Revisiting Node Affinity Prediction in Temporal Graphs
This work addresses a common task in temporal graph learning with applications in social networks and recommender systems, but it is incremental as it builds on existing heuristics and models.
The paper tackles the problem of node affinity prediction in temporal graphs, where existing models are outperformed by simple heuristics, and proposes NAViS, a model that exploits the equivalence between heuristics and state space models, achieving state-of-the-art performance on the TGB benchmark.
Node affinity prediction is a common task that is widely used in temporal graph learning with applications in social and financial networks, recommender systems, and more. Recent works have addressed this task by adapting state-of-the-art dynamic link property prediction models to node affinity prediction. However, simple heuristics, such as Persistent Forecast or Moving Average, outperform these models. In this work, we analyze the challenges in training current Temporal Graph Neural Networks for node affinity prediction and suggest appropriate solutions. Combining the solutions, we develop NAViS - Node Affinity prediction model using Virtual State, by exploiting the equivalence between heuristics and state space models. While promising, training NAViS is non-trivial. Therefore, we further introduce a novel loss function for node affinity prediction. We evaluate NAViS on TGB and show that it outperforms the state-of-the-art, including heuristics. Our source code is available at https://github.com/orfeld415/NAVIS