Table 6 The importance analysis of subgraphs.

From: Heterogeneous graph convolutional network for rumor detection with multi-level interactive fusion and graph reconstruction

Twitter15 dataset

Method

Acc

NR

FR

TR

UR

\({F}_{1}\)

\({F}_{1}\)

\({F}_{1}\)

\({F}_{1}\)

Only global (attention)

0.911

0.952

0.917

0.907

0.867

Only adaptive gate

0.875

0.912

0.859

0.893

0.836

Our method

0.920

0.958

0.911

0.920

0.889

Twitter16 dataset

Method

Acc

NR

FR

TR

UR

\({F}_{1}\)

\({F}_{1}\)

\({F}_{1}\)

\({F}_{1}\)

Only global (attention)

0.908

0.899

0.901

0.957

0.872

Only adaptive gate

0.875

0.854

0.835

0.905

0.903

Our method

0.929

0.946

0.911

0.926

0.933

Back to article page