Advanced search
Start date
Betweenand

A recommendation system as a service for real estate market based on geographic information

Grant number: 16/14823-7
Support type:Research Grants - Innovative Research in Small Business - PIPE
Duration: September 01, 2017 - August 31, 2019
Field of knowledge:Physical Sciences and Mathematics - Computer Science
Cooperation agreement: FINEP - PIPE/PAPPE Grant
Principal Investigator:Paulo Scarpelini Neto
Grantee:Paulo Scarpelini Neto
Company:Enterup Tecnologia em Sistemas Ltda - ME
City: São José do Rio Preto
Co-Principal Investigators:Guilherme Priólli Daniel
Assoc. researchers:Carlos Roberto Valêncio
Associated research grant:15/08191-5 - A hybrid recommendation framework for real estate based on geographic information and big data, AP.PIPE
Associated scholarship(s):19/01091-6 - A recommendation system as a service for real estate market based on geographic information, BP.TT
18/06897-6 - A recommendation system as a service for real estate market based on geographic information, BP.TT
17/23534-1 - Searching for a suitable real state has been a slow task to the potential end-user due to the amount of available options, BP.TT
+ associated scholarships 17/22167-5 - A recommendation system as a service for real estate market based on geographic information, BP.TT
17/22070-1 - A recommendation system as a service for real estate market based on geographic information, BP.PIPE
17/22044-0 - A recommendation system as a service for real estate market based on geographic information, BP.PIPE - associated scholarships

Abstract

The real estate market is taken as one of the pillars of the modern economy. In Brazil the recent decades has provided a significant growth in this sector, mainly driven by the appreciation of real estate in big cities and in the less populated regions, furthermore, real estate projects and the demand for new properties remained in growth, mainly in the program "Minha Casa Minha Vida" from Federal Government. Front of this reality the real estate supply has grown considerably and the customers have faced a considerable amount of options, especially on the Internet where real estate websites gather in a single platform a large volumes of properties. In this context, the search for a property that meets the customer's needs has become a slow task, hindering the negotiations and decreasing the billing market. The recommendation systems consist of applications that help in the search for products and services by suggesting items that have a suitable combination with the expectations of users, consequently this type of system is considered an alternative to improve the buying experience in real estate. The development of a Recommendation System with a real estate focus consists of an interesting opportunity, but the existing techniques of suggestion are not meeting the specifics of this market satisfactorily ­ this is because the pricing, supply and demand of properties have specific characteristics that are directly affected by the location of the real estate projects. Moreover, recent research shows that current solutions for recommendation where the algorithms are performed off-line and batch, do not meet the online massive processing demand required by current search engines used in the real estate market. The project aims to build a Recommendation System as a Service based on technical feasibility analysis performed during Phase 1 of the PIPE/FAPESP project, which involved the creation and validation of a hybrid recommendation framework for real estate which included the development of an innovative strategy of recommendation based on the geographical location of the properties, as well as combination of this approach with other recommendation strategies strongly discussed in the literature. The creation of this kind of recommendation system has two main challenges: 1) build a recommendation system as a service, including challenges related to the storage of large volumes of data and scalable processing in real-time; 2) improve the recommendation framework built in Phase 1 in order to apply it in a real environment, including challenges related to improvement of the recommendation algorithm based on spatial data, as well as treatments for specific problems to recommendation algorithms such as Cold Start, First Rating and sparsity. The recommendation system is an essential part of the business model under development by EnterUp Technology, which is based on an increase in sales conversion rate for estate agents and brokers through a decrease in search time for ideal property and improving buying experience offered to customers. (AU)