Logo
ScienceStack
Table of Contents
Fetching ...
Sign In

An Empirical Study on the Characteristics of Bias upon Context Length Variation for Bangla

Jayanta Sadhu, Ayan Antik Khan, Abhik Bhattacharjee, Rifat Shahriyar

TL;DR

A clear dependency of bias metrics on context length is demonstrated, highlighting the need for nuanced considerations in Bangla bias analysis and considering this work as a stepping stone for bias measurement in the Bangla Language.

Abstract

Pretrained language models inherently exhibit various social biases, prompting a crucial examination of their social impact across various linguistic contexts due to their widespread usage. Previous studies have provided numerous methods for intrinsic bias measurements, predominantly focused on high-resource languages. In this work, we aim to extend these investigations to Bangla, a low-resource language. Specifically, in this study, we (1) create a dataset for intrinsic gender bias measurement in Bangla, (2) discuss necessary adaptations to apply existing bias measurement methods for Bangla, and (3) examine the impact of context length variation on bias measurement, a factor that has been overlooked in previous studies. Through our experiments, we demonstrate a clear dependency of bias metrics on context length, highlighting the need for nuanced considerations in Bangla bias analysis. We consider our work as a stepping stone for bias measurement in the Bangla Language and make all of our resources publicly available to support future research.

An Empirical Study on the Characteristics of Bias upon Context Length Variation for Bangla

TL;DR

A clear dependency of bias metrics on context length is demonstrated, highlighting the need for nuanced considerations in Bangla bias analysis and considering this work as a stepping stone for bias measurement in the Bangla Language.

Abstract

Pretrained language models inherently exhibit various social biases, prompting a crucial examination of their social impact across various linguistic contexts due to their widespread usage. Previous studies have provided numerous methods for intrinsic bias measurements, predominantly focused on high-resource languages. In this work, we aim to extend these investigations to Bangla, a low-resource language. Specifically, in this study, we (1) create a dataset for intrinsic gender bias measurement in Bangla, (2) discuss necessary adaptations to apply existing bias measurement methods for Bangla, and (3) examine the impact of context length variation on bias measurement, a factor that has been overlooked in previous studies. Through our experiments, we demonstrate a clear dependency of bias metrics on context length, highlighting the need for nuanced considerations in Bangla bias analysis. We consider our work as a stepping stone for bias measurement in the Bangla Language and make all of our resources publicly available to support future research.
Paper Structure (31 sections, 10 equations, 9 figures, 6 tables)

This paper contains 31 sections, 10 equations, 9 figures, 6 tables.

Table of Contents

  1. Introduction
  2. Linguistic Characteristics of Bangla: Gender Perspectives
  3. Methodology
  4. Baseline: WEAT and SEAT
  5. Contextualized Embedding Association Test (CEAT)
  6. Log Probability Bias Score Test
  7. Data Preparation
  8. Stimuli
  9. Contextualized Word Embedding
  10. Context Aware Templates
  11. Results and Evaluation
  12. Case Study: Effect of Context Variation on CEAT
  13. Case Study: Effect of Context Variation on Log Probability Bias Scores
  14. Conclusion
  15. Sample words and sentences for each experiment
  16. ...and 16 more sections

Figures (9)

  • Figure 1: Comparison between models on the change of effect size due to segment length variation. The variations for all categories are shown (from C1-C9). CEAT was done separately for definite segment length with sample size N=1000. (only statistically significant values with $p < 0.005$ are shown)
  • Figure 2: Prior Bias Score vs Corrected Bias Score diagrams for sentence structures S1 to S5 and negative traits. Experiment run on BanglaBERT (Large) Generator.
  • Figure 3: Male vs Female terms used for aggregation
  • Figure 4: Examples of Words (English Translations under each row) in Different WEAT Categories
  • Figure 5: Examples of sentences for different experiments
  • ...and 4 more figures