HIL - Artificial Intelligence to facilitate property searches - System of recommendations with spatial and non-spatial visualisation for property search (2017)

Summary:

This project is an extension of a partnership between researchers in sociology, social psychology and IT and the Home in Love start-up, which provides a web platform to facilitate housing searches. The aim of the project is to improve the spatial recommendation procedures by proposing neighbourhoods to people seeking housing.
It consists of two parts. The first concerns the analysis and categorisation of neighbourhoods through a multidisciplinary perspective. Based on existing data, this involves producing an original typology of neighbourhoods in the Lyon conurbation and formalising these results and selected indicators in a computer model. The second part concerns the recommendation itself, with the aim of developing a method for matching the types of neighbourhood and the profiles of purchasers using the Home in Love platform.

The neighbourhood profiling is based on a database compiling three types of sources: public statistical data (notably that of INSEE at the scale of IRIS), information obtained from data mining and contained in property adverts (including semantics, press articles), and data collected by Home in Love on household preferences, expressed in the choice of adverts they view, in the preparation of neighbourhood types. This phase involves the close collaboration of researchers in the human and social sciences (data collection, choice of relevant indicators and classification methods) and researchers in information science (construction of the database, creation of an ontology).

With regard to matching, the data exploited to characterise the users and their project are supplied by Home in Love: the history of users' visits to the platform and in particular the adverts they view, their opinions on these adverts and the responses to the questionnaire on their housing project they fill in on signing up to the website. This will involve developing the recommendation algorithm integrating the neighbourhood analysis results into the recommendation procedure.

The ultimate goal of the project is the development of a neighbourhood recommendation system for people wishing to move house, based on neighbourhood categorisation integrating lessons from sociology and social psychology. The first part of the research will lead to the creation of a database of Lyon neighbourhoods and the development of a classification able to be extrapolated to other areas. The second part will give rise to the development of a recommendation algorithm taking into account neighbourhood types and the preferences of households visiting the Home in Love website.