University of Washington - MSIM Capstone 2022
FLIPGRID PLATFORM HEALTH
Empowering Healthy Online Learning
OVERVIEW
The Flipgrid team at Microsoft believes that a healthy online learning platform is a combination of the sense of safety and the level of engagement from its users on the site. Using PowerBI as a vessel for surfacing trends, our mission was to help Flipgrid refine how they measure the health of their platform. In doing so, we want to discover patterns that help them shape their platform policies and feature developments in order to make their community a source of good connections, healthy experiences and a safe environment for all users.
Our Approach
01
Exploratory Data Analysis
The first part of the project was spent exploring literature related to online community health and investigating available data to gather information and create a knowledge base for the team. The initial data analysis helped the team understand the type of data being collected by Flipgrid, thereby informing our query development to measure health and safety on the platform.
03
Build Visualizations in Power BI
Once we derived metrics, and isolated the available data via KQL, the queries we imported into PowerQuery to create visualizations in PowerBI.
02
Designing Logical Queries
With the help of mentoring sessions from our sponsors as well as additional training materials, we spent time developing our knowledge of the Kusto Query Language (KQL) to help us design queries.
04
Consolidate visualizations into a comprehensive dashboard for Flipgrid
Visualizations were organized for the Flipgrid team so that they can quickly access insights derived from our analysis. Our project output was added to an existing analytical dashboard managed by Flipgrid.
Benefits
Our analysis helped measure health on the Flipgrid platform from a couple of different angles. First, we determined the types and quantity of negative content most often impacting the learning communities as well as the users associated with the flagged content. Another outcome of this work produced an offender index in order to classify users based on the types and frequency of offenses being committed. Finally we surfaced moderation trends, which can be used to help standardize the governance model for the internal safety team. The metrics we surfaced in the functional area of platform safety can be combined with others (i.e. content quality, engagement) to calculate relative health index for the platform. This benchmark metric can be used to validate feature implementations and policy updates by Flipgrid moving forward.
PROJECT SPONSOR
Flipgrid, Inc. is an online learning platform that is hosted and managed by the Microsoft Corporation. Flipgrid is categorized as a social learning platform, with the mission to “empower educators as they help students define their voices, share their voices and respect the diverse voices of others”. Utilized by educators from kindergarten through higher education programs as well as for community users of all sorts, the goal in any grid community is to provide a platform that enhances the voice of all members. In traditional classrooms, sometimes not all students are heard. Using Flipgrid, users are able to create custom video responses that can be thoughtfully planned in an environment that allows them to be comfortable. Flipgrid has a critical goal to make sure its user community is happy, healthy and safe.
THE PROBLEM
Like any online community where users are encouraged to share content (images, comments, videos, external links, etc.), there are opportunities for negative or explicit content to emerge. This type of negative content can detract from the good experiences of others. To mitigate this non-normative behavior and protect the integrity of their communities, Flipgrid employs moderation resources in order to keep the site as safe as possible. Flipgrid's user community is growing fast with about one in three U.S. teachers having utilized Flipgrid within their classrooms for any number of the hundreds of use cases it can support. The health and safety team wanted to be more proactive about how they measure platform health to protect communities and cater the best user experience possible.
OUR APPROACH
01
Exploratory Data Analysis
The first part of the project was spent exploring literature related to online community health and investigating available data to gather information and create a knowledge base for the team. The initial data analysis helped the team understand the type of data being collected by Flipgrid, thereby informing our query development to measure health and safety on the platform.
02
Designing Logical Queries
With the help of mentoring sessions from our sponsors as well as additional training materials, we spent time developing our knowledge of the Kusto Query Language (KQL) to help us design queries.
03
Build Visualizations in Power BI
Once we derived metrics, and isolated the available data via KQL, the queries we imported into PowerQuery to create visualizations in PowerBI.
04
Consolidate visualizations into a comprehensive dashboard for Flipgrid
Visualizations were organized for the Flipgrid team so that they can quickly access insights derived from our analysis. Our project output was added to an existing analytical dashboard managed by Flipgrid.
BENEFITS
Our analysis helped refine how health is measured on the Flipgrid platform. First, we investigated the negative content most often impacting the communities as well as the users associated with the flagged content. Another outcome of this work produced an offender index to classify users based on the types and frequency of offenses being committed. The metrics we surfaced in the functional area of platform safety can be combined with others (i.e. content quality, engagement) to calculate the relative health of the platform. This benchmark metric can be used to validate feature implementations and policy updates by Flipgrid moving forward.
TEAM SaaS
TEAM SUCCESS AS A SERVICE
The Capstone team consists of graduate students at the University of Washington Information School and part of the Masters of Science of Information Management program.
Charles Frahm
Project Manager
Shruti Dixit
BI Engineer
Amiya Renavikar
Tech Lead
Shivang Dalal
BI Product Manager