LinkedIn

S. Geyik, Stuart Ambler, K. Kenthapadi

We present a framework for quantifying and mitigating algorithmic bias in mechanisms designed for ranking individuals, typically used as part of web-scale search and recommendation systems. We first propose complementary measures to quantify bias with respect to protected attributes such as gender and age. We then present algorithms for computing fairness-aware re-ranking of results. For a given search or recommendation task, our algorithms seek to achieve a desired distribution of top ranked results with respect to one or more protected attributes. We show that such a framework can be tailored to achieve fairness criteria such as equality of opportunity and demographic parity depending on the choice of the desired distribution. We evaluate the proposed algorithms via extensive simulations over different parameter choices, and study the effect of fairness-aware ranking on both bias and utility measures. We finally present the online A/B testing results from applying our framework towards representative ranking in LinkedIn Talent Search, and discuss the lessons learned in practice. Our approach resulted in tremendous improvement in the fairness metrics (nearly three fold increase in the number of search queries with representative results) without affecting the business metrics, which paved the way for deployment to 100% of LinkedIn Recruiter users worldwide. Ours is the first large-scale deployed framework for ensuring fairness in the hiring domain, with the potential positive impact for more than 630M LinkedIn members.

Fedor Borisyuk, Shihai He, Yunbo Ouyang, Morteza Ramezani, Peng Du, Xiaochen Hou, Chen-Chen Jiang, Nitin Pasumarthy, Priya Bannur, B. Tiwana, Ping Liu, Siddharth Dangi, Daqi Sun, Zhuotao Pei, Xiao Shi, Sirou Zhu, Qianqi Shen, Kuang-Hsuan Lee, David Stein, Baolei Li, Haichao Wei, Amol Ghoting, Souvik Ghosh

In this paper, we present LiGNN, a deployed large-scale Graph Neural Networks (GNNs) Framework. We share our insight on developing and deployment of GNNs at large scale at LinkedIn. We present a set of algorithmic improvements to the quality of GNN representation learning including temporal graph architectures with long term losses, effective cold start solutions via graph densification, ID embeddings and multi-hop neighbor sampling. We explain how we built and sped up by 7x our large-scale training on LinkedIn graphs with adaptive sampling of neighbors, grouping and slicing of training data batches, specialized shared-memory queue and local gradient optimization. We summarize our deployment lessons and learnings gathered from A/B test experiments. The techniques presented in this work have contributed to an approximate relative improvements of 1% of Job application hearing back rate, 2% Ads CTR lift, 0.5% of Feed engaged daily active users, 0.2% session lift and 0.1% weekly active user lift from people recommendation. We believe that this work can provide practical solutions and insights for engineers who are interested in applying Graph neural networks at large scale.

Nicole Mishnick, D. Wise

Social media has revolutionized communication and changed how society accesses and receives information. As social media has become more prevalent, companies' advertising and marketing strategies worldwide have changed. In order to reach their target audience, organizations, including universities, have shifted their marketing plans to include social media. Research shows that social media campaigns enable universities to build positive relationships with potential undergraduate and graduate students. However, previous research on postgraduate social media use focuses on social media as a collective tool and does not analyze engagement by each platform. This study aimed to determine which social media platform, LinkedIn, Instagram, or Facebook, would have the highest engagement, as measured by likes, comments, and shares. Welch’s NOVA indicated a statistically significant difference in the engagement between platforms. However, post hoc analysis only showed statistically significant differences between Facebook and Instagram. These findings suggest that while Facebook may yield higher engagement than Instagram and LinkedIn, universities should consider all platforms when utilizing social media as a recruitment tool in higher education.

Fedor Borisyuk, Mingzhou Zhou, Qingquan Song, Siyu Zhu, B. Tiwana, Ganesh Parameswaran, Siddharth Dangi, Lars Hertel, Qiang Charles Xiao, Xiaochen Hou, Yunbo Ouyang, Aman Gupta, Sheallika Singh, Dan Liu, Hailing Cheng, Lei Le, Jonathan Hung, S. Keerthi, Ruoyan Wang, Fengyu Zhang, Mohit Kothari, Chen Zhu, Daqi Sun, Yun Dai, Xun Luan, Sirou Zhu, Zhiwei Wang, Neil Daftary, Qianqi Shen, Chen-Chen Jiang, Haichao Wei, Maneesh Varshney, Amol Ghoting, Souvik Ghosh

We present LiRank, a large-scale ranking framework at LinkedIn that brings to production state-of-the-art modeling architectures and optimization methods. We unveil several modeling improvements, including Residual DCN, which adds attention and residual connections to the famous DCNv2 architecture. We share insights into combining and tuning SOTA architectures to create a unified model, including Dense Gating, Transformers and Residual DCN. We also propose novel techniques for calibration and describe how we productionalized deep learning based explore/exploit methods. To enable effective, production-grade serving of large ranking models, we detail how to train and compress models using quantization and vocabulary compression. We provide details about the deployment setup for large-scale use cases of Feed ranking, Jobs Recommendations, and Ads click-through rate (CTR) prediction. We summarize our learnings from various A/B tests by elucidating the most effective technical approaches. These ideas have contributed to relative metrics improvements across the board at LinkedIn: +0.5% member sessions in the Feed, +1.76% qualified job applications for Jobs search and recommendations, and +4.3% for Ads CTR. We hope this work can provide practical insights and solutions for practitioners interested in leveraging large-scale deep ranking systems.

Shani Orgad

Digital spaces such as LinkedIn, the world’s largest professional digital network, constitute central sites for self-promotion, where job seekers and the employed present their polished “best” professional selves. However, in recent years, LinkedIn members are increasingly publishing accounts that highlight their vulnerabilities and struggles. This article examines the emergence of vulnerability on LinkedIn by analyzing how vulnerability is articulated in a sample of 40 posts (2021–2023). It identifies three genres: (1) Triumph over tragedy: vulnerability as a vector for self-growth and resilience; (2) Snap: vulnerability as a breaking point; and (3) Subversive commentary on self-promotion. On one hand, posting vulnerability on LinkedIn is a strategic form of digital self-branding, which monetizes vulnerability and depoliticizes its meanings. At the same time, vulnerability posts have the potential to form a basis for resistance to digital and work cultures’ glorification of overwork, individualized resilience and self-sufficiency, and the constant pressure to self-promote.