At Facebook, we apply statistics to generate insights and to improve decision making for a business that touches billions of people across the globe. It is increasingly important that we advance our statistical methodologies so we can make the best decisions for our community, our products and our infrastructure.
We also believe that our statistical challenges are interesting to a broad community of researchers. In order to support academic work that addresses our challenges and opportunities while producing generalizable knowledge, Facebook is pleased to offer three $50K research grants.
Researchers can submit proposals to address problems of applied statistics that have direct applications for producing more effective insights and decisions for data scientists and researchers. We have a large, active and diverse community of practitioners, so we are interested in a varied set of statistical topics, including but not limited to: experimentation, forecasting, predictive modeling, survey modeling and sampling. Here we describe some specific examples of topics that are important within Facebook, but we will gladly review proposals for areas that we have not yet considered.
- Design and analysis of experiments. Facebook uses randomized experiments to measure the benefits of the improvements we make to our products. We seek to maximize the useful information we learn from these experiments by improving how they are designed and analyzed. We are interested in methodologies which allow us to extend or enhance the standard experimentation framework: improving the precision of effect estimates, measuring/detecting heterogenous or time-varying effects, and estimating effects of many-valued or continuous-valued treatments. We are also actively interested in research on adaptive experimentation such as Bayesian optimization and reinforcement learning.
- Surveys, nonresponse bias and missing data. In order to improve our understanding of people’s preferences and experiences, Facebook researchers conduct surveys and design targeted feedback mechanisms. We are interested in practical methodologies that can reduce respondent burden, adjust for and measure nonresponse bias, account for measurement error, and make valid inferences for small and under-represented subpopulations.
- Forecasting. Facebook uses forecasts to detect anomalies, to set goals, and to plan effectively. There is no one-size-fits-all solution to the variety of forecasting problems we encounter in practice, and we would like to expand the set of available methodologies for robust forecasting in the diverse settings we encounter in practice. Areas of special interest include forecasting multiple related time-series and clustering/classification of time-series into related groups or categories.
- Statistical models of complex social processes. Facebook’s products help connect billions of people around the world and we often think of our products and systems as massive, time-varying networks. We would like to improve our models for understanding how social processes evolve on these networks over time, with an eye toward designing better products and making better decisions for people on Facebook.
- Causal inference with observational data. Often researchers at Facebook would like to answer causal questions even when it is not possible to conduct product tests. For instance, we may want to measure the effects of external events or understand the potential causes for anomalies that we observe in our data. Proposals in this area should improve our ability to suggest potential hypotheses for interesting phenomena or to credibly estimate the effects of known causes.
- Efficient sampling and prevalence measurement. An important problem for Facebook is monitoring and measuring how well we do at removing content that violates our Community Standards from our site, such as hate speech, spam and nudity. We are broadly interested in methods that can improve the speed, correctness and certainty with which we can measure the prevalence or reach of items/users with a rare property over time.
- Efficiency and correctness of human analysts. With hundreds of data scientists at Facebook working on a wide variety of research questions, we are interested in helping practitioners work more efficiently and effectively. Proposals in this area should aim to help us understand how analysts work with data in practice and how their work can be improved through better education, collaboration, visualizations, tools and processes.
We will prioritize proposals that are pragmatic, use-inspired and grounded in empirical applications. As a community of tool-builders, we will pay special attention to proposals that include plans for producing useful, easy-to-use and open-source software as part of the research output.
Payment will be made to the proposer’s host university as an unrestricted gift.
Proposals should include
- A summary of the project. Provide a clear and concise statement (maximum of 3 pages) explaining the area of focus, a description of expected contribution of the project, any relevant prior work, and a timeline with milestones and expected outcomes.
- Curriculum Vitae of the principal researchers
- A one-paragraph biography of the principal researchers
- A proposed budget description (1 page) including an approximate cost of the award and explanation of how funds would be spent
Timing and Dates
- Applications are now open. Applications close October 30th, 2018, 5:00 pm PST.
- Successful awardees will be notified by email by November 30th, 2018.
For questions related to this request for research proposals, please email firstname.lastname@example.org.