ETHOS (Evaluative Text Heuristic for Objective Scrutiny) is a pioneering argument analysis system that evaluates the logical structure, validity, and rhetorical quality of text. Designed to serve as an AI arbiter of rational discourse, ETHOS provides comprehensive analysis of argumentation quality through sophisticated NLP techniques.
Using a combination of rule-based analysis and neural language models (BERT), ETHOS dissects texts to identify claims, premises, logical fallacies, and rhetorical devices—creating a complete picture of argument structure and quality. The system then visualizes these relationships and provides detailed assessments of logical soundness and rhetorical clarity.
🌟 Key Features
🧩 Advanced Argument Structure Analysis: Extracts claims and premises from text with high accuracy using dependency parsing and linguistic markers
🔗 Semantic Argument Linking: Uses BERT embeddings to establish relationships between premises and claims based on semantic similarity
⚠️ Sophisticated Fallacy Detection: Identifies common logical fallacies (Ad Populum, Hasty Generalization, False Dichotomy) through both rule-based and ML-based approaches
🎭 Rhetoric Device Recognition: Detects rhetorical questions, strong sentiment expressions, and superlatives that may influence argument perception
📊 Logical Soundness Scoring: Evaluates the overall logical quality of arguments with nuanced ratings
📈 Comprehensive Visualization: Generates clear argument graphs showing support relationships between premises and claims
💬 Natural Language Explanations: Provides human-readable analysis of logical structure and potential issues
🚀 Why ETHOS is Revolutionary
ETHOS represents a significant advancement in automated argument analysis:
Deeper than Syntax: Goes beyond simple keyword matching by using semantic understanding to link arguments
Multi-dimensional Analysis: Examines both structural (argumentation form) and rhetorical (persuasive techniques) aspects simultaneously
Visual Representation: Transforms complex argumentative structures into intuitive graphs
Balanced Assessment: Provides nuanced evaluation rather than binary judgments about argument quality
Transparency: Clearly highlights the evidence for each assessment, allowing users to understand the system's reasoning
💡 Intended Uses
ETHOS is designed for applications requiring sophisticated argument analysis:
📝 Academic Research: Analyze argument structures in papers, essays, and scholarly works
🎓 Educational Assessment: Evaluate student essays and argumentation skills
📰 Media Analysis: Assess the logical quality of news articles, editorials, and opinion pieces
💼 Professional Writing: Improve the logical coherence and persuasiveness of business communications
🗣️ Debate Preparation: Analyze strengths and weaknesses in argumentative positions
📊 Policy Evaluation: Examine the logical structure of policy proposals and position papers
🛠️ System Architecture
ETHOS consists of several modular components that work together seamlessly:
argument_analyzer: Extracts argument components (claims and premises) using dependency parsing and linguistic indicators
logic_analyzer: Identifies logical fallacies through rule-based patterns and machine learning
rhetoric_analyzer: Detects rhetorical devices using sentiment analysis and linguistic patterns
argument_visualizer: Creates graph-based visualizations of argument structure using semantic similarity
synthesis_engine: Combines all analyses to produce overall evaluations of argument quality
nlp_utils: Provides NLP infrastructure including spaCy and BERT embedding functionality
📊 Performance Highlights
ETHOS demonstrates strong capabilities in:
Argument Component Identification: Accurately extracts claims and premises with confidence scores
Semantic Relationship Detection: Successfully links related premises and claims based on meaning
Fallacy Recognition: Identifies common fallacies with explanatory context
Multi-language Support: Extend capabilities to additional languages
Interactive Visualizations: Enhanced visual representations with interactive elements
API Implementation: RESTful API for seamless integration with other applications
📋 Citation
If you use ETHOS in your research, please cite:
@software{ethos_2025,
author = {Abdullah Kocaman},
title = {ETHOS: The AI Arbiter of Rational Discourse},
year = {2025},
url = {https://huggingface.co/NextGenC/ETHOS},
version = {2.2},
description = {A comprehensive system for argument analysis, fallacy detection, and rhetorical evaluation}
}
ETHOS - Rasyonel Söylem için Yapay Zeka Hakemi
ETHOS Logo
Durum
Lisans
🔍 Model Açıklaması
ETHOS (Değerlendirici Metin Buluşsal Objektif İnceleme Sistemi), metinlerin mantıksal yapısını, geçerliliğini ve retorik kalitesini değerlendiren öncü bir argüman analiz sistemidir. Rasyonel söylemin yapay zeka hakemi olarak hizmet etmek üzere tasarlanan ETHOS, sofistike NLP teknikleri aracılığıyla kapsamlı argümantasyon kalitesi analizi sağlar.
Kural tabanlı analiz ve sinir dili modellerinin (BERT) bir kombinasyonunu kullanan ETHOS, iddiaları, gerekçeleri, mantık safsatalarını ve retorik araçları tanımlamak için metinleri detaylı bir şekilde inceler—böylece argüman yapısı ve kalitesinin tam bir resmini oluşturur. Sistem daha sonra bu ilişkileri görselleştirir ve mantıksal sağlamlık ve retorik netlik hakkında ayrıntılı değerlendirmeler sunar.
🌟 Temel Özellikler
🧩 Gelişmiş Argüman Yapısı Analizi: Bağımlılık ayrıştırma ve dilbilimsel belirteçleri kullanarak metinden iddiaları ve gerekçeleri yüksek doğrulukla çıkarır
🔗 Anlamsal Argüman Bağlama: Anlamsal benzerliğe dayalı olarak gerekçeler ve iddialar arasında ilişkiler kurmak için BERT gömüleri kullanır
⚠️ Sofistike Safsata Tespiti: Hem kural tabanlı hem de ML tabanlı yaklaşımlarla yaygın mantık safsatalarını (Ad Populum, Aceleci Genelleme, Yanlış İkilem) tespit eder
🎭 Retorik Araç Tanıma: Argüman algısını etkileyebilecek retorik soruları, güçlü duygu ifadelerini ve üstünlük belirteçlerini tespit eder