Project Respect

Following Love by Numbers the year before, we worked with Google again in 2018 to build Project Respect, for the Sydney Gay and Lesbian Mardi Gras. We asked the LGBTIQ+ community to write positive identity statements about themselves. These statements are then made available as part of an open source data set, to teach machine learning models about how the LGBTIQ+ community speak about themselves.

The web experience consists of a visualisation of statements written by the community in an interactive single page application.

The dataset will be used to positively identify words that the LGBTIQ+ community uses to talk about themselves.

In the submission form, we asked users to list 3 identity labels they identify with. The Perspective API was used to identify the most toxic term. With each submission, the goal was to reduce the toxicity score of that term.

The site was built using the Google App Engine secure scaffold, in order to comply with the security requirements and best practices that Google demanded for hosting on their domain. The visualisations were done with vanilla javascript. Vue.js was used to power the front-end and bring the experience together.