Substructures in graphs follow a power-law distribution as words in natural languages. Graphs have such connections to natural languages and thus can be modeled by the neural language models.
Some results have been hidden because they may be inaccessible to you
Show inaccessible results