Dashboard
Live overview of narrative trends, channel activity metrics, and disinformation propagation patterns detected by the Vector system.
Narrative Volume Over Time
Interactive chart coming soon
Active Channels
Channel monitoring stats
Top Narratives
Narrative cluster rankings
Actor Network
Entity relationship graph
Sentiment Trends
Sentiment distribution over time
The live dashboard is under development. In the meantime, analytical outputs are available in the Analysis section, and the underlying data can be accessed via the Dataset page.
Technical Decisions
A transparent log of architectural and technical decisions made while building the Vector system.
Open Data Model for Research Access Deciding how to share Vector's collected data and analysis outputs with the research community while respecting privacy and responsible use. Read decision HDBSCAN for Narrative Clustering After extracting AAT frames from messages, we need to group similar frames into narrative clusters to identify recurring disinformation themes. Read decision Actor-Action-Target Frame Extraction Model We needed a structured representation for narrative content that could be extracted automatically from unstructured Telegram messages. Read decision Snowball Sampling for Channel Discovery We needed a systematic method to discover and select which Telegram channels to monitor, starting from a limited set of known disinformation sources. Read decision PostgreSQL for Primary Storage Choosing a database for storing raw messages, extracted frames, and channel metadata. The system needs to handle millions of records with complex queries for narrative analysis. Read decision Telegram as Primary Data Source Vector needed to choose an initial platform for narrative monitoring. Options included Twitter/X, Facebook, Telegram, and VKontakte. Read decision