Modeling Graph Structure via Relative Position for Text Generation from Knowledge Graphs
This work addresses graph-to-text generation for natural language processing applications, representing an incremental improvement with a novel attention mechanism.
The authors tackled the problem of generating text from knowledge graphs by introducing Graformer, a Transformer-based model that uses graph self-attention based on shortest path lengths to capture global patterns, achieving strong performance on AGENDA and WebNLG benchmarks with fewer parameters.
We present Graformer, a novel Transformer-based encoder-decoder architecture for graph-to-text generation. With our novel graph self-attention, the encoding of a node relies on all nodes in the input graph - not only direct neighbors - facilitating the detection of global patterns. We represent the relation between two nodes as the length of the shortest path between them. Graformer learns to weight these node-node relations differently for different attention heads, thus virtually learning differently connected views of the input graph. We evaluate Graformer on two popular graph-to-text generation benchmarks, AGENDA and WebNLG, where it achieves strong performance while using many fewer parameters than other approaches.