In brief
The candidate will study how to design a textual language for representing floor plans of apartments and also investigate how to evaluate floor plans automatically.
Background & motivation
Real-estate transactions involve complex assets with high value. Actors in the instant buyer (iBuyer) business, like Oslo-based startup Solgt.no, buy apartments directly from private customers without involving third parties like a real-estate broker. These actors must systematically and quickly evaluate the properties of an apartment before making an offer. One key property of an apartment is the floor plan. Floor plans have inherent structure and real-estate domain experts have informal rules about good and bad floor plans. Today, those who buy apartments in Norway need to study floor plan images manually, but in this work we assume that floor plans are available digitally
Problem
First, the candidate will study the problem of how to represent all floor plans textually using one common format. Second, the candidate will design and implement algorithms for tasks such as comparing floor plans or scoring floor plans with respect to quality as defined by domain experts. The starting point for the work will be existing work on domain-specific languages (DSLs). Third and if time, the candidate will study the problem of how to assist real-estate domain experts in representing existing floor plans (i.e., images), using the previously defined DSL for floor plans. This could take the form of an interactive tool for use by a domain expert, but with minimal user requirements, for example, no knowledge of the DSL necessary. Image analysis techniques for processing floor plans will not be part of the work, but the ability to visualize floor plans represented in the DSL could be useful.
Solgt.no and Norsk Regnesentral, together with other academic and industrial partners, are currently engaged in a multi-year research and innovation project called “Decision support system for professional real-estate price estimation” thus providing a context for the master thesis project
The candidate
The candidate should be interested in domain-modeling and (domain-specific) language design, and real-estate. Knowledge of subjects such as programming languages and formal modeling or logical methods would be advantageous. In addition, the candidate should have practical programming experience, ideally with a functional programming language, in order to realize the idea of the thesis.
Institutional setting
Supervision will take place in the offices of Norsk Regnesentral in Kristian Nygaards hus, the building next door to IFI. In addition, the candidate will have the opportunity to interact with technologists and real-estate domain experts at the startup company Solgt.no.
Getting in touch
For more information contact Bjarte M. Østvold, email: bjarte@nr.no