Reviews From Latest Submitted Papers

These past week, a couple of my follow-up works on the preliminary Tracing the Visitor’s Eye case study based on Florence have been accepted. I will be presenting the position paper “Assessing pervasive user-generated content to describe tourist dynamics” at the First International Workshop on Trends in Pervasive and Ubiquitous Geotechnology and Geoinformation on the need and methods to understand the users practices behind the generation of digital footprints and their relation to the urban space. In addition, the paper “Digital footprinting: uncovering the presence and movements of tourists from user-generated content” will be published in the IEEE Pervasive Computing Magazine, special issue on Pervasive User-Generated Content. It showcases the ability to uncover the presence and movements of tourists from geo-referenced photos they explicitly make public, as well as from network data implicitly generated by users of mobile phones.

Besides encouraging reviews on an “exciting area of research with lots of potential” some comments suggest some areas I should focus my current and future data analysis on:

Reveal the unexpected
- “it would be interesting to hear whether any of the findings were unusual/unexpected”
- “delve deeper into how “digital footprints” provide novel observations/results compared with traditional knowledge (expend to other nationalities)”
- “also focus on people taking photos in their own city (how are they different?)”

In consequence, I am exploring methods to reveal abnormal behaviors from time series, and to profile users from their use of the system and their mobility.

Space and user profiling
- “other automatic group categorization are possible (type of camera?), segmentation of the cellphone data as well”
- “what areas are more prone to be taken with a camera phone than a digital camera? and how does this evolves?”
- “hypothesis: tagged data is likely to be of fairly high quality (accurate and semantically meaningful)”
- “different types of tourists may geocode photos differently (American tourists are much less adept at geocoding location apart from thos most famous landmarks than Italian tourists”

Validation
- “conclusions are conjectures and not yet validated (they are obvious and not too controversial)”
- “methods of validation will likely have to be developed”
- “how to assess the impact of the biases in the data?”
- “in what ways does the passively collected data accurately reflect reality?”
- “additional investigation into the reasons for the discrepancies between datasets seems warranted. Do the demographics of photographers match the demotraphics of the mobile phone users? This additional analysis may help more convincingly explain the validation discrepencies and provide an avenue for correcting biases.”

In consequence, I am collecting hard data from traditional surveys that should represent the reality to compare to and calibrate to if they correlate.

Heterogeneity of the analysis
- “tie more together explicitly each analysis (presence and flow)”
- “are there other types of analysis (besides presence, flow, semantics)”

In consequence, I have to set an exhaustive list of what can be analyzed for each type of digital footprint and digital shadow. Then explain the scope of each type of analysis and how they relate to each other.
In consequence, I have extended the features that characterize users (e.g. type of camera, taken time and upload time difference, categories of tags used, …) in a correlation matrix. Use of proximity function and clustering techniques to define groups of users. Use of Characterization techniques as seen previously in Methods to Study Flickr Users Behaviors.