A Toolkit for Transparency in Dataset Documentation – Google AI Weblog

As machine studying (ML) analysis strikes towards large-scale fashions able to quite a few downstream duties, a shared understanding of a dataset’s origin, growth, intent, and evolution turns into more and more vital for the accountable and knowledgeable growth of ML fashions. Nevertheless, data about datasets, together with use and implementations, is usually distributed throughout groups, people, and even time. Earlier this 12 months on the ACM Convention on Equity, Accountability, and Transparency (ACM FAccT), we printed Information Playing cards, a dataset documentation framework aimed toward rising transparency throughout dataset lifecycles. Information Playing cards are transparency artifacts that present structured summaries of ML datasets with explanations of processes and rationale that form the info and describe how the info could also be used to coach or consider fashions. At minimal, Information Playing cards embrace the next: (1) upstream sources, (2) information assortment and annotation strategies, (3) coaching and analysis strategies, (4) meant use, and (5) choices affecting mannequin efficiency.

In follow, two important elements decide the success of a transparency artifact, the flexibility to determine the data decision-makers use and the institution of processes and steering wanted to accumulate that data. We began to discover this concept in our paper with three “scaffolding” frameworks designed to adapt Information Playing cards to a wide range of datasets and organizational contexts. These frameworks helped us create boundary infrastructures, that are the processes and engagement fashions that complement technical and purposeful infrastructure mandatory to speak data between communities of follow. Boundary infrastructures allow dataset stakeholders to seek out frequent floor used to supply numerous enter into choices for the creation, documentation, and use of datasets.

In the present day, we introduce the Information Playing cards Playbook, a self-guided toolkit for a wide range of groups to navigate transparency challenges with their ML datasets. The Playbook applies a human-centered design method to documentation — from planning a transparency technique and defining the viewers to writing reader-centric summaries of complicated datasets — to make sure that the usability and utility of the documented datasets are nicely understood. We’ve created participatory actions to navigate typical obstacles in establishing a dataset transparency effort, frameworks that may scale information transparency to new information varieties, and steering that researchers, product groups and corporations can use to supply Information Playing cards that mirror their organizational rules.

The Information Playing cards Playbook incorporates the newest in equity, accountability, and transparency analysis.

The Information Playing cards Playbook

We created the Playbook utilizing a multi-pronged method that included surveys, artifact evaluation, interviews, and workshops. We studied what Googlers needed to find out about datasets and fashions, and the way they used that data of their day-to-day work. Over the previous two years, we deployed templates for transparency artifacts utilized by fifteen groups at Google, and when bottlenecks arose, we partnered with these groups to find out acceptable workarounds. We then created over twenty Information Playing cards that describe picture, language, tabular, video, audio, and relational datasets in manufacturing settings, a few of which are actually accessible on GitHub. This multi-faceted method offered insights into the documentation workflows, collaborative information-gathering practices, data requests from downstream stakeholders, and evaluation and evaluation practices for every Google group.

Furthermore, we spoke with design, coverage, and know-how specialists throughout the trade and academia to get their distinctive suggestions on the Information Playing cards we created. We additionally included our learnings from a collection of workshops at ACM FAccT in 2021. Inside Google, we evaluated the effectiveness and scalability of our options with ML researchers, information scientists, engineers, AI ethics reviewers, product managers, and management. Within the Information Playing cards Playbook, we’ve translated profitable approaches into repeatable practices that may simply be tailored to distinctive group wants.

Actions, Foundations, and Transparency Patterns

The Information Playing cards Playbook is modeled after sprints and co-design practices, so cross-functional groups and their stakeholders can work collectively to outline transparency with a watch for real-world issues they expertise when creating dataset documentation and governance options. The thirty-three accessible Actions invite broad, important views from all kinds of stakeholders, so Information Playing cards might be helpful for choices throughout the dataset lifecycle. We partnered with researchers from the Accountable AI group at Google to create actions that may mirror concerns of equity and accountability. For instance, we have tailored Analysis Gaps in ML practices right into a worksheet for extra full dataset documentation.

Obtain readily-available exercise templates to make use of the Information Playing cards Playbook in your group.

We’ve fashioned Transparency Patterns with evidence-based steering to assist anticipate challenges confronted when producing clear documentation, supply finest practices that enhance transparency, and make Information Playing cards helpful for readers from completely different backgrounds. The challenges and their workarounds are primarily based on information and insights from Googlers, trade specialists, and educational analysis.

Patterns assist unblock groups with advisable practices, warning in opposition to frequent pitfalls, and steered alternate options to roadblocks.

The Playbook additionally consists of Foundations, that are scalable ideas and frameworks that discover basic elements of transparency as new contexts of information modalities and ML come up. Every Basis helps completely different product growth levels and consists of key takeaways, actions for groups, and useful sources.

Playbook Modules

The Playbook is organized into 4 modules: (1) Ask, (2) Examine, (3) Reply, and (3) Audit. Every module accommodates a rising compendium of supplies groups can use inside their workflows to deal with transparency challenges that continuously co-occur. Since Information Playing cards have been created with scalability and extensibility in thoughts, modules leverage divergence-converge pondering that groups might already use, so documentation isn’t an afterthought. The Ask and Examine modules assist create and consider Information Card templates for organizational wants and rules. The Reply and Audit modules assist information groups full the templates and consider the ensuing Information Playing cards.

In Ask, groups outline transparency and optimize their dataset documentation for cross-functional decision-making. Participatory actions create alternatives for Information Card readers to have a say in what constitutes transparency within the dataset’s documentation. These handle particular challenges and are rated for various intensities and durations so groups can mix-and-match actions round their wants.

The Examine module accommodates actions to determine gaps and alternatives in dataset transparency and processes from user-centric and dataset-centric views. It helps groups in refining, validating, and operationalizing Information Card templates throughout a corporation so readers can arrive at affordable conclusions concerning the datasets described.

The Reply module accommodates transparency patterns and dataset-exploration actions to reply difficult and ambiguous questions. Matters coated embrace making ready for transparency, writing reader-centric summaries in documentation, unpacking the usability and utility of datasets, and sustaining a Information Card over time.

The Audit module helps information groups and organizations arrange processes to guage accomplished Information Playing cards earlier than they’re printed. It additionally accommodates steering to measure and monitor how a transparency effort for a number of datasets scales inside organizations.

In Observe

A knowledge operations group at Google used an early model of the Lenses and Scopes Actions from the Ask modules to create a custom-made Information Card template. Apparently, we noticed them use this template throughout their workflow until datasets have been handed off. They used Information Playing cards to take dataset requests from analysis groups, tracked the varied processes to create the datasets, collected metadata from distributors chargeable for annotations, and managed approvals. Their experiences of iterating with specialists and managing updates are mirrored in our Transparency Patterns.

One other information governance group used a extra superior model of the actions to interview stakeholders for his or her ML health-related initiative. Utilizing these descriptions, they recognized stakeholders to co-create their Information Card schema. Voting on Lenses was used to rule out typical documentation questions, and determine atypical documentation wants particular to their information kind, and vital for choices continuously made by ML management and tactical roles inside their group. These questions have been then used to customise present metadata schemas of their information repositories.


We current the Information Playing cards Playbook, a steady and contextual method to dataset transparency that intentionally considers all related supplies and contexts. With this, we hope to determine and promote practice-oriented foundations for transparency to pave the trail for researchers to develop ML programs and datasets which can be accountable and profit society.

Along with the 4 Playbook modules described, we’re additionally open-sourcing a card builder, which generates interactive Information Playing cards from a Markdown file. You’ll be able to see the builder in motion within the GEM Benchmark undertaking’s Information Playing cards. The Information Playing cards created have been a results of actions from this Playbook, through which the GEM group recognized enhancements throughout all dimensions, and created an interactive assortment software designed round scopes.

We acknowledge that this isn’t a complete answer for equity, accountability, or transparency in itself. We’ll proceed to enhance the Playbook utilizing classes discovered. We hope the Information Playing cards Playbook can turn into a strong platform for collaboratively advancing transparency analysis, and invite you to make this your personal.


This work was performed in collaboration with Reena Jana, Vivian Tsai, and Oddur Kjartansson. We need to thank Donald Gonzalez, Dan Nanas, Parker Barnes, Laura Rosenstein, Diana Akrong, Monica Caraway, Ding Wang, Danielle Smalls, Aybuke Turker, Emily Brouillet, Andrew Fuchs, Sebastian Gehrmann, Cassie Kozyrkov, Alex Siegman, and Anthony Keene for his or her immense contributions; and Meg Mitchell and Timnit Gebru for championing this work.

We additionally need to thank Adam Boulanger, Lauren Wilcox, Roxanne Pinto, Parker Barnes, and Ayça Çakmakli for his or her suggestions; Tulsee Doshi, Dan Liebling, Meredith Morris, Lucas Dixon, Fernanda Viegas, Jen Gennai, and Marian Croak for his or her help. This work wouldn’t have been attainable with out our workshop and research contributors, and quite a few companions, whose insights and experiences have formed this Playbook.

Newsletter Updates

Enter your email address below to subscribe to our newsletter

Leave a Reply