks to apply LLMs to the real estate data by introducing a novel interpretable recommandation system.

Comment

Inspiration

Real estate investment has always been a dynamic field, influenced by property attributes, market trends, economic conditions, and investor sentiment. In recent years, large language models (LLMs) have emerged as powerful tools for analyzing and predicting various aspects of the real estate market.

What it does

This project seeks to apply LLMs to the real estate data by introducing a novel interpretable and predictable price estimatation and recommandation system. With its capability of processing multimodal input data(image, text and table), this smart agent RealEstateGPT will facilitate the interpretation of various data sources and data formats, and effectively generate useful market analysis, property valuation and customer assistance in real estate market.

How we built it

  1. Data Preprocessing: clean and preprocess input unstructured data with cognitive recognization model to extract meaningful information
  2. Embedding: extracted data in different format are converted to unified embeddings with fine-tuned LLM
  3. ML Prediction: vector embeddings are fed into trained ML model to get the predictions
  4. User Interaction: users are allowed to input natural language queries or commands, which the LLM then processes to interact with the embeddings. ## Challenges we ran into
  5. Mapping user preferences to attribute tags in db
  6. High relevante property recommendation ## What's next for ImmoAgent Market promotion and customer acquisition

Built With

Share this project:

Updates