Résumé
Property value depends on multiple factors, some, like surface or number of rooms, are easy to quantify, but others, like amenities or views, are often described. Large Language Models (LLMs) offer an opportunity to quantify data from text descriptions and, as such, present a new way of analysing real estate offers to identify factors determining the price. An LLM is a type of artificial intelligence (AI) program that can recognise and generate text. LLMs use machine learning called deep learning (DL) to understand how characters, words and sentences function together. DL involves the probabilistic analysis of unstructured data, which ultimately makes it possible to recognise differences between pieces of content without human intervention. We are proposing a new method of classifying non-standard house and apartment characteristics to identify additional quantifiable dimensions that can be used in pricing estimation. Based on the analysis of real estate offers retrieved from websites in Luxembourg, we try to confirm if factors such as greenery or proximity to services can be extracted from the descriptive section.
langue originale | Anglais |
---|---|
état | Publié - 6 avr. 2024 |
Evénement | National Scientific Conference „e-Factory of Science” – XI edition - online, Pologne Durée: 6 avr. 2024 → 6 avr. 2024 Numéro de conférence: 11 https://promovendi.pl/efos11/ |
Une conférence
Une conférence | National Scientific Conference „e-Factory of Science” – XI edition |
---|---|
Titre abrégé | EFOS11 |
Pays/Territoire | Pologne |
période | 6/04/24 → 6/04/24 |
Adresse Internet |