Interfacing an LSP Homebuyer Assistant System with Real Estate APIs


Navchetan Singh

Oral Defence Date: 



TH 434


Professors Jozo Dujmovic and Arno Puder


In this project we present a programming system (LSPhome) which implements a comprehensive model for quantitative evaluation, comparison and selection of homes in online real estate. The model is based on the logic scoring of preferences method for system evaluation. This project aims at decomposing each online real estate home data into logic attributes and finding out the degree of satisfaction that each home offers to the buyer. The primary focus of this project is HomeSelector program, which is a component of LSPhome. HomeSelector program uses the application programming interface data offered by an online real estate data provider and Google Places API. The data are provided in XML format and HomeSelector program uses un-marshalling technique along with Java threads to extract data. The extraction process uses multithreading to optimize performance. HomeSelector program also uses filtering algorithms to eliminate unsuitable homes and reduce the workload. HomeSelector program processes the filtered list of homes and creates numeric values of all home attributes. The resulting data are used as inputs to two other programs, LSPloc and LSPcalc, which calculate a score for the quality of location and the quality of home, respectively. LSPhome system combines these two scores to produce a final value score for each home under evaluation and creates a sorted list of homes. We present the structure and components of LSPhome, the structure of home evaluation criterion, and a sample evaluation and comparison of homes. We also introduce and use shade diagrams, a new version of rectangular diagrams that represent the aggregation process in LSP criteria.


LSPhome, HomeSelector, LSPloc, LSPcalc, online real estate, un-marshalling, filtering, system attribute tree, elementary criterion, shade dia-gram


Navchetan Singh