Table 1.

GENETAG corpus statistics The 20K sentences were split into four subsets called Train, Test, Round1 and Round2.

	Train	Test	Round1	Round 2	Total
Number of Sentences	7,500	2,500	5,000	5,000	20,000
Number of Words	204,195	68,043	137,586	137,977	547,801
Number of Tagged Genes = G	8,935	2,987	5,949	6,125	23,996
Total Number of Alternative Forms of Gene Names in G	6,583	2,158	4,275	4,505	17,531
Number of Gene Names in G with Alternative Forms = N	4,675	1,522	3,057	3,186	12,440
Average Number of Alternatives per Gene Name in N	1.66	1.67	1.62	1.65	1.65