Beyond Conventional Text Rewriting: The First Unified Bidirectional Generative AI Expression Intensity Attenuation Framework for Human–AI Communication

The intensity of an expression substantially influences how information is perceived, interpreted, and responded to during human–human and human–AI communication. Expressions containing strong criticism, emotional intensity, certainty, urgency, demands, or confrontational language may communicate an intended message accurately while simultaneously introducing an unnecessarily high level of communicative intensity. Although previous research has investigated controllable text generation, sentiment manipulation, politeness transfer, attribute intensity control, and text detoxification, expression intensity attenuation has not been sufficiently established as a unified, independently controllable, and bidirectional communication capability encompassing both human-originated and AI-originated expressions. This paper introduces the Unified Bidirectional Expression Attenuation Framework (UBEAF), a conceptual framework designed to reduce the intensity of selected expressions while preserving their underlying semantic content, communicative intention, factual information, and relevant contextual constraints. Unlike conventional style transfer, which primarily focuses on transforming text between stylistic categories, UBEAF defines attenuation as an independent communicative operation that can be applied to an entire message, a sentence, a phrase, or a selected expression. The framework distinguishes four operational configurations: AI-to-human output attenuation, human-to-AI input attenuation, user-controlled expression attenuation, and AI-controlled expression attenuation. It additionally introduces adjustable attenuation levels, attribute-specific intervention, selective span processing, semantic preservation, reversibility, and adaptive intensity calibration. A formal task definition, operational architecture, evaluation methodology, and research propositions are developed to establish a foundation for future empirical investigation. To the best of our knowledge, this paper presents the first unified conceptualization of bidirectional, direction-independent, and selectively controllable expression intensity attenuation as an independent capability for generative AI-mediated human communication. The proposed framework provides a foundation for developing communication systems capable of adjusting not only what is expressed, but also how strongly it is expressed.

Authors

Publication Details

Journal
Zenodo (CERN European Organization for Nuclear Research)
Published
2026-11-17
DOI
https://doi.org/10.5281/zenodo.24298233
Primary Topic
Speech and dialogue systems
Type
preprint
Controls
|||
ALL TIME
JAN
FEB
MAR
APR
MAY
JUN
JUL
AUG
SEP
OCT
NOV
preprint

Beyond Conventional Text Rewriting: The First Unified Bidirectional Generative AI Expression Intensity Attenuation Framework for Human–AI Communication

Jincheng Zhang
6 links
Zenodo (CERN European Organization for Nuclear Research)
Speech and dialogue systems
preprint

Beyond Conventional Text Rewriting: The First Unified Bidirectional Generative AI Expression Intensity Attenuation Framework for Human–AI Communication

Jincheng Zhang
preprint en

Abstract

The intensity of an expression substantially influences how information is perceived, interpreted, and responded to during human–human and human–AI communication. Expressions containing strong criticism, emotional intensity, certainty, urgency, demands, or confrontational language may communicate an intended message accurately while simultaneously introducing an unnecessarily high level of communicative intensity. Although previous research has investigated controllable text generation, sentiment manipulation, politeness transfer, attribute intensity control, and text detoxification, expression intensity attenuation has not been sufficiently established as a unified, independently controllable, and bidirectional communication capability encompassing both human-originated and AI-originated expressions. This paper introduces the Unified Bidirectional Expression Attenuation Framework (UBEAF), a conceptual framework designed to reduce the intensity of selected expressions while preserving their underlying semantic content, communicative intention, factual information, and relevant contextual constraints. Unlike conventional style transfer, which primarily focuses on transforming text between stylistic categories, UBEAF defines attenuation as an independent communicative operation that can be applied to an entire message, a sentence, a phrase, or a selected expression. The framework distinguishes four operational configurations: AI-to-human output attenuation, human-to-AI input attenuation, user-controlled expression attenuation, and AI-controlled expression attenuation. It additionally introduces adjustable attenuation levels, attribute-specific intervention, selective span processing, semantic preservation, reversibility, and adaptive intensity calibration. A formal task definition, operational architecture, evaluation methodology, and research propositions are developed to establish a foundation for future empirical investigation. To the best of our knowledge, this paper presents the first unified conceptualization of bidirectional, direction-independent, and selectively controllable expression intensity attenuation as an independent capability for generative AI-mediated human communication. The proposed framework provides a foundation for developing communication systems capable of adjusting not only what is expressed, but also how strongly it is expressed.

Zenodo (CERN European Organization for Nuclear Research)
Speech and dialogue systems
AI Navigator

Ask Laika to Summarize, Analyze, and Connect papers live on the map.

Summarize Papers & Methodologies

Extract key findings, datasets, and comparative methods across publications.

Benchmark Rankings & Visual Analytics

Rank top research institutions, authors, funders, topics, and journals by Field-Weighted Citation Impact (FWCI) and paper volume with instant charts.

Connect Distant Disciplines

Bridge topological clusters on the map to find hidden collaborative intersections.

Laika access coming soon