Fig 1.
COS aims to learn MeSH term embeddings based on three data sources: corpus (the green block), ontology (the orange block), and semantic predications (the blue block). The structured relationships between MeSH terms are different across the data sources. The learned MeSH term embeddings should contain the information from all data sources.
Fig 2.
COS firstly generates MeSH term sequences from each data source. It then samples each group of generated sequences to the same number of sequences and merges them into one set of MeSH term sequences. Finally, COS learns the MeSH term embeddings based on the sequences set.
Table 1.
Summary of related work.
Table 2.
Important notation.
Table 3.
The four semantic predications datasets and their statistics.
Table 4.
Edge prediction results for treat.
Table 5.
Edge prediction results for interact.
Table 6.
Edge prediction results for cause.
Table 7.
Edge prediction results for affect.
Table 8.
The edge prediction results of COS using different sampling strategies.
Table 9.
The edge prediction results of COS and COS without using the semantic predications data source.