Table 1.

Eight data sets constructed from our drug-suicide relation (DSR) corpus and their respective properties.

Data set	Split ratio for training and test data sets (training:test)	Categorization of relation classifications	Order of entities
1	90%:10%	None and ADEa	No
2	80%:20%	None and ADE	Yes
3	90%:10%	None and ADE	No
4	80%:20%	None and ADE	Yes
5	90%:10%	None, ADE, suicide means, and treatment	No
6	80%:20%	None, ADE, suicide means, and treatment	Yes
7	90%:10%	None, ADE, suicide means, and treatment	No
8	80%:20%	None, ADE, suicide means, and treatment	Yes

^aADE: adverse drug event.