1.Data Collection &Preprocessing
Data Cleaning
Normalization
Time
-
series
2.Feature Extraction
Spatial Features(CNN)
Temporal Features(LSTM)
Textual Features(BERT)
3.Multi
-
modal Feature
Fusion(Proposed
Method:Knowledge
-
Guided
Attentive Fusion)
Corss
-
attention Mechanism
External Knowledge
Integration
4.Model Training & Optimization
Initial Model
Backpropagation
Loss Function
(Cross
-
Entropy
Regularization)
Updated
Parameters
Threshold Met?
5.Final Classification
& Prediction
NO
YES