Name-Disambiguation

Semantic features learning

• step 1: Generate all text file
• step 2: Train word2vec model
• step 4: Sample paper triples
• step 5: Train FCN by triple loss
• step 6: Output semantic features

Non-semantic(Relational) features learning

• step 1: Generate paperID-authors/paperID-organizations/paperID-venues pairs file ordered by name
• step 2: Generate heterogeneous information network ordered by name
• step 3: Train HeGAN ordered by name
• step 4: Output relational features trained in generator and discriminator

Cluster

• step 1: Calculate similarity matrices
• step 2: Run DBSCAN to cluster
• step 3: Assign lonelymountains to cluster to form a cluster itself