Property Recommendation System With Geospatial Data Analytics Andnatural Language Processing For Urban Land Use
Date of Award
Master of Science In Computer Engineering Degree
Electrical And Computer Engineering
Dr. Yongjian Fu
Dr. Pong Chu
Recently Cuyahoga County has been tremendously improved as properties are being constructed, renovated, or altered for new land use transactions on a nearly daily basis. Most existing property recommendation systems for the area simply rely on surface-level information and user history data to produce recommendations while failing to prioritize factors according to their importance and utilizing the location based complex information efficiently. This is leading them to become stagnant and simplistic in their approach and their accuracy is worsening as there are too many factors to be considered and location based complex yet useful information such as land use aspects of neighboring areas or information about people who are living or working in the area are often hard to be discovered. To combat these issues, this thesis proposes a modern property recommendation system with new approaches: 1) Employing data analytic methods to discover complex location based geospatial knowledge from big data processing, 2) Collecting and deriving summary information on people demographic data in the neighbor, and 3) Adopting natural language processing techniques for a user given phrase query to generate accurate candidate sets. Our recommendation system consists of three key components: 1) Using derived geospatial knowledge as new features and viewpoints for a better overall understanding of neighbor for a given property. 2) Incorporating Hotspot Analysis and data analytic methods to identify which areas are the v most ideal for each type of properties based on current and history data. 3) Allowing a user query in a sentence or phrase through natural language text processing techniques to create accurate candidates to tailor recommendations to a given individual user to return the Top-N ranked results. The experimental results show the effectiveness of these new approaches.
Riehl, Sean K., "Property Recommendation System With Geospatial Data Analytics Andnatural Language Processing For Urban Land Use" (2020). ETD Archive. 1219.