Linguistically-Controlled Paraphrase Generation

TL;DR

This paper presents LingConv, a novel encoder-decoder framework for controlled paraphrase generation that allows fine-grained manipulation of linguistic attributes and improves output quality through an iterative quality control mechanism.

Contribution

LingConv introduces a new method for controlling 40 linguistic attributes in paraphrase generation and incorporates a quality control process to enhance accuracy and semantic fidelity.

Findings

Reduces attribute error by up to 34%.

Quality control mechanism improves attribute matching by 14%.

Enhances reliability of controlled paraphrase generation.

Abstract

Controlled paraphrase generation produces paraphrases that preserve meaning while allowing precise control over linguistic attributes of the output. We introduce LingConv, an encoder-decoder framework that enables fine-grained control over 40 linguistic attributes in English. To improve reliability, we introduce a novel inference-time quality control mechanism that iteratively refines attribute embeddings to generate paraphrases that closely match target attributes without sacrificing semantic fidelity. LingConv reduces attribute error by up to 34% over existing models, with the quality control mechanism contributing an additional 14% improvement.