InFillmore: Frame-Guided Language Generation with Bidirectional Context
This work addresses the challenge of controlling semantics in language generation for NLP applications, offering a flexible approach for various use scenarios, though it appears incremental as it builds on existing infilling methods.
The paper tackled the problem of bidirectional-context conditional language generation (infilling) by proposing a structured extension using Frame Semantic theory, resulting in frame-guided generation that allows explicit manipulation of intended infill semantics with minimal loss in distinguishability from human-generated text, as confirmed by automatic and human evaluations.
We propose a structured extension to bidirectional-context conditional language generation, or "infilling," inspired by Frame Semantic theory (Fillmore, 1976). Guidance is provided through two approaches: (1) model fine-tuning, conditioning directly on observed symbolic frames, and (2) a novel extension to disjunctive lexically constrained decoding that leverages frame semantic lexical units. Automatic and human evaluations confirm that frame-guided generation allows for explicit manipulation of intended infill semantics, with minimal loss in distinguishability from human-generated text. Our methods flexibly apply to a variety of use scenarios, and we provide a codebase and interactive demo available from https://nlp.jhu.edu/demos/infillmore.