TYPES
Towards transparency and privacy in the online advertising business

ECML/PKDD 2017 – Dealing with shortest paths in evolving graphs and imposed system workload imbalance

Disseminating Partner: TID

Type of disseminating activity: Conference

Target-profile / audience: Scientific Community (Higher Education, Research), Industry, Civil Society, Policy Makers

Number of attendants: 35

Section of the project covered by dissemination: Activities related to WP3 and WP4 on the design of algorithms that can perform dynamic graph mining (e.g., graph connecting advertisers with trackers and publishers).

Overview of feedback received: Many researchers provided feedback on the methods presented and challenged some assumptions on the dynamicity of the ecosystems considered. However, it was agreed that the methods still need to consider dynamics of the graphs under investigation due to ever-changing properties of the graph nodes and edges.

Reason for choosing this particular event to disseminate the project: ECML/PKDD is a top conference on machine learning, data mining and graph mining, and in particular, the large-scale time-dependent graphs workshop which extended the invitation for the keynote.

Integration with future exploitation plans: The talk presented work on scalable machine learning on dynamic graphs, that will be taken into utilization when TID will deploy the WIT (web proxy for privacy safeguarding) within the Niji corporate product.

Relevant links: http://tdlsg-ecmlpkdd17.isima.fr/index.php?action=home

Other comments: –

Leave a Reply