The Everyday Respect project aims to provide multimodal, multi-perspective machine learning models to study communication between officers and drivers during traffic stops. We will deploy these tools to study patterns in the communication between LAPD officers and drivers, publish these findings in academic journals, and present them to the Los Angeles Board of Police Commissioners.
Project Phases
Project Goal: This research project aims to facilitate transparency and accountability with respect to officer-driver communication during traffic stops. The project is technically focused, developing new machine-learning tools and answering empirical research questions about patterns in officer-driver communication.
This project involves five steps:
Phase 1: Stakeholder-Defined Metrics of “Good” Communication
Engage with community stakeholders to define metrics of “good” communication during motor vehicle stops. This step includes surveys, interviews and focus groups and is critical to develop measures that reflect diverse stakeholder views and capture the dimensions of communication most salient to each of our stakeholder groups.
Phase 2: Multimodal, Multi-Perspective, Machine Learning Models
Incorporate community and officer perspectives to construct machine learning models that automatically rate appropriateness of communications during stops. Critically, in this phase we employ a diverse set of human annotators to view and score communication, reflecting multiple perspectives from members of the community and retired officers.
Phase 3: Analysis of ~30,000 Stops
Apply these models to analyze approximately 30,000 stops conducted by the LAPD.
Phase 4: The Causes of “Good” Communication & (De)escalation
Use the generated data, along with supplementary contextual information about the stops, to establish a scalable, evidence-based framework defining when and how communication dynamics result in an effective and respectful interaction, but also when they fail and result in escalations.
Phase 5: Public release of new tools.
Present analysis back to our community and law enforcement stakeholders regarding patterns we observe in communication during motor vehicle stops. We will also publicly release the machine-learning tools we develop, facilitating learning and accountability across the country.