Online behavioral data is a valuable source of insights for researchers. However, data collected passively via tracking meters contains Personal Identifiable Information (PII). With the GDPR into force, the value of online behavioral data is constrained by the risk of disclosing PII. We present a machine-learning solution that significantly reduces the risk of revealing PII when sharing browsing data.
Online behavioral data is a valuable source of insights for researchers. However, data collected passively via tracking meters contains Personal Identifiable Information (PII). With the GDPR into force, the value of online behavioral data is constrained by the risk of disclosing PII. We present a machine-learning solution that significantly reduces the risk of revealing PII when sharing browsing data.
Posting and sharing photos is considered one of the most popular online activities. In Britain, Dutton et al., characterized this activity as the most frequent and engaging Internet activity. In the US, 62% of the adult Internet users stated that they post or share pictures on the Internet (PewResearch, 2013). Despite its relevance, no research has been done exploring the use of sharing photos to answer survey questions. Since most of the mobile devices nowadays have a camera, and most of the mobile users are used to upload photos to the Internet (Ibid), there is a clear opportunity to test the possibilities and limitations of answering questions by taking and sharing photos.
In the digital age, market researchers are looking for new methodologies to understand consumer behaviour. Behavioural data collection through passive metering adds value to survey based data collection, as it measures objective data better because it is not limited to human memory. Data collected in a passive way can be perceived as intrusive to the research participant's privacy and intimacy. In our presentation we intend to provide a model to empower the participants, carefully protect their personal information and at the same time preserve the utility of the data collected. This model can help researchers to effectively deal with difficulty in recruiting research participants as well as help prevent biases in the collected data.