Identifying and Understanding Human Values in Text: A Tailorable LLM-Based Architecture

Forum topic · 小凯 · 2026-05-29

Summary

This paper introduces an LLM-based architecture for detecting and quantifying the intensity of human values in text, whether expressed explicitly or implicitly. As intelligent systems grow more autonomous, aligning their decisions with human values is becoming a key research focus beyond traditional utility-maximization models. Unlike prior approaches tied to specific value theories or dependent on complex prompt engineering, the proposed pipeline is modular and theory-agnostic, consisting of three coordinated modules: one that generates structured value specifications from the foundational texts of any theoretical framework, one that annotates text using these specifications, and one that assigns support/opposition levels based on rhetorical and semantic evidence. This modular design separates the conceptualization of values from the detection task, producing a scalable, reproducible pipeline adaptable to different value theories. Experiments on the ValueEval dataset demonstrate competitive detection performance. The work offers researchers and practitioners a flexible tool for value-aware AI decision-making and content analysis. Source: arXiv 2605.27373.

Paper Overview

Field: AI / Natural Language Processing
Authors: Eduardo de la Cruz Fernández, Marcelo Karanik, Sascha Ossowski
Published: 2026-05-28
arXiv: 2605.27373

Introduction

As intelligent systems become more autonomous, the scientific community is focusing on decision-making mechanisms that incorporate ethical and moral considerations, moving beyond traditional utility-maximization models. A key aspect of this goal is assessing how well decisions align with human values — which requires reliably identifying those values in text, whether explicit or implicit.

Key points

The paper proposes an LLM-based architecture to detect and quantify the intensity of human values in text.
It avoids the limitations of previous approaches that are tied to a specific value theory or require complex prompt engineering.
The architecture consists of three coordinated modules:
A module that generates structured value specifications from the foundational texts of any theoretical framework.
A module that annotates text using these specifications.
A module that assigns support/opposition levels based on rhetorical and semantic evidence.
The modular design separates the conceptualization of values from the detection task, yielding a scalable, reproducible pipeline adaptable to different theories.
Experiments on the ValueEval dataset show the pipeline achieves competitive detection performance.

Significance

By decoupling value theory from value detection, this work provides a tailorable framework for value-aware NLP — useful for ethical AI alignment, argument mining, and content analysis across diverse theoretical traditions.

Original Abstract (excerpt)

> As intelligent systems become more autonomous, the scientific community focuses on creating decision-making mechanisms that include ethical and moral considerations, unlike traditional utility-maximisation models. To achieve this, a key aspect is assessing how well these decisions align with human values. To this end, a promising line of research is centred on developing approaches based on Large Language Models (LLMs) to identify human values from text, whether explicit or implicit... This paper introduces a LLM-based architecture to detect and quantify the intensity of human values in text, avoiding the limitations of previous approaches tied to specific value theory or complex prompt engineering. The architecture comprises three coordinated modules: one that generates structured value specifications from the foundational texts of any theoretical fr...

*Auto-collected on 2026-05-29.*