Automated factchecking in the era of fake news
summary
Full Fact describes a working automated fact‑checking engine that scans Hansard, online papers and TV subtitles, matches lines to a database and surfaces prior checks and new claims in real time. The team of ~12 runs code in Python and Django, uses Apache Solr and NLP (Stanford CoreNLP) and reports concrete hits like 142 sightings of a single claim. Demos include live subtitle matching, a Trends dashboard for repeaters and Facebook Live integration; accuracy and false positives remain work in progress.
AI-written summary · about
chapters (11)
- 0:00Post-truth era and fake news
- 3:18Term fake news is misleading
- 5:06Full Fact’s mission since 2010
- 9:00Complexity of manual fact-checking
- 16:33Why automating is hard
- 23:41Building automated engine
- 27:38Trends for tracking misinformation
- 32:12Live fact-checking tools
- 36:30Real-time on Facebook Live
- 38:16Future plans for platform
- 40:02Q&A
related talks
31mThe Fast and the Curious: Harnessing AI in Financial Data Extraction at Scale
30mBuilding Reliable AI Agents for Publishing: A DSPy-Based Quality Assurance Framework
45mZero to Shipped in 30 mins with AI Coding Assistants
17mNatural Language Processing in Automation
08mNews Ninja: Gamified Annotation of Linguistic Bias in Online News
42mBeyond spell checking - what else can we check automatically?
20mKnowledge Scout: From Talks and Slides to Smart Summaries
46mMagic of Automation and Everyday Chores