Pooling Attention: Evaluating Pretrained Transformer Embeddings for Deception Classification

TL;DR

This study evaluates the effectiveness of pretrained Transformer embeddings, especially BERT, for deception detection, showing that simple pooling methods and lightweight classifiers can achieve strong performance and robustness.

Contribution

It systematically compares pooling strategies and classifier types for Transformer embeddings, highlighting the robustness and effectiveness of attention-based encodings in deception classification.

Findings

BERT embeddings with logistic regression outperform neural baselines.

Simple max or average pooling enhances robustness to sequence truncation.

Transformer-based encodings serve as solid, architecture-centric foundations for veracity tasks.

Abstract

This paper investigates fake news detection as a downstream evaluation of Transformer representations, benchmarking encoder-only and decoder-only pre-trained models (BERT, GPT-2, Transformer-XL) as frozen embedders paired with lightweight classifiers. Through controlled preprocessing comparing pooling versus padding and neural versus linear heads, results demonstrate that contextual self-attention encodings consistently transfer effectively. BERT embeddings combined with logistic regression outperform neural baselines on LIAR dataset splits, while analyses of sequence length and aggregation reveal robustness to truncation and advantages from simple max or average pooling. This work positions attention-based token encoders as robust, architecture-centric foundations for veracity tasks, isolating Transformer contributions from classifier complexity.