Request for proposals: Machine Learning on Graph for Interest-based Personalization

About

Facebook AI is pleased to invite university faculty to respond to this call for proposals about machine learning on graph data for interpretable personalized recommendation.

At Facebook we are looking to provide users with great experiences in terms of providing highly relevant content, capturing the niche, and truly personal interests and contextual preferences, and being intuitive and interpretable for user agency to control what they see.

We aim to develop a broader and deeper understanding across the industry of how to personalize user experience. We are building the next generation of ML modeling and platform technology centered around people and are committed to driving this innovation forward in an interpretable and trustworthy manner. To model users’ interest in a comprehensive way, we hope to leverage recent advances in machine learning on graph data, where various modeling techniques can be jointly built over a unified graph that organizes all the relevant entities and interactions.

Facebook AI is soliciting proposals to help accelerate research that targets our next personalized recommendation for better capturing users interest and fulfilling high level purposes in an interpretable and trustworthy approach, with focus on techniques of machine learning on graph. We hope to work with researchers to influence how we build our products, and bring values to our users and the whole society.

Up to two proposals will be selected, for projects at a cost of up to $75,000 each. Payment will be made to the Principal Investigator’s host university in the form of a sponsored research agreement with Facebook per the Terms and Conditions set forth below. We strongly encourage researchers from diverse backgrounds and of diverse abilities to apply.

Deadline for submissions are on March 15, 11:59pm PT.

Areas of Interest

Many highly impactful products pertaining to FB embrace a strong connectivity to users and their engaged items. From the machine learning perspective, it implies there is a user-centric contextual graph underlying various input data and models, motivating us to explore such a graph for decoding user interests and making personalized recommendations accordingly. Despite the remarkable progress in personalization, a fundamental question remains nascent: how can we make the personalized learning process and results be explainable, transparent, and fair? The answer to this question benefits a variety of our products for user experience and facilitates trustworthy interaction with users. Therefore, the research objective of this project is to explore the power and beauty of such user-centric contextual graphs on applications interesting to FB, with areas of interest that include, but are not limited to, the following:

1. Improving interest-based relevance in personalized recommendation

  • How to leverage contextual graph capture users’ niche and truly personal interests and contextual preferences?

  • How to optimize the usage of user interest-related data through a graph with heterogeneous entities/links, as such data are usually scattered and unstructured in practice?

  • How to effectively model user interest using a graph based approach from various heterogeneous sources and to capture the dynamics (interest changes with respect to time variance, seasonality, location, context, and evolves over time)?

  • How to improve cold start and make fast interest profiling for new users/creators introduced to a contextual graph?

  • How to improve the freshness (i.e. newly emerging contents) and democracy (i.e. not dominated by social celebrities or a few highly active creators)

2. Improving the understanding of users’ interest through ML modeling techniques

  • How to make the personalized recommendation intuitive and interpretable to our users, and give user agency to control what they see?

  • How to incorporate desired new angles for the interest representation using a graph based approach, such as explicitness, explainability, context-awareness and dynamics?

  • How to shift recommendation from who and what to how and why, leveraging a variety of sources, factors and interactions that can be represented by a heterogenous graph in a unified way?

  • How to address filter bubbles in personalized recommendation, such as to improve the balance between interest diversity (exploration) and exploitation?

  • How to bring up diversity and exploration into new topics/interests when relevant entities/factors are linked as a graph?

3. Optimizing user cohort characterization

  • How to leverage contextual graph and user cohort analysis to mitigate various unwanted biases in recommendation?

  • How to design efficient large scale graph clustering algorithms to model collective user segments in the use case of user cohort analysis, for industrialized recommendation applications?

  • How to improve hierarchical clustering for user-interest based recommendation and explore different interest granularity levels for sourcing and ranking?

Requirements

Prior to submitting a proposal, please confirm that your institution will agree to a contract for sponsored research from Facebook. You may assume that all IP rights shall belong to the university and the agreement will operate under “Open Science” terms meaning that all research will be available for publication in the public sphere. Application materials include:

  • A summary of the project (1–2 pages), in English, explaining the area of focus, a description of techniques, any relevant prior work, and a timeline with milestones and expected outcomes. You may choose to use this proposal template (optional).

  • A draft budget description (one page) including an approximate cost and explanation of how funds would be spent .

  • Curriculum Vitae for all project participants

  • Contact details for 1) A legal representative responsible for research contracts , and 2) an administrator or financial contact responsible for payments and invoicing.

Eligibility

  • Proposals must comply with applicable US and international laws, regulations and policies.

  • Applicants must be current full-time faculty at an accredited academic institution that awards research degrees to PhD students.

  • Applicants must be the Principal Investigator on any resulting award.

  • Facebook cannot consider proposals submitted, prepared or to be carried out by individuals residing in, or affiliated with an academic institution located in, a country or territory subject to comprehensive U.S. trade sanctions.

  • Government officials (excluding faculty and staff of public universities, to the extent they may be considered government officials), political figures, and politically affiliated businesses (all as determined by Facebook in its sole discretion) are not eligible.

Terms and Conditions

Please read these terms carefully before proceeding.

Facebook’s decisions will be final in all matters relating to Facebook RFP solicitations, including whether or not to select a proposal for funding and the interpretation of Facebook RFP Terms and Conditions. By submitting a proposal, applicants affirm that they have read and agree to these Terms and Conditions.

  • Facebook is authorized to evaluate proposals submitted under its RFPs, to consult with outside experts, as needed, in evaluating proposals, and to select or choose not to select proposals for funding using criteria determined by Facebook to be appropriate and at Facebook’s sole discretion. Facebook’s decisions will be final in all matters relating to its RFPs, and applicants agree not to challenge any such decisions.

  • Facebook will not be required to treat any part of a proposal as confidential or protected by copyright, and may use, edit, modify, copy, reproduce and distribute all or a portion of the proposal in any manner for the sole purposes of administering the Facebook RFP website and evaluating the contents of the proposal.

  • Personal data submitted with a proposal, including name, mailing address, phone number, and email address of the applicant and other named researchers in the proposal may be collected, processed, stored and otherwise used by Facebook for the purposes of administering Facebook’s RFP website, evaluating the contents of the proposal, and as otherwise provided under Facebook’s Privacy Policy.

  • Neither Facebook nor the applicant is obligated to enter into a business transaction as a result of the proposal submission. Facebook is under no obligation to review or consider the proposal.

  • Feedback provided in a proposal regarding Facebook products or services will not be treated as confidential or protected by copyright, and Facebook is free to use such feedback on an unrestricted basis with no compensation to the applicant. The submission of a proposal will not result in the transfer of ownership of any IP rights.

  • Applicants represent and warrant that they have authority to submit a proposal in connection with a Facebook RFP and to grant the rights set forth herein on behalf of their organization. All funding provided by Facebook in connection with this RFP shall be used only in accordance with applicable laws and shall not be used in any way, directly or indirectly, to facilitate any act that would constitute bribery or an illegal kickback, an illegal campaign contribution, or would otherwise violate any applicable anti-corruption or political activities law.

  • Funding for RFP proposals selected by Facebook will be provided pursuant to a sponsored research agreement between Facebook and the applicant’s organization. Applicants understand and acknowledge that their organization will need to agree to the terms and conditions of such sponsored research agreement to receive funding for the RFP proposal.

To apply

Please submit proposal materials to utl-aiproposals@fb.com

Deadline for submissions are on March 15, 11:59pm PT.