Fake Information & Hatred Language Detection

How Sigma’s AI platform can help monitor online sources and detect fake news and hatred language.

The Challenge

The client wanted to have statistics on hate speech and fake news on social media and in the news respectively. They also wanted to classify hate speech and fake news to be able to detect trends and topics.

Sigma’s AI Natural Language Processing and Understanding platform can monitor on line sources of information and social media to, among other functions, detect fake news and hate speech.

The Solution

Sigma configured its crawlers to access the sources of information requested by the client as well as reliable sources of information. The collected information was processed by Sigma’s AI Natural Language Processing and Understanding platform. It clustered similar news, found relations between them, extracted information, and compared it to the information from different sources to make a decision on the reliability of the news. It also looked for hate speech using Natural Language Processing and Understanding technology as well as machine learning.

The AI platform was adapted and fine tuned to the specific task by manually annotating a reference corpus. Sigma performed the collection and annotation of the corpus.

The Outcome

The resulting system was able to identify hate speech with 95% accuracy and fake news with over 75% accuracy.
There is a second phase in which more data is being collected and annotated to increase accuracy.