TEXTUAL AND DIGITAL TRACE DATA

Textual data in Surveys:

Open-ended questions in surveys are often ignored because they are difficult to analyze. Nevertheless, they can provide further insight into the phenomenon to be measured. 

I'm applying my expertise in text mining and NPL to analyse open-ended survey questions in the Supply Chain module of the Carbon Disclosure Project (CDP) Dataset.

The objective is to understand the rationale behind collaboration between suppliers and clients. We derive a new taxonomy for the reasons for collaborations applying topic modelling.

Digital Trace Data:

Big data, digital traces, and digital behavioural data are well-known terms in social science.  Essentially, it relates to the traces left online by human behavior and interactions. Some examples are social media posts, google trends and sensor data.

I studied the quality issues of social media data, and specifically Twitter, for the construction of social indicators.

I'm also interested in the construction of socio-economic indicators from social media and their integration with traditional data.

To collect such data I use web scraping and APIs. I apply different methodologies including dicionary-based analysis, supervised and unsupervised learning techniques.

Check my publication and coference presentation list for more insight on my work!