Lexical Features Are More Vulnerable, Syntactic Features Have More Predictive Power

TL;DR

This study compares the vulnerability of lexical and syntactic features in noisy text, finding lexical features are more sensitive to modifications but syntactic features have greater impact on classification performance.

Contribution

It provides a comparative analysis of how lexical and syntactic features are affected by text noise and their influence on model performance.

Findings

Lexical features are more sensitive to text modifications.

Syntactic features have a stronger impact on classification performance.

Results validated across multiple datasets and language types.

Abstract

Understanding the vulnerability of linguistic features extracted from noisy text is important for both developing better health text classification models and for interpreting vulnerabilities of natural language models. In this paper, we investigate how generic language characteristics, such as syntax or the lexicon, are impacted by artificial text alterations. The vulnerability of features is analysed from two perspectives: (1) the level of feature value change, and (2) the level of change of feature predictive power as a result of text modifications. We show that lexical features are more sensitive to text modifications than syntactic ones. However, we also demonstrate that these smaller changes of syntactic features have a stronger influence on classification performance downstream, compared to the impact of changes to lexical features. Results are validated across three datasets representing different text-classification tasks, with different levels of lexical and syntactic complexity of both conversational and written language.