We will rename the each stage to reflect the functionality of these stages in the pipeline. For example,
-
Processing
stays -
Construction
$\leftarrow$ Embedding
-
Filtering
stays -
Labelling
$\leftarrow$ GNN
-
Segmenting
stays
Or, one rename them as follows
-
data_processing
$\leftarrow$ Processing
-
edge_construction
orgraph_contruction
$\leftarrow$ Embedding
-
edge_filtering
$\leftarrow$ Filtering
-
edge_labelling
$\leftarrow$ GNN
-
graph_segmenting
$\leftarrow$ Segmenting
The output directories can be fixed in config files to reflect the name of each stages:
-
data_processing
orfeature_store
-
edge_construction
orgraph_contruction
$\leftarrow$ embedding_processed
-
edge_filtering
$\leftarrow$ filter_processed
-
edge_labelling
$\leftarrow$ gnn_processed
-
graph_segmenting
$\leftarrow$ segments_processed
NOTE: edge labelling or graph labelling stage is infact edge classification stage of the pipeline.