Machine Learning Service Journalism for Newsrooms (Round 6)

United Kingdom Project type: prototype Tanya Cordrey


Technology can change the economics of service journalism. This project aims to provide news organisations with the ability to build out service journalism at scale through combining machine learning with the knowledge of crowds. There are millions of reviews, left by regular people, who have tried and tested almost every product. The goal is to analyse these reviews for publishers and to identify which products consistently perform the best over time. If successful, this project not only offers improved monetisation in terms of affiliate fees revenue but also the ability to to experiment and scale service journalism which plays such an important role in subsidising hard-to-monetise editorial areas such as investigative journalism.

The solution

Consumers browsing products often search news organisations for accurate reviews before they make a purchase. This in turn produces a reliable stream of affiliate revenues e.g. NYT Wirecutter. Although major news organisations produce some highly regarded and trusted product recommendations they simply cannot produce the range and volume demanded by users because of the prohibitive cost of editorial product research relative to revenues. This project will build components for publishers that use machine learning to assess public user reviews in order to make reliable product recommendations. This can be made free to publishers, giving them additional ways to monetise traffic.