# Tracking Temporal Evolution of Graphs using Non-Timestamped Data

## TL;DR

This paper introduces YoutubeGraph-Dyn, a novel dynamic graph dataset from YouTube interactions, enabling research on temporal graph evolution without relying on timestamped data, and demonstrates its utility through community detection and forecasting tasks.

## Contribution

The paper presents a new dataset and methodology for creating evolving graphs from non-timestamped data, facilitating dynamic graph learning research.

## Key findings

- Dataset contains 416 snapshots over 104 days
- State-of-the-art algorithms effectively detect community migration
- Forecasting models perform well on non-timestamped data

## Abstract

Datasets to study the temporal evolution of graphs are scarce. To encourage the research of novel dynamic graph learning algorithms we introduce YoutubeGraph-Dyn (available at https://github.com/palash1992/YoutubeGraph-Dyn), an evolving graph dataset generated from YouTube real-world interactions. YoutubeGraph-Dyn provides intra-day time granularity (with 416 snapshots taken every 6 hours for a period of 104 days), multi-modal relationships that capture different aspects of the data, multiple attributes including timestamped, non-timestamped, word embeddings, and integers. Our data collection methodology emphasizes the creation of time evolving graphs from non-timestamped data. In this paper, we provide various graph statistics of YoutubeGraph-Dyn and test state-of-the-art graph clustering algorithms to detect community migration, and time series analysis and recurrent neural network algorithms to forecast non-timestamped data.