Ontotext Proud of Its Work on the PHEME Project for Computing Veracity

Wednesday, June 28, 2017

pheme logo

Ontotext is pleased to announce that together with our partners, we have completed the work on the EU-funded PHEME project ‘Computing Veracity Across Media, Languages, and Social Networks’ than ran from 1 January, 2014 through 31 March, 2017.

At the end of this project, Ontotext is taking stock of the work we have done in helping to create a computational framework for automatic discovery and verification of information at scale and fast. Named after the goddess of rumours and fame in the Greek mythology, Pheme, the project was aimed at developing a smart way to alert users to rumors and misinformation. PHEME focused on modeling, identifying, and verifying phemes (internet memes with added truthfulness or deception), as they spread across media, languages, and social networks.

As part of the project, Ontotext participated in several work packages, lending its expertise and experience in semantic technologies to help create algorithms for detecting and classifying rumours, and for harnessing the project technologies into practical applications to be used by digital journalists.

The data acquired and enriched by the different components developed in PHEME were included in a Knowledge repository implemented in Ontotext’s semantic graph database GraphDB. The datasets used as the factual knowledge sources came from the Linked Open Data (LOD) cloud: FactForge, DBpedia, OpenCyc, and Linked Life Data.

However, Ontotext’s work on the project was not limited to GraphDB. Using our extensive experience in text analytics, we have also provided concept extraction and enrichment. By interlinking people, organizations and locations from unchecked social-media streams to the rich contextual information about them in our knowledge base, we were able to know more. This makes it easier to recognize whether something could be a rumor or not.

Ontotext also took active part in harnessing the project technologies into practical applications to be used by digital journalists. We and our project partners developed an open-source journalism dashboard prototype that showcased the technologies and aimed to assist journalists with news gathering and verification tasks.

In addition, Ontotext and its project partners developed the fact-checking assistant Hercule. It is a web-based interface that helps journalists to explore news, video and social media relevant to a given topic, to filter and visually analyze them by time, impact, controversiality, factuality and source, and to validate claims against proprietary and open data.

The innovation features of Hercule include extracting, semantically enriching, and indexing structured claims. It also helps associating impact, controversiality and veracity scores as well as providing context for fact-checking by linking to open and government data, and content archives.

By lending our expertise and experience in semantic technologies to the PHEME project, we contributed to the building of technologies for checking veracity that can help organizations scale up content verification, reduce costs and build trust among social media users.

Learn more about the PHEME project and see how GraphDB turns data into knowledge.


For more information, contact Doug Kimball, Chief Marketing Officer at Ontotext