Multi-Agent Large Language Model Based Emotional Detoxification Through Personalized Intensity Control for Consumer Protection
#Emotional Detoxification #Multi-Agent LLM #Attention Economy #Consumer Protection #Information Sanitization #Emotional Balance #Keito Inoshita #arXiv
📌 Key Takeaways
- MALLET system uses four specialized agents to reduce emotional stimulation in content
- The system demonstrated up to 19.3% reduction in stimulus scores while preserving meaning
- Effectiveness varied by content category, with greatest impact on Sports, Business, and Tech content
- The system allows personalized emotional detoxification without restricting access to original content
📖 Full Retelling
Researcher Keito Inoshita introduced a novel Multi-Agent LLM-based Emotional deToxification (MALLET) system on arXiv on February 26, 2026, designed to protect consumers from excessive emotional stimulation in the attention economy that impairs calm decision-making. The innovative system consists of four specialized agents working in concert: Emotion Analysis Agent quantifies stimulus intensity using a 6-emotion BERT classifier; Emotion Adjustment Agent rewrites texts into two presentation modes, BALANCED (neutralized text) and COOL (neutralized text with supplementary information); Balance Monitoring Agent aggregates weekly consumption patterns to generate personalized advice; and Personal Guide Agent recommends presentation modes based on individual sensitivity. Testing on 800 AG News articles demonstrated significant stimulus score reduction of up to 19.3% while maintaining semantic preservation, with near-zero correlation between these two factors indicating they can be independently controlled. The research revealed category-specific effectiveness, with substantial reductions (17.8-33.8%) in Sports, Business, and Sci/Tech categories, though limited impact in the World category where facts are inherently high-stimulus. The system provides a framework for supporting calm information reception without restricting access to original content, addressing a growing concern in digital media consumption.
🏷️ Themes
Artificial Intelligence, Consumer Protection, Digital Wellbeing
📚 Related People & Topics
Attention economy
Economic view of human attention as a commodity
The attention economy refers to the incentives of advertising-driven companies, in particular, to maximize the time and attention their users give to their product. Attention economics is an approach to the management of information that treats human attention as a scarce commodity and applies econo...
Consumer protection
Protection of consumers against unfair practices
Consumer protection is the practice of safeguarding buyers of goods and services, and the public, against unfair practices in the marketplace. Consumer protection measures are often established by law. Such laws are intended to prevent businesses from engaging in fraud or specified unfair practices ...
Entity Intersection Graph
No entity connections available yet for this article.
Original Source
--> Computer Science > Artificial Intelligence arXiv:2602.23123 [Submitted on 26 Feb 2026] Title: Multi-Agent Large Language Model Based Emotional Detoxification Through Personalized Intensity Control for Consumer Protection Authors: Keito Inoshita View a PDF of the paper titled Multi-Agent Large Language Model Based Emotional Detoxification Through Personalized Intensity Control for Consumer Protection, by Keito Inoshita View PDF HTML Abstract: In the attention economy, sensational content exposes consumers to excessive emotional stimulation, hindering calm decision-making. This study proposes Multi-Agent LLM-based Emotional deToxification , a multi-agent information sanitization system consisting of four agents: Emotion Analysis, Emotion Adjustment, Balance Monitoring, and Personal Guide. The Emotion Analysis Agent quantifies stimulus intensity using a 6-emotion BERT classifier, and the Emotion Adjustment Agent rewrites texts into two presentation modes, BALANCED (neutralized text) and COOL (neutralized text + supplementary text), using an LLM. The Balance Monitoring Agent aggregates weekly information consumption patterns and generates personalized advice, while the Personal Guide Agent recommends a presentation mode according to consumer sensitivity. Experiments on 800 AG News articles demonstrated significant stimulus score reduction (up to 19.3%) and improved emotion balance while maintaining semantic preservation. Near-zero correlation between stimulus reduction and semantic preservation confirmed that the two are independently controllable. Category-level analysis revealed substantial reduction (17.8-33.8%) in Sports, Business, and Sci/Tech, whereas the effect was limited in the World category, where facts themselves are inherently high-stimulus. The proposed system provides a framework for supporting calm information reception of consumers without restricting access to the original text. Subjects: Artificial Intelligence (cs.AI) Cite as: arXiv:2602.23123 [c...
Read full article at source