What Makes an Image Tagger Fair?

Barlas, Pınar; Kleanthous, Styliani; Kyriakou, Kyriakos; Otterbacher, Jahna

doi:10.1145/3320435.3320442

Cited by 12 publications

(6 citation statements)

References 21 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…Annotators (16 papers). In this work, human annotators assign gender for users, providers or subjects [3,30,70]. In [20], the authors use a dataset where the gender of book authors was annotated by library professionals.…”

Section: Gender Determinationmentioning

confidence: 99%

Much Ado About Gender: Current Practices and Future Recommendations for Appropriate Gender-Aware Information Access

Pinney,

Raj,

Hanna

et al. 2023

Preprint

View full text Add to dashboard Cite

Information access research (and development) sometimes makes use of gender, whether to report on the demographics of participants in a user study, as inputs to personalized results or recommendations, or to make systems gender-fair, amongst other purposes. This work makes a variety of assumptions about gender, however, that are not necessarily aligned with current understandings of what gender is, how it should be encoded, and how a gender variable should be ethically used. In this work, we present a systematic review of papers on information retrieval and recommender systems that mention gender in order to document how gender is currently being used in this field. We find that most papers mentioning gender do not use an explicit gender variable, but most of those that do either focus on contextualizing results of model performance, personalizing a system based on assumptions of user gender, or auditing a model's behavior for fairness or other privacyrelated issues. Moreover, most of the papers we review rely on a binary notion of gender, even if they acknowledge that gender cannot be split into two categories. We connect these findings with scholarship on gender theory and recent work on gender in humancomputer interaction and natural language processing. We conclude by making recommendations for ethical and well-grounded use of gender in building and researching information access systems. CCS CONCEPTS• Social and professional topics → Gender; • Information systems → Information retrieval.

show abstract

Section: Gender Determinationmentioning

confidence: 99%

Much Ado About Gender: Current Practices and Future Recommendations for Appropriate Gender-Aware Information Access

Pinney,

Raj,

Hanna

et al. 2023

Preprint

View full text Add to dashboard Cite

show abstract

“…While there exists all these mathematical fairness definitions and metrics, they tend to be conflicting and it is impossible to comply with all of them simultaneously, as shown by Chouldechova et al [38]. Consequently, few papers [18,62,105,106,195] study how the fairness of data-driven decisionsupport systems is perceived in order to choose the most relevant definitions taking into account stakeholders' preferences and mathematical trade-offs. Srivastava et al [173] show that one simple definition of fairness (demographic parity) solely matches the expectations of users of hypothetical systems.…”

Section: Conflicting Perceptions Of Fairnessmentioning

confidence: 99%

“…After choosing a fairness definition, deciding how to transform age into a categorical attribute can have direct bias consequences. Defining protected classes (male, black, [10][11][12][13][14][15][16][17][18][19][20][21][22][23]) or (male, black, [10][11][12][13][14][15][16][17][18][19][20][21][22][23][24][25]) as protected attributes would both surface and measure different biases. Different mappings of age to its protected class "young" can create different system behaviours: the granularity of the categories chosen would influence both the performance and fairness of the trained inference model.…”

Section: Bias-aware Schema Designmentioning

confidence: 99%

Managing bias and unfairness in data for decision support: a survey of machine learning and data engineering approaches to identify and mitigate bias and unfairness within data management and analytics systems

2021

View full text Add to dashboard Cite

The increasing use of data-driven decision support systems in industry and governments is accompanied by the discovery of a plethora of bias and unfairness issues in the outputs of these systems. Multiple computer science communities, and especially machine learning, have started to tackle this problem, often developing algorithmic solutions to mitigate biases to obtain fairer outputs. However, one of the core underlying causes for unfairness is bias in training data which is not fully covered by such approaches. Especially, bias in data is not yet a central topic in data engineering and management research. We survey research on bias and unfairness in several computer science domains, distinguishing between data management publications and other domains. This covers the creation of fairness metrics, fairness identification, and mitigation methods, software engineering approaches and biases in crowdsourcing activities. We identify relevant research gaps and show which data management activities could be repurposed to handle biases and which ones might reinforce such biases. In the second part, we argue for a novel data-centered approach overcoming the limitations of current algorithmic-centered methods. This approach focuses on eliciting and enforcing fairness requirements and constraints on data that systems are trained, validated, and used on. We argue for the need to extend database management systems to handle such constraints and mitigation methods. We discuss the associated future research directions regarding algorithms, formalization, modelling, users, and systems.

show abstract

“…Another set of works in the HCI domain, analyzes crowdsourced data from the OpenStreetMap to detect any potential biases such as gender and geographic information bias [30,109]. In a similar vein, two other studies run a crowdsourcing study to detect any bias on human versus algorithmic decision making [52,7]. Green and Chen [52] run a crowdsourcing study to examine the influence of algorithmic risk assessment to human decision making, while Barlas et.…”

Section: Direct or Indirect Discrimination Discoverymentioning

confidence: 99%

“…al. [7] compared human and algorithmic generated descriptions of people images in a crowdsourcing study in an attempt to identify what is perceived as fair when describing the depicted person. The execution of a crowdsourcing study for discrimination detection has also been used in IR systems [40,88].…”

Section: Direct or Indirect Discrimination Discoverymentioning

confidence: 99%

Mitigating Bias in Algorithmic Systems -- A Fish-Eye View

Orphanou,

Otterbacher,

Kleanthous

et al. 2021

Preprint

Self Cite

View full text Add to dashboard Cite

Mitigating bias in algorithmic systems is a critical issue drawing attention across communities within the information and computer sciences. Given the complexity of the problem and the involvement of multiple stakeholders -including developers, end-users and third-parties -there is a need to understand the landscape of the sources of bias, and the solutions being proposed to address them. This survey provides a "fish-eye view," examining approaches across four areas of research. The literature describes three steps toward a comprehensive treatment -bias detection, fairness management and explainability management -and underscores the need to work from within the system as well as from the perspective of stakeholders in the broader context.

show abstract

What Makes an Image Tagger Fair?

Cited by 12 publications

References 21 publications

Much Ado About Gender: Current Practices and Future Recommendations for Appropriate Gender-Aware Information Access

Much Ado About Gender: Current Practices and Future Recommendations for Appropriate Gender-Aware Information Access

Managing bias and unfairness in data for decision support: a survey of machine learning and data engineering approaches to identify and mitigate bias and unfairness within data management and analytics systems

Mitigating Bias in Algorithmic Systems -- A Fish-Eye View

Contact Info

Product

Resources

About