The Dataset Nutrition Label: A Framework To Drive Higher Data Quality Standards

Abstract

Artificial intelligence (AI) systems built on incomplete or biased data will often exhibit problematic outcomes. Current methods of data analysis, particularly before model development, are costly and not standardized. The Dataset Nutrition Label (the Label) is a diagnostic framework that lowers the barrier to standardized data analysis by providing a distilled yet comprehensive overview of dataset "ingredients" before AI model development. Building a Label that can be applied across domains and data types requires that the framework itself be flexible and adaptable; as such, the Label is comprised of diverse qualitative and quantitative modules generated through multiple statistical and probabilistic modelling backends, but displayed in a standardized format. To demonstrate and advance this concept, we generated and published an open source prototype with seven sample modules on the ProPublica Dollars for Docs dataset. The benefits of the Label are manyfold. For data specialists, the Label will drive more robust data analysis practices, provide an efficient way to select the best dataset for their purposes, and increase the overall quality of AI models as a result of more robust training datasets and the ability to check for issues at the time of model development. For those building and publishing datasets, the Label creates an expectation of explanation, which will drive better data collection practices. We also explore the limitations of the Label, including the challenges of generalizing across diverse datasets, and the risk of using "ground truth" data as a comparison dataset. We discuss ways to move forward given the limitations identified. Lastly, we lay out future directions for the Dataset Nutrition Label project, including research and public policy agendas to further advance consideration of the concept.

arXiv commentFirst Draft May 2018

References in corpus (3)

• Man is to Computer Programmer as Woman is to Homemaker? Debiasing Word Embeddings
• A Nutritional Label for Rankings
• BayesDB: A probabilistic programming system for querying the probable implications of data

Cited by in corpus (41)

• On the Opportunities and Risks of Foundation Models
• Fairness in Machine Learning: A Survey
• Quality at a Glance: An Audit of Web-Crawled Multilingual Datasets
• Auditing of AI: Legal, Ethical and Technical Approaches
• Documentation Matters: Human-Centered AI System to Assist Data Science Code Documentation in Computational Notebooks
• Talking datasets: Understanding data sensemaking behaviours
• Ethics-Based Auditing of Automated Decision-Making Systems: Intervention Points and Policy Implications
• Data and its (dis)contents: A survey of dataset development and use in machine learning research
• Towards Accountability for Machine Learning Datasets: Practices from Software Engineering and Infrastructure
• Energy Usage Reports: Environmental awareness as part of algorithmic accountability
• Addressing "Documentation Debt" in Machine Learning Research: A Retrospective Datasheet for BookCorpus
• Predictive Multiplicity in Classification
• Between Subjectivity and Imposition: Power Dynamics in Data Annotation for Computer Vision
• Studying Up Machine Learning Data: Why Talk About Bias When We Mean Power?
• ABOUT ML: Annotation and Benchmarking on Understanding and Transparency of Machine Learning Lifecycles
• Multisource AI Scorecard Table for System Evaluation
• A framework for fostering transparency in shared artificial intelligence models by increasing visibility of contributions
• Identifying Bias in AI using Simulation
• FAT Forensics: A Python Toolbox for Algorithmic Fairness, Accountability and Transparency
• NaturalProofs: Mathematical Theorem Proving in Natural Language
• Artificial Intelligence Strategies for National Security and Safety Standards
• SoK: Machine Learning Governance
• Biases in Data Science Lifecycle
• German's Next Language Model
• An Ethical Highlighter for People-Centric Dataset Creation
• Trust and Transparency in Contact Tracing Applications
• A Methodology for Creating AI FactSheets
• Comprehensive and Comprehensible Data Catalogs: The What, Who, Where, When, Why, and How of Metadata Management
• Teaching Responsible Data Science: Charting New Pedagogical Territory
• Modeling Disclosive Transparency in NLP Application Descriptions
• Confounding Tradeoffs for Neural Network Quantization
• Green Lighting ML: Confidentiality, Integrity, and Availability of Machine Learning Systems in Deployment
• Know Your Model (KYM): Increasing Trust in AI and Machine Learning
• A Review of Bangla Natural Language Processing Tasks and the Utility of Transformer Models
• Detecting discriminatory risk through data annotation based on Bayesian inferences
• Statistical Learning to Operationalize a Domain Agnostic Data Quality Scoring
• Garbage In, Garbage Out? Do Machine Learning Application Papers in Social Computing Report Where Human-Labeled Training Data Comes From?
• Nose to Glass: Looking In to Get Beyond
• We Need to Talk About Data: The Importance of Data Readiness in Natural Language Processing
• Patterns Count-Based Labels for Datasets
• Outlining Traceability: A Principle for Operationalizing Accountability in Computing Systems