Visual stories teamwork

I’ve written, researched and helped develop award-winning visual stories such as this 2023 women in politics article that won the Society of News Design (SND) Politics category, an analysis of 123 years of federal budget speeches.

Channel Nine Hackathon 2023

I presented and helped develop an AI solution for image searching using Open AI CLIP (Contrastive Language–Image Pre-training), built on zero-shot transfer learning.

CLIP can learn visual concepts from natural language supervision and apply better captions making it easier to split content into relevant categories.

This semantic and relational search also works for captioning, keeping metadata consistent, accurate and up to date. It has been used by the New York Times and other corporations. We developed a live prototype called “Supercharge DCX”. 

By using other AI models via transfer learning it could be possible to have suggested images for stories based on keywords in the article. 

By training an AI model on library data we may find photos of famous people in the archive before they were famous such as NSW Police Commissioner Karen Webb who would have been photographed working at highway patrol or in other settings.

2016 Walkley hackathon

The Global Editors Network and The WalkleyFoundation hosted the Editor’s Lab hackathon in March 2016, I also represented Fairfax Media in the 2018 event.

I was among the 14 teams invited to compete, and while our entry didn’t win, it was a great experience. Our entry Comment Sweeper is a tool for journalists and readers to better filter, understand and engage with comments posted to news stories on social media and news websites.

Newscorp may develop our prototype and we were mentioned as one of the outstanding projects at the hack in a few forums. An example from the ABC link below:

From memory, there were a few other projects that stood out to me:

● NewsCorp and IBM teamed up to solve the problem of finding sources for follow up stories by analysing the vast amounts of text in their comments sections and Facebook pages, looking for indicators that the audience member has a personal experience that could enhance their stories.