Form follows Function: Text-to-Text Conditional Graph Generation based on Functional Requirements
This addresses the problem of automated graph generation for downstream tasks like molecule and knowledge graph design, though it is incremental as it builds on existing LLM and graph methods.
The paper tackles the problem of generating graphs based on functional requirements described in text, by fine-tuning a pretrained large language model with an inductive bias that incorporates graph structure through message passing layers. Results show that the proposed approach generates graphs that more closely meet the functional requirements, outperforming baselines by a statistically significant margin.
This work focuses on the novel problem setting of generating graphs conditioned on a description of the graph's functional requirements in a downstream task. We pose the problem as a text-to-text generation problem and focus on the approach of fine-tuning a pretrained large language model (LLM) to generate graphs. We propose an inductive bias which incorporates information about the structure of the graph into the LLM's generation process by incorporating message passing layers into an LLM's architecture. To evaluate our proposed method, we design a novel set of experiments using publicly available and widely studied molecule and knowledge graph data sets. Results suggest our proposed approach generates graphs which more closely meet the requested functional requirements, outperforming baselines developed on similar tasks by a statistically significant margin.