Nowadays, it is possible to "report" and share information from almost anywhere in the world. This is regularly done before professional journalists arrive at the scene of an event (e.g. a political demonstration or a natural disaster). As a consequence, journalists increasingly turn to Social Media to find both news and background information. Verifying content in easy, transparent and relatively fast ways is becoming more and more relevant, especially when taking into consideration (a) the sheer quantity of content found in Social Networks, and (b) the fact that a lot of content consists of hoaxes, rumours or deliberately misleading information (e.g. propaganda, spin).
The REVEAL project focuses on verification technologies, tools and strategies. It aims to develop tools, components and strategies that aid journalists in identifying, assessing and verifying user-generated content (UGC) on Social Networks.
Journalism: The challenge for journalists and independent media is to report on developments and occurrences across entire countries and analyze the situation. Often it is difficult to separate truth from lies, propaganda from facts. Especially without or limited direct access to the region, journalists have to rely on other accessible sources, with much content coming from Social Media.
Enterprise: Software forums are the main place for end users, experts, developers, students, scientists to share their knowledge and get useful information from others. Metrics such as community members’ replies, ratings and comments can be utilised to measure modalities such contributor reputation, history, popularity or influence.
Geoparsing, Geosemantics, Trust and Credibility Modelling - Highly scalable real-time geoparsing of regions, streets and buildings running on an Apache Storm cluster. Natural language processing to extract claims (e.g. fake or genuine images), source attribution (e.g. via BBC News) and geosemantic context useful for cross-checking facts. Knowledge-based trust and credibility model to cross-check evidence with known facts about an event for semi-automated verification.
Decision Support System and Visualization - Situation assessment framework for aggregating and visualizing large volumes of social media content. Multi-dimensional interactive visualizations allowing data to be clustered and sub-clusters so analysts can quickly get to evidence relevant to verification of breaking news.
Social media crawling - Social media crawler platform (Twitter, You Tube, Instagram, Facebook etc.) able to run 24/7 on many parallel news stories.
Journalist Decision Support System - JDSS.
Scalable Python Geoparse Library Released - geoparsepy.
Bias in Linked Open Data - BLinD.
Geoparsing and Real-time Social Media Analytics, ESRC Seminar, Oct 2016
REVEAL Project overview - trust and credibility analysis, RDSM-2015 Workshop @ WWW-2015
Journalist Decision Support System (JDSS), REVEAL R&D results, Jan 2017
Scalable Python Geoparse Library Released - geoparsepy, REVEAL R&D results, Dec 2016
Geoparse Benchmark Open Dataset from the University of Southampton IT Innovation Centre, REVEAL R&D results, Jan 2016
Observational study - how do journalists really verify user generated content, REVEAL R&D results, Aug 2016
Verification of Eyewitness Media - An Observational Study, REVEAL Interview, Aug 2016
Geoparsing and Geosemantics for Social Media: Spatio-Temporal Grounding of Content Propagating Rumours to support Trust and Veracity Analysis during Breaking News, ACM Transactions on Information Systems (TOIS) 34, 3, Article 16 (April 2016)
Real-time Crisis Mapping of Natural Disasters using Social Media, IEEE Intelligent Systems, 2014
REVEAL has received EC research funding.