Abstract
Curves of trends of content interaction events (e.g., views, shares, likes, etc.) versus time are of interest to social media and content-sharing services. Clustering trends can be used to classify, profile, and understand content creators or influencers. Clustering can also be used to recognize abusers, e.g., users that attempt to generate fraudulent views or likes in an effort to boost advertising revenue. This disclosure describes techniques to define similarity functions, or equivalently, distance measures, to enable the clustering of trends into trend types.
Creative Commons License
This work is licensed under a Creative Commons Attribution 4.0 License.
Recommended Citation
Reznik, Aviv and van Delft, Bart, "Clustering of Curve Types using Similarity Scores", Technical Disclosure Commons, (June 09, 2021)
https://www.tdcommons.org/dpubs_series/4365