Social Data: Biases, Methodological Pitfalls, and Ethical Boundaries

By Alexandra Olteanu, Carlos Castillo, Fernando Diaz, Emre Kıcıman. 2019

737 citations recorded by the source corpus.

Social data in digital form—including user-generated content, expressed or implicit relations between people, and behavioral traces—are at the core of popular applications and platforms, driving the research agenda of many researchers. The promises of social data are many, including understanding “what the world thinks” about a social issue, brand, celebrity, or other entity, as well as enabling better decision-making in a variety of fields including public policy, healthcare, and economics. Many academics and practitioners have warned against the naïve usage of social data. There are biases and inaccuracies occurring at the source of the data, but also introduced during processing. There are methodological limitations and pitfalls, as well as ethical boundaries and unexpected consequences that are often overlooked. This paper recognizes the rigor with which these issues are addressed by different researchers varies across a wide range. We identify a variety of menaces in the practices around social data use, and organize them in a framework that helps to identify them. “For your own sanity, you have to remember that not all problems can be solved. Not all problems can be solved, but all problems can be illuminated.” –Ursula Franklin1

Subfields: Communication, Sociology and Political Science, Transportation, Evaluation & Benchmarks, Safety, Robustness & Interpretability, AI Ethics, Fairness & Society

How researchers see this paper

The researchers whose trusted networks rank it highest, strongest advocate first. Every researcher gets a different order.

In Communication

In Sociology and Political Science

In Transportation

In Evaluation & Benchmarks

In Safety, Robustness & Interpretability

In AI Ethics, Fairness & Society

Citations in the corpus

Language (Technology) is Power: A Critical Survey of “Bias” in NLP (2020, background): “What are the impacts of annotation guidelines, annotator assumptions and perceptions (Olteanu et al., 2019; Sap et al., 2019; Geiger et al., 2020), and annotation aggregation processes (Pavlick and Kwiatkowski, 2019)?”
Re-imagining Algorithmic Fairness in India and Beyond (2021, background): “5This issue is somewhat related to what [148] call “Non-individual agents”.”
Measurement and Fairness (2019, background): “, [27, 41, 50].”
Toward Gender-Inclusive Coreference Resolution (2019, background): “annotation of hate-speech that annotator positionality matters (Olteanu et al., 2019).”
The Pushshift Reddit Dataset (2020, background): “Even if the “golden age” of API-driven computational social science and social computing research had not closed in the shadow of privacy scandals, it was nevertheless characterized by enormous inefficiencies in data collection and inequalities in access (Manovich 2011; Puschmann 2019), ethically-su”
A Survey on Bias and Fairness in Machine Learning (2019, background): “Linking bias arises from differences in the attributes of networks obtained from user connections, interactions, or activity [99].”
Knowledge Infused Learning (K-IL): Towards Deep Incorporation of Knowledge in Deep Learning (2019, methodology): “Moreover, the ML models that are deployed in the absence of transparency and accountability (Rudin 2019) and trained on biased datasets, can lead to grave consequences, such as potential social discrimination and unfair treatment (Olteanu et al. 2019).”
A Survey of Race, Racism, and Anti-Racism in NLP (2021, background): “The representativeness of data sets is a well-discussed issue in social-oriented tasks, like inferring public opinion (Olteanu et al., 2019), but this issue is also an important considera-5 tion in ‘neutral’ tasks like parsing (Waseem et al., 2021).”
The ethics of algorithms: key problems and solutions (2020, neutral): “Data quality—the timeliness, completeness and correctness of a dataset—constrains the questions that can be answered using a given dataset (Olteanu et al. 2016).”
When Are Search Completion Suggestions Problematic? (2020, background): “As in our study, the type of scenarios being surfaced is bounded by how the experimental or training datasets are sampled and generated, as well as by the quality of crowd annotations and self-reports [57].”
Detection and Resolution of Rumours in Social Media (2017, background): “Researchers attempt to clean social media data [Gao et al. 2014] and try to get rid of population biases [Olteanu et al. 2016] to understand how social media shapes society’s perceptions on issues, products, people, etc. [Goodman et al. 2011].”
A Hierarchy of Limitations in Machine Learning (2020, background): “The decision of what data to bring to a particular problem supersedes all other limitations and questions (Mallows, 1998), but data has also been covered extensively, most recently around social media data in comprehensive review by Olteanu et al. (2019).”
Algorithmic Bias in Education (2021, background): “While raising overlapping concerns with Suresh and Guttag (2020), Olteanu et al. (2019) highlight the potential for bias specifically in the analysis of social data, such as user-generated content, online behavior and networks.”
A Critical Review of Online Social Data: Biases, Methodological Pitfalls, and Ethical Boundaries (2018, background): “Our tutorial has a companion survey paper [26].”
Rehumanized Crowdsourcing: A Labeling Framework Addressing Bias and Ethics in Machine Learning (2019, background): “In regards to biases originating from crowd work, several prior works have looked at biases introduced by the process of labeling via crowds (e.g., [3, 5, 6, 8, 9, 23, 24, 26, 34]).”
Collecting a Large-Scale Gender Bias Dataset for Coreference Resolution and Machine Translation (2021, background): “…∈ [−100, 100], and positive 7According to Mehrabi et al. (2021), Population bias occurs when “statistics, demographics, representatives, and user characteristics are different in the user population represented in the dataset or platform from the original target population” (Olteanu et al., 2019).”
Stereotypical Bias Removal for Hate Speech Detection Task using Knowledge-based Generalizations (2019, background): “h bias sensitive attributes (“protected attributes”) explicitly known. While there have been efforts on identifying a broad range of unintended biases in the research and practices around social data [18], only recently some research work has been done about bias detection and mitigation in the fair”
The Limits of Abstract Evaluation Metrics: The Case of Hate Speech Detection (2017, background): “, identify online hate speech for content moderation) [4, 6].”