INFSCI 3350: Doctoral Seminar: Fall 2012

Projects

  • Francis Fofie and Qun Yu
    • Use the LBSN data from Brightkite and Gowalla (SNAP dataset) and if you can get it, the data from the Barcelona Bicing site.
    • Determine whether correlations exist between the check-ins and availability of bikes/spaces at various times where possible. If there are no overlapping times, try to do the same evaluation based on (general) days of the week and times of day. Note that you have to constrain the check-ins to Barcelona. You may have to do additional work to locate the venues accurately. If the Bicing data is not available, just compare the dynamics of the Brightkite and Gowalla datasets for different locations over time.
    • Use this paper to see if there are better models for predicting the time-series associated with biking availability or check-ins (aggregate or individual).
  • Marcela Gomez and Xiao Ma
    • Use the dataset from American Tower that has the locations and other information (e.g., tower heights) associated with base station towers that they own in the US.
    • Document the limitations of this dataset.
    • Use this dataset to develop model(s) for base station locations and coverages of base stations. The output of your work must be an algorithm that places base stations in a given area that is statistically similar to the actual locations of base stations. Towards this, you may have to analyze the data in many different ways (spatial - around cities, in rural areas, etc., by type - what kind of tower etc.)
    • Analyze the coverage overlaps (at various frequencies) in the different models that you produce and compare it with actual data.
  • Khalid Alkobayer and Xerandy
    • Use the dataset from the SSL Landscape paper and do a thorough analysis of it. First reproduce the analysis in the SSL Landscape paper. Next, take into consideration categories of sites, not just the top ranked sites in Alexa and the geography into account. Try more sophisticated analysis than what you see in the paper. Analyze whether Phishing sites are using certificates by checking with phishtank.com.
  • Abdulaziz Alashaikh
    • Talk with Anh Le and first use the data he used for this paper to do a multiway analysis to develop a better model for ETT in a network. If you can come up with a good model, we may be able to add to this paper and send it to a journal. Also, check the CRAWDAD site to see what other wireless measurements are available and see if a similar multi-way data analysis is useful.
  • William Garrard
    • Use the VA dataset to determine what factors make websites more accessible. Try to categorize users to see if there are differences among the way they may use websites. Use the sequence analysis from the palindromes paper to see if it applies. Try other models based on your analysis. (William Garrard)
  • Ke Zhang
    • Use the Gowalla dataset to see if friendships motivate check-ins. See whether it is possible to create a "ranking" of friends, since not all friends may motivate check-ins.
back to top