Creating a medical dictionary using word alignment: The influence of sources and resources

Table 1 Partition characteristics

Partition	Content	Word correlation	Character correlation	Word ratio difference	Rubrics	English rubric average number (standard deviation) of words	Swedish rubric average number (standard deviation) of words	English unique words	Swedish unique words	English unique words per rubric	Swedish unique words per rubric
All	All terminology systems	0.78	0.79		38,575	3.7 (2.9)	3.3 (3.0)	17,679	25,848	0.5	0.7
1	MeSH, one word in either English or Swedish rubric			0.56	13,514	1.5 (0.7)	1.0 (0.1)	11,267	13,581	0.8	1.0
2	MeSH, more than one word in both English and Swedish rubrics	0.52	0.71	0.30	5,568	2.6 (0.8)	2.3 (0.7)	5,434	6,443	1.0	1.2
3	ICF, whole	0.69	0.79	0.53	1,496	4.7 (2.5)	4.2 (2.8)	991	1,263	0.7	0.8
4	KSH97-P, whole	0.70	0.67	0.49	968	4.0 (2.5)	3.5 (2.4)	1,324	1,382	1.4	1.4
5	ICD-10, except chapter 2 level 4	0.77	0.75	0.37	10,791	5.2 (3.0)	5.2 (3.4)	5,144	7,219	0.5	0.7
6	NCSP, except chapter N	0.64	0.63	0.38	4,137	5.8 (2.7)	5.0 (2.5)	1,758	2,347	0.4	0.6
7	ICD-10, chapter 2 level 4	0.38	0.45	0.71	713	3.6 (2.2)	6.3 (2.7)	443	535	0.6	0.8
8	NCSP, chapter N	0.55	0.48	0.25	1,388	9.4 (2.6)	7.7 (2.3)	249	285	0.2	0.2

Content of the partitions.
Kendall's tau-b correlation between the English rubrics and corresponding Swedish rubrics according to number of words and number of characters and average absolute differences between the ratio for all rubrics in the partition and the grand mean of the different terminology partitions.
Number of parallel rubrics, average number and standard deviation of words per rubrics, number of unique words, and number of average unique words per rubric of the different terminology partitions.

ISSN: 1472-6947