Infodemic: Taiwan Disinformation Understanding for Pandemic
The Diplomat: Taiwan is at the Frontline of the Disinformation Wars
Since 2018, international reports and studies have shown that Taiwan is facing more and more severe disinformation and media manipulation from the outsider.
In times of Sars-CoV-2 Pandemic
Disinformation spring up among the Sars-CoV-2 pandemic, causing chaos and panic worldwide. Due to the spread of misleading information, the world has seen societal and economic upheavals, such as the global toilet-paper panic hoarding and the face mask diplomacy polemics in Taiwan.
To help the public discover disinformation in the midst of the global pandemic, Taiwan AI Labs developed the disinformation understanding system, which collects comprehensive information sources and generates a navigation knowledge graph by Natural Language Processing(NLP) to help people understand the formative bias by comprehensiveness. This AI-powered system aims at providing the public a more holistic view of the world. To try the system, visit the Infodemic.cc, a Taiwan real-time media understanding website.
Comprehensive Information Sources
AI Labs monitors public data on 20+ popular news, social and streaming media in Taiwan, including best selling news reports, twitter and youtube to observe how information be manipulated and compare the features of information and users on different media.
Natural Language Processing Powered Solution
AI Labs involves AI technology to human languages understanding and information communication analysis, translates complex information phenomena into understandable conclusions. This solution detects suspicious information and sources early by learning their features, and helps people, governments and related organizations prepare to take countermeasures.
Information Knowledge Graph
Applying word segmentation, NER(Named Entity Recognition), OIE(Open Information Extraction) , clustering, classifying and news generation model to build the architecture inside every piece of information and between informations.
Knowing the Formative Bias by Comprehensiveness
To identify the informative bias, AI Labs models the quality of news content using three evaluation indicators: credibility, sentiment, and neutrality scores. Islander Satellite also investigates the stances, sentiments, and writing styles for each user on social media. With intensive feature engineering on their comments and social links, Islander is able to explore the deviation and relevance across users as well as trending topics, which is a key to detecting disinformation propagation.
Case 1: TV News Monitoring
AI Labs cooperated with the National Communication Commission(NCC) to build a TV news monitoring system, aiming to maintain the neutrality of TV news. This system monitors 24 channels from 2019/9/9 to 2020/1/5. For more information, try out our Taiwan TV news monitoring system.
Case 2: Social Network Understanding
Is mask donation advantageous to Taiwan’s worldwide Internet volume? The understanding of Twitter would show the answer. Report Link
Case 3: User Understanding
Disinformation might be spread by social network users. Let user understanding reveal those users’ intentions. Report Link
