Do you have any great research ideas involving smartphones? Fed up working with aggregated data and want to see how individuals use their smartphones? We have a massive, global dataset (100 billion records, 17,000 devices) of real-world smartphone usage data.
The UbiComp/ISWC 2014 Programming Competition offers you the chance to access this dataset and publish your discoveries at the Device Analyzer Workshop. Publications will appear in the ACM Digital Library and the Ubicomp Adjunct Proceedings. We are also offering a prize of $250 to the best overall entry.
The Device Analyzer workshop will take place on Sunday 14th September. All welcome. Come along if you want to find out more about Device Analyzer or what other researchers have done with the data collected by the platform.
The proceedings for the 2014 programming competition can be found as part of the adjunct proceedings, here.
Device Analyzer is a Android smartphone application which runs on standard handsets and collects metadata, including:
Device Analyzer collects so much data it's impossible to summarise all of the data above. The complete list of what is collected is available here.
We now have over 100 billion records of Android smartphone usage from over 17,000 devices across the globe. Many of our contributors have agreed to share much of their data with other researchers, and in this competition we invite researchers from around the world to put this data to good use.
Some participating individuals have requested parts of their data are redacted in the released data set. This typically includes location and installed/running applications. Should your analysis aim to extract aggregate statistics from these redacted fields please email us to see if we are able to run your analysis on our servers and send you the summary statistics.
We're looking for novel research contributions using Device Analyzer data, which may include, but are not limited to:
Examples might include:
The competition is split into five phases:
A sample dataset from the Device Analyzer archive is available. Please email us with your your name and a brief description of your idea. We will email back a download link for the data and also any advice we have to offer on how to use the data to help solve the problem or idea you have. Note: you are not allowed to attempt to de-anonymise individuals in the dataset.
We strongly recommend you undertake Phase 2 after Phase 1. Access to the full dataset requires you to provide a summary of your idea and sign the terms and conditions. Please email us with your name and a description of your idea in less than 250 words. We will email back any advice we have to offer on how to use the data to help solve the problem or idea you have along with the terms and conditions for you to sign. Once we have received your signed terms and conditions, we will enable access to the full archive of data.
Papers should be submitted by email by the end of Sunday 22nd June 2014. We expect papers will be split into two sections. The first section (typically 2-3 pages) should outline your research question, overall approach, related work and key findings. The second section (typically 3-4 pages) and should describe the algorithmic steps required to process the device analyzer data, as well as graphs or other forms of summary which back up your key findings. The overall paper must not exceed 6 pages in length and must be formatted using the same template as the Ubicomp 2014 workshop papers (LaTeX | Word). Workshop publications will appear in the ACM Digital Library and the Ubicomp Adjunct Proceedings.
The competition committee will review the papers and produce a shortlist of papers to be presented and discussed at the workshop at UbiComp. Papers will be assessed on the soundness of the approach and the utility of the results to the community.
Paper presentations and demos will take place at the workshop held at UbiComp. The winner of the overall prize will be announced at the end of the workshop.