Table 1.

The table summarizes the number of sentences in the training set, which was detected by ‘Sentence Gene Matcher’ as relevant to a gene and also annotated to have a gene function

Passage type	With gene function	Total	%
front	26	67	39
title_2	149	797	19
abstract	225	1253	18
paragraph	1700	20 703	8
fig_title_caption	17	412	4
fig_caption	99	6009	2
table_title_caption	0	47	0
title_1, title_3, title_4	0	26	0

The different passage types are ‘front’ for the title of the article, ‘title_1’ refers to section headings like ‘Introduction’, ‘title_2’ is the section subheadings that sometimes describes the specific topic/finding of the section, ‘title_3’ and ‘title_4’ are more deeper levels of section headings, ‘abstract’ is the abstract content, ‘fig_title_caption’ is the title of a figure caption and ‘fig_caption’ is the caption of the figure, ‘table_title_caption’ is the caption of a table.