House Seeking

Visualizaton, Retrieval and Analytics of Location-Centered Data

What we have done?

  • Map-based Visual exploration on location-centred multi-dimensional data (WSDM’19, EuroVIS’18, JVLC’18)
  • Map-based Spatial Object Recommendation with Explainability (SIGIR’20)
  • Representative Spatial Object Selection (SIGMOD’18)
  • Realtime Popular House (POI) and Area (AOI) Monitoring (ICDE’17)

What problems we are solving?

  • Geo-located Multi-dimensional & Multi-granularity Data Visualization: (1) Aggregation of results at different levels of detail for efficient visualization rendering (Data Sampling techniques, Approximate clustering); (2) Visualization Design of the Geo-located Multi-dimensional Data.
  • Spatial Keyword (Range) Search for Facility Seeking: (1) Find house candidates which can reach all the facilities before further inspection/exploration - coles supermarkets (for food), good clinics or pathology center (for health), top-50 public schools (for education), train stations (for transportation), Bunnings (for house renovation), etc; (2) Find the smallest region that can cover all my expected facilities in this surburb.
  • Location-centered Data Integration & Cleaning: Collect information centered around a property from different channels (besides the information of a real estate property itself), including but not limited to: (1) Regional Profile (age, sex, income, safety, density, etc.); (2) Transportational Profile (real time home-to-office travel time); (3)Education Profile (school zones, school rankings).
  • Selection of Objects For Visualization & Interactive User Exploration over the Map: How to select the most representative k-size objects while they are not close to each other (which can maintain the visibility of all selected objects) at all levels of granularities when users want to see them in the map? How to maintain a seamless user experiences (i.e. keep selecting the up-to-date k-size objects) when users interact with the map by zooming and moving to keep changing their region of interest? Can we allow users to choose their own preferred similarity metrics and our approach is independent of that?
  • Spatial object recommendation with Explainability: When spatial granularity matters: What does a user expect when using a personalized map service for spatial object recommendation? We observe that the following questions may matter to cater for users’ intention: (1) what does the system recommend for users at different stages of data exploration when they zoom in or zoom out the map application? (2) why does the system recommend a particular spatial object to the users? In this project, we focus on the POI recommender system in a multi-level manner by involving spatial granularity information. To this end, we expect that users can be recommended effective POIs when they have personal requirements on spatial granularity, or even they do not provide any specific intentions on where to go. Our model can completely explore the intrinsic spatial containment relationship between a POI point and a suburb then provide recommendations for varying spatial granularity. Moreover, users will be acknowledged about additional hints on why such a recommendation is generated.
  • How to exploit the visualization techniques to strike a balance between structured query and ranked retrieval: Structured query like SQL may give few or no results, while there is no one-size-fits-all ranking model to meet every individual buyer’s requirement, not to mention the buyer’s requirement may change w.r.t. the search result. What shall we do? Customer A wants to buy a house with a swimming pool under the budget of 1M; however, there is no such house, but there are few houses asking for 1.01M with swimming pools. We provide a visualized comparison for those “borderline” houses, for users to quickly identify which attributes of her interest do not meet her requirement.



HomeSeeker assists users in understanding the local areas and local real estate market, exploring and finding candidate properties based on their individual requirements, and visually comparing properties/suburbs in multiple aspects.
Tutorials Point

Concave Cubes

ConcaveCubes is a cluster-based data cube to support interactive visualization of large-scale multidimensional urban data.
Tutorials Point


AOI-Visualization is a powerfull visulization framework to ssupport efficient and effective visualization of user-defined urban areas of interest in an interactive manner.
Tutorials Point