Symbolic Graph Embedding using Frequent Pattern Mining
This work addresses the need for interpretable graph embeddings in relational data mining, offering a scalable and efficient solution for domains requiring symbolic representations, though it is incremental as it builds on existing inductive logic programming ideas.
The authors tackled the problem of learning interpretable node representations in graphs by proposing Symbolic Graph Embedding (SGE), which uses frequent pattern mining on node neighborhoods to create symbolic features, resulting in performance that outperforms DeepWalk and matches metapath2vec on a venue classification task, especially with small data, while scaling to millions of nodes and edges.
Relational data mining is becoming ubiquitous in many fields of study. It offers insights into behaviour of complex, real-world systems which cannot be modeled directly using propositional learning. We propose Symbolic Graph Embedding (SGE), an algorithm aimed to learn symbolic node representations. Built on the ideas from the field of inductive logic programming, SGE first samples a given node's neighborhood and interprets it as a transaction database, which is used for frequent pattern mining to identify logical conjuncts of items that co-occur frequently in a given context. Such patterns are in this work used as features to represent individual nodes, yielding interpretable, symbolic node embeddings. The proposed SGE approach on a venue classification task outperforms shallow node embedding methods such as DeepWalk, and performs similarly to metapath2vec, a black-box representation learner that can exploit node and edge types in a given graph. The proposed SGE approach performs especially well when small amounts of data are used for learning, scales to graphs with millions of nodes and edges, and can be run on an of-the-shelf laptop.