Research papers

View our latest published research and white papers

Abstract

This study aims to explore the application of large-scale language models in detecting and reducing gender and age biases in job advertisements. To establish gender and age bias detectors, we trained and tested various large-scale language models, including RoBERTa, ALBERT, and GPT-2, and found that RoBERTa performed the best in detecting gender and age biases. Our analysis based on these models revealed significant male bias in job ads, particularly in the information and communication technology, manufacturing, transportation and logistics, and services industries. Similarly, research on age bias revealed a preference for younger applicants, with limited demand for older candidates in job ads. Furthermore, we explored the application of natural language generation using ChatGPT to mitigate gender bias in job advertisements.

We generated two versions of job ads: one adhering to gender-neutral language principles and the other intentionally incorporating feminizing language. Through user research, we evaluated the effectiveness of these versions in attracting female candidates and reducing gender bias. The results demonstrated significant improvements in attracting female candidates and reducing gender bias for both versions. Overall, gender bias was reduced, and the appeal of job ads to female candidates was enhanced. The contributions of this study include an in-depth analysis of gender and age biases in job advertisements in Australia, the development of gender and age bias detectors utilizing large-scale language models, and the exploration of natural language generation methods based on ChatGPT to mitigate gender bias. By addressing these biases, we contribute to the creation of a more inclusive and equitable job market.

Authors

Ruochen Mao, Liming Tan, Rezza Moieni, Nicole Lee

Get this paper

Abstract

In today’s rapidly evolving technology landscape, predicting future revenue trends is crucial for strategic decisionmaking. This research aims to forecast the revenue trajectories of leading tech companies from 2024 to 2029 by analyzing historical data from 2018 to 2023. Using advanced statistical models including ARIMA, ETS, and TBATS, and the Root Mean Square Error (RMSE) as a key accuracy metric to assess and compare model performance. Our findings highlight TBATS model that excels in adapting to the tech sector’s seasonal fluctuations and trends, incorporating cultural and market diversity, and providing valuable insights for industry leaders, investors, and policymakers.

This study fills a critical gap in long-term revenue prediction literature and demonstrates the significant role of advanced analytics in strategic planning for the technology industry, considering the influence of diverse cultural and market factors. While the TBATS model’s prowess is evident, future studies could examine the model’s predictive performance over an extended period to more thoroughly assess its viability as a long-term forecasting tool.

Authors

Neha Durgadmath, Rezza Moieni, Nicole Lee

Get this white paper

Abstract

Diversity Atlas provides real-time analytic insight into granular cultural and demographic diversity data to help organizations understand the diverse richness of their teams, which accurately reflects humanity’s rich cultural diversity. Diversity Atlas enables organizations to collect data that can transform the way an organization measures, understands, acknowledges and acts on its diversity, inclusion and equity (ID&E) strategies.

Authors

AWS

Get this white paper

Abstract

Even a decade ago, the field of Diversity Equity and Inclusion—as a field and as a specific job title—did not exist. Today, the field has hit an apex—likely as a result of the corporate zeitgeist with the “business case” for diversity consistently in the forefront of business news, demonstrating organisations with diverse workforces are more innovative, more profitable and have better staff retention. This research study aims to provide a comprehensive understanding of Diversity, Equity and Inclusion (DEI) practitioners in Australia by analyzing LinkedIn profiles. The study examines career patterns, job titles, industry distribution, and gender representation among DEI officers. Using a dataset of 1000 profiles, we investigate average employment durations, career shifts, and part-time DI roles. Our research utilizes descriptive statistics and visualizations to highlight key insights. By leveraging LinkedIn data, we enhance insights into the roles, tenure, and impact of DEI practitioners, contributing to a deeper understanding of this field’s dynamics.

Authors

Nicole Lee, Rezza Moieni & Peter Mousaferiadis

Get this paper

Abstract

This research paper explores the role of language in shaping cultural and social attitudes towards gender and the importance of using gender-inclusive language to promote gender equality and eliminate gender bias. Specifically, we analysed the Victorian government website in Australia over some years from 1970-2023 to Measure and compare the evolution of its language across this period in terms of gender neutrality. To conduct this research, we created three datasets by scraping data from three different websites. The first two datasets comprise a list of masculine and feminine words used in the English language, while the third dataset comprises a list of gender-neutral words approved by the Victorian government. We compared this list of words with the data provided by the Victorian government website, and based on this analysis, we assessed how gender-neutral the website was. The findings show that using gender-inclusive language is a powerful way to promote gender equality and eliminate gender bias. Moreover, our analysis reveals that the Victorian government website has become more gender-neutral over the years, which signals their commitment to promoting gender equality and inclusivity. Overall, our research underscores the importance of using gender-inclusive language in all communication, including website content, to create a more inclusive and equitable society.

Authors

Arushi Raichur, Nicole Lee & Rezza Moieni

Get this paper

Abstract

There are discussions about the importance of diversity in literature and in the media and minimizing gaps between minorities and majorities. In order to see if a community is making progress in minimizing these gaps and to measure success, there is an interest in being able to predict the diversity of communities given currently prevailing. There are well-designed data forecasting algorithms in data science using large data sets. However, diversity data has only been collected over the last few decades. This paper adopts algorithms formulated by Grey and ARIMA (Auto-Regressive Integrated Moving Average), using small data to predict the likely diversity of a cohort for a time in the near future. Our results demonstrate there is more confident forecasting for “country of birth”, but in terms of predicting linguistic and religious diversity, due to the changeable nature of these factors throughout an individual’s life, we would require further data to make any accurate prediction.

Authors

Rezza Moieni, Peter Mousaferiadis & Leyla Roohi

Get this paper

Abstract

Discussions about the importance of cultural diversity are abundant in professional literature and mainstream media. Despite intense interest, there is significant knowledge gap in what is meant by diversity and how it can be measured. This article proposes a set of quantifiable dimensions of diversity that can be benchmarked, compared over time, evaluated against adjustable variables, and used as the basis for recommendations about improved business performance. This research builds on literature that has identified existing models for measuring diversity. From previous research emerges a framework for determining a model for quantifying key parameters of cultural diversity. While our previous work considered these parameters to be ethnicity, language, and belief, in this article, we propose a new method for measuring cultural diversity that expands the primary categories to four. We introduce an index for measuring cultural diversity of groups based on four distinct measures of ethnicity, languages, worldviews (beliefs), and countries of birth of the people in the community. This index has been developed into a digital tool called Diversity Atlas.

Authors

Rezza Moieni, Peter Mousaferiadis & Leyla Roohi

Get this paper

Abstract

For diversity to serve as a competitive advantage and for companies to be successful, the diversity of their workforce should be synchronized with the community they are doing business in. This degree of similarity in diversity of an organisation to its customer base is termed as Mutuality. High synchronization builds a healthy relationship between an organisation and its customers leads to better customer service, thus, radically improve their business performance. However, at present, this concept of diversity & mutuality has been analytically neglected and most organisations in both the public and private sectors are grappling with this aspect. Thus, the goal of this research is to design an empirical formula for the Mutuality Index by using the method of cosine similarity which captures the orientation (the angle) of one attribute of diversity in an organisation with its customer base and determines the similarity between them. The research explains how this data-driven approach can help organisations become better at measuring, understanding, tracking and delivering more informed and better diversity strategies which eventually enhances their business performance and induces a stronger bottom line (Profits)

Authors

Rezza Moieni, Peter Mousaferiadis & Prateek Pateel

Get this paper