Projects: Fighting Fake News

DECaDE researchers are engaged in socio-technical research projects tackling the dissemination of Fake News e.g. on Social Media Platforms.

Visual Content Provenance.

Team: John Collomosse, Tu Bui, Alex Black.

Fake news and misinformation are centuries-old societal problems, exacerbated by the ease with which digital imagery is manipulated and shared. Improving awareness of the origins or ‘provenance’ of content can help users make more informed decisions on whether to trust the media they encounter online.

Emerging standards for image attribution embed provenance information securely within metadata, which is carried around with the media file. However image metadata is commonly stripped when uploaded and shared on social platforms, and may be replaced by bad actors to misattribute an image to a false story. One solution is to lookup provenance information in a shared, public immutable data store by visually matching images via ‘content fingerprinting’; i.e. a compact hash of image or video content.

DECaDE researchers are actively exploring both aspects of this technology. First, novel computer vision ('AI') methods to compute a robust visual hash of media content. Second, the exploration of distributed ledger technology (DLT/Blockchain) to store those hashes immutably without a centralised point of control or trust. This work takes early work exploring a similar idea for digital preservation in Project ARCHANGEL to the next stage, exploring how a combination of AI fingerprinting and DLT can be used to underscore integrity e.g. in photo journalism.

Multimodal strategies of misinformation. The case of COVID vaccines.

Team: Itziar Castello, Andrew Hill, Glenn Parry.

The objective of this project is to understand the characteristics (emotional work and symbolic work) of the leading misinformation discourse on social media (e.g. Twitter). Work will analyse the posts/tweets of an actor or small group involved in using visuals and words to elicit emotions through moral shock, and then transforming those emotions into emotional energy to perpetuate their misinformation discourse.

Researchers will capture historical posts/tweets (including text and images) of an individual or a group of linked actors advocating a distinct misinformation issue. In this work research will investigate the vaccines created for COVID19 and follow the discourse from the leading actor/group wishing to spread fear or misinformation regarding those vaccines. Relatively small amounts of data will be captured to enable analysis through quantitative and qualitative means. Quantitative analysis will identify the centrality of the posts/tweets and how the messaging is amplified. Qualitative analysis will examine the emotional strategies employed within discourse.