Update README.md

README.md

@@ -1,22 +1,69 @@
 ---
 title: >-
-  Employing Large Language Models to Identify Biased Language in Job
-  Descriptions.
+  Large Language Models for Detecting Bias in Job Descriptions.
 emoji: 🌍
 colorFrom: gray
 colorTo: pink
 sdk: static
 pinned: false
 license: apache-2.0
-short_description: Advancing NLP for bias detection in job ads.
+short_description: LLms for Detecting Bias in Job Descriptions.
 ---
 
-Our research aims to advance the application of Natural Language Processing (NLP) techniques in the detection of biases in job advertisements, which can affect applicant diversity by embedding prejudices within text such as age and gender.
+Abstract—This study explores the application of large language
+(LLM) models for detecting implicit bias in job descriptions,
+an important concern in human resources that shapes applicant
+pools and influences employer perception. We compare different
+LLM architectures—encoder, encoder-decoder, and decoder mod-
+els—focusing on seven specific bias types. The research questions
+address the capability of foundation LLMs to detect implicit
+bias and the effectiveness of domain adaptation via fine-tuning
+versus prompt tuning. Results indicate that fine-tuned models
+outperform non-fine-tuned models in detecting biases, with Flan-
+T5-XL emerging as the top performer, surpassing the zero-shot
+prompting of GPT-4o model. A labelled dataset comprising gold,
+silver, and bronze-standard data was created for this purpose
+and open-sourced1 to advance the field and serve as a valuable
+resource for future studies.
 
-Our hypothesis is that LLMs, due to their reasoning skills, can de-bias the HR ecosystem. We aim to validate this through three research questions:
 
-1) Can foundation LLMs faithfully identify implicit and subtle bias in job descriptions? Ideally, we hope to study the implicit information extraction capability of LLMs without being trained explicitly for the task. We plan to test various LLMs and scales (<= n billion-parameter models) to establish our hypothesis.
-2) Which prompting method (zero-shot, few-shot, or chain-of-thought) most efficiently identifies bias in text? We aim to carry out comprehensive experiments on different prompting strategies already proposed in the literature. We also aim to propose different prompting templates covering the three settings and document our observations across the LLMs.
-3) Does domain adaptation via fine-tuning foundational LLMs improve on prompt tuning? We aim to perform domain adaptation by fine-tuning the selected models on our datasets. We will seek to investigate if fine-tuning improves or lowers their performance under the same prompt settings.
+Short Introduction:
+In human resources, bias affects both employers and em-
+ployees in explicit and implicit forms. Explicit bias is
+conscious and controllable, but can be illegal in employment
+contexts. Implicit bias is subtle, unconscious, and harder to
+address. Implicit bias in job descriptions is a major
+concern as it shapes the applicant pool and influences appli-
+cants’ decisions. Bias in the language of job descriptions can
+affect how attractive a role appears to different individuals and
+can impact employer perception. The challenge is to efficiently
+identify and mitigate these biases.
 
-NOTE: Work in progress.  The benchmark dataset and models aren't complete.
+
+The application of large language models (LLMs) for de-
+tecting bias in job descriptions is promising but underexplored.
+This study examines the effectiveness of various LLM archi-
+tectures (encoder, encoder-decoder, decoder) less than 10 billion parameters in detecting implicit
+bias.
+
+
+We conceptualise the task of identifying implicit bias in
+job descriptions as a multi-label classification problem, where
+each job description is assigned a subset of labels from a
+set of eight categories—age, disability, feminine, masculine,
+general exclusionary, racial, sexuality, and neutral. This study
+investigates two primary research questions:
+
+1) Can foundation LLMs accurately detect implicit bias in
+job descriptions without specific task training? We evalu-
+ate the performance of three topical decoder-only models
+under four distinct prompt settings, assessing their ability
+to extract relevant information from job descriptions and
+identify implicit bias.
+
+2) Does domain adaptation via fine-tuning foundational
+LLMs outperform prompt tuning for detecting implicit
+bias in job descriptions? We fine-tune models with vary-
+ing architectures as text-classifiers on task-specific data
+and compare their performance to that of prompt-tuned
+models.