The following example searches a file describing the HAMLET package for some of its main keywords with the following optional settings:
The output listing is as follows :-
WORD-SEARCHING IS INSENSITIVE TO CASE.
WORD COUNTS ............................................
WORD FREQUENCY % VOCAB. % TEXT CONTEXT UNITS
context* 39 11.02 0.97 31
dimension* 18 5.08 0.45 16
frequenc* 24 6.78 0.60 19
hamlet 37 10.45 0.92 34
joint 27 7.63 0.67 24
MINISSA 12 3.39 0.30 11
scaling 16 4.52 0.40 14
text* 64 18.08 1.60 54
vocabulary 28 7.91 0.70 25
word* 89 25.14 2.22 59
4011 words were read from the text file.
354 of these were in the search list, and
134 context-units were counted.
JOINT FREQUENCIES ......................................
for a FIXED CONTEXT LENGTH of 30 words:
i 1 2 3 4 5 6 7 8 9
+-------------------------------------------------------------------------
context* 1 |
dimension* 2 | 6
frequenc* 3 | 10 7
hamlet 4 | 7 7 9
joint 5 | 11 8 16 10
MINISSA 6 | 4 6 4 7 6
scaling 7 | 2 8 4 7 6 6
text* 8 | 15 5 6 11 10 6 8
vocabulary 9 | 5 1 6 8 5 2 2 11
word* 10 | 14 5 10 13 15 4 6 29 16
STANDARDISED JOINT INDEX VALUES ........................
Jaccard coefficient - ignores joint non-occurrence
i 1 2 3 4 5 6 7 8 9
+-------------------------------------------------------------------------
context* 1 |
dimension* 2 | 0.15
frequenc* 3 | 0.25 0.25
hamlet 4 | 0.12 0.16 0.20
joint 5 | 0.25 0.25 0.59 0.21
MINISSA 6 | 0.11 0.29 0.15 0.18 0.21
scaling 7 | 0.05 0.36 0.14 0.17 0.19 0.32
text* 8 | 0.21 0.08 0.09 0.14 0.15 0.10 0.13
vocabulary 9 | 0.10 0.03 0.16 0.16 0.11 0.06 0.05 0.16
word* 10 | 0.18 0.07 0.15 0.16 0.22 0.06 0.09 0.35 0.24
The standardised matrix is then submitted to Multidimensional Scaling with the following result:
Interpretation of MDS solutions may be assisted by reference to Hierarchical Clustering. Smallest space analysis, on the other hand, is generally claimed to produce more easily interpreted geometric solutions in fewer dimensions than metric procedures like Factor Analysis, as well as being more versatile in detecting ordered structures in the data.The following are the results of applying Correspondence Analysis to the same example:
For further discussion of the procedures used here, see the references.
