Using social media data in pedestrian and cyclists research

Subject

The research of pedestrian and cycle traffic and transportation operations involves various aspects. Performing such studies require huge information. Currently, it is based on expensive data sources which are not scalable. In the meantime, the advent of Web-enabled technologies makes social media data as a promising source of knowledge to tap for pedestrian and cyclists research. The main goal of this research is to further the scientific understanding of how social media could be used as the valuable data source in pedestrian and cyclists research. This goal is hindered by a number of scientific problems, including the understanding distinction in data quality compared with typical data sources, which makes the models and algorithms not fittable. The performance of the data-driven models and algorithms based on social media data is also unclear.

Scientific challenges

One of the scientific challenges is social media data quality, such as data bias due to the composition of its reference population, the content of the posts contains, the retrieval limitations imposed by the social media platform, and users are not always trustworthy in their account of what happens around them. A second challenge is provided by the schema and semantic heterogeneity of social media data, compared to traditional social data sources, and among different social media platforms. Another challenge is the enrichment of social media data with higher-order properties about its creator, or its content. Last but not the least, the different properties of social media data also call for a new class of data-driven models and algorithms to collect and provide information for pedestrian and cyclist modeling studies.

Societal relevance

Using social media as data source reduces the cost of performing pedestrian and cyclist research, and particularly make it in a scalable and near online manner. It also provides more insights and semantic information than ever before.

Vincent Xun Gong

Start/end date: Jun 2016 – Jun 2020
Daily supervisor: Winnie Daamen, Alessandro Bozzon
Promotor: Serge Hoogendoorn
/* */