LinguaSynth: Heterogeneous Linguistic Signals for News Classification

2025-06-26

AI Summary

To address the poor interpretability, high computational cost, and strong black-box nature of deep learning models in news classification, this paper proposes a lightweight, interpretable classification framework based on multi-source linguistic feature fusion. The method systematically integrates five heterogeneous linguistic signals—lexical (part-of-speech tagging), syntactic (dependency structures), named entity (NER), word-level (Word2Vec), and document-level (Doc2Vec) representations—and empirically validates their complementary contributions to distributional semantics via feature interaction analysis. A transparent logistic regression classifier is employed as the prediction module. Evaluated on the 20 Newsgroups dataset, the framework achieves 84.89% accuracy—outperforming the TF-IDF baseline by 3.32%—while maintaining low computational overhead and full model transparency. This work establishes a novel paradigm for efficient, interpretable NLP, reconciling high performance with rigorous explainability and practical deployability.

Technology Category

• Natural Language Processing: Interpretability, Analysis, and Evaluation of NLP Models
• Machine Learning: Multi-class/Multi-label Learning & Extreme Classification
• Computer Vision: Interpretability, Explainability, and Transparency

Application Category

• Web Mining and Content Analysis: Large pretrained models with web data
• Search and Retrieval-Augmented AI: Multilingual and cross-lingual Web search
• Semantics and Knowledge: Methods to enhance, augment, integrate or synergize semantic models such as knowledge graphs and LLMs

Abstract

Deep learning has significantly advanced NLP, but its reliance on large black-box models introduces critical interpretability and computational efficiency concerns. This paper proposes LinguaSynth, a novel text classification framework that strategically integrates five complementary linguistic feature types: lexical, syntactic, entity-level, word-level semantics, and document-level semantics within a transparent logistic regression model. Unlike transformer-based architectures, LinguaSynth maintains interpretability and computational efficiency, achieving an accuracy of 84.89 percent on the 20 Newsgroups dataset and surpassing a robust TF-IDF baseline by 3.32 percent. Through rigorous feature interaction analysis, we show that syntactic and entity-level signals provide essential disambiguation and effectively complement distributional semantics. LinguaSynth sets a new benchmark for interpretable, resource-efficient NLP models and challenges the prevailing assumption that deep neural networks are necessary for high-performing text classification.

Problem

• Improving interpretability in NLP models
• Enhancing computational efficiency for text classification
• Integrating diverse linguistic signals effectively

Innovation

• Integrates five complementary linguistic feature types
• Uses transparent logistic regression model
• Achieves high accuracy without deep neural networks