Rumor-related social conversations evolve over time, making thread sequencing and text preparation relevant research questions.
THE APPROACH
How it comes together.
Collection helpers and preprocessing prepare tweet-conversation inputs. The experiment constructs time-interval sequences and an LSTM–pooling–LSTM model, while Django routes and templates provide an interface around the research snapshot.
UNDER THE HOOD
Follow the workflow.
Trace the components. Select a node for its role in the system.
Historical research implementation overview6 COMPONENTS
INTERACTIVE 3DRumor Detection Research
PythonDjangoKeras
Rumor Detection Research / workflow
Move to tilt · Drag to rotate · Arrow keysDrag to rotate · Tap a component6 components · 5 connections
2D diagram
Trace connections 5 relationships
Tweet threads and labels→Text preprocessing
Text preprocessing→Time-interval features
Time-interval features→Keras sequence experiment
Keras sequence experiment→Research classification output
Research classification output→Django interfaceinterface integration requires revalidation
+
Select a component to see how it fits into the system.
CAPABILITIES
What’s inside.
01
Tweet collection helpers and text cleaning for links, handles, symbols, and emoji
02
Time-interval sequence construction for conversation data