Information Technology Reference
In-Depth Information
T
ABLE
6.
Statistics of text collections used in experiments.
STATISTICS
SMALL
LARGE
Characters
18,621
1,231,109
Words
2,060
173,145
After stopping
1,200
98,234
Index size
1.31 Kb
109.0 Kb
T
ABLE
6.
Statistics of text collections used in experiments.
Collection
Small
Large
File size (Kb)
18.2
1,202.3
Index size (Kb)
1.3
109.0
Number of words
2,060
173,145
— after stopping
1,200
98,234
Fig. 11.13
Two versions of a table. The
upper
version is poor. Because there is no sense of table
hierarchy—all the elements are at the same level—headings and content must be differentiated by
case. Different units have been used for file sizes in different lines (assuming characters are one byte
each). Units are mentioned explicitly in the last line, and the precision is inconsistent. The heading
of the first column is unnecessary and the table has too many horizontal lines. In the
lower
version
there are no vertical lines. Rows of the same type are now adjacent so that they can be compared by
the reader, and the hierarchy between the total number of words and the number of words excluding
stopwords is visually indicated. Note that the values of different units do not need to be vertically
aligned on the decimal point or presented with the same precision
rows may be partitioned or have internal structure. The hierarchy can be indicated in
several ways: rows or columns can be separated by double lines, single lines, or white
space; headings can span several columns; labels can refer to several rows. Deeper
structure—which is sometimes necessary but is usually unwise—can be indicated
by markup within the table such as embedded headings. (A complex table is shown
in Fig.
11.14
.) The items below a column head should be of the same kind or about
the same thing. Items to the right of a row label should all be properties of that label.
The column of labels does not need to have a heading, but this position, the top-left
corner of the table, should not be a label for the other column headings. If there is
no heading for the column of labels, leave the position blank.
Tables should be open and uncluttered, with ample white space. Don't have too
many horizontal or vertical rules. In particular, there is no need to have a rule between
