Calculate the Similarity Between Data

ASPL User's Guide v 1.00
© 2024 by SetSphere.COM

9-1

Similarity Operations

Similarity is a cosine based calculation of the angle between vectors in a plane. Setting aside its implementation, ASPL provides the similarity or its abbreviation sim as a command to calculate the similarity between your datasets. In this example, we will load the sample workspace RANDONEBIT and calculate the similarity between the randomly generated groups in the workspace:

# aspl -wsname RANDONEBIT -groupingclass BITGROUP

aspl> sim v1 v2 v3
display the similarity between v1 v2 and v3

aspl> simi123 = sim v1 v2 v3
assign the similarity calculatiion to COS variable simi123

aspl> c
display the COS symbol table

aspl> ?c simi123
interrogate the similarity variable simi123

aspl> ??c simi123
really interrogate simi123

aspl> c sorted
display COS symbol table in sorted mode

aspl> v sorted
display the symbol table in sorted mode

aspl> @c simi123
display the archived COS variables for simi123

aspl> @@c simi123
display the dependencies for COS variables simi123

aspl> quit
leave ASPL without saving the workspace RANDONEBIT

A similarity variable is interrogated by preceding it with the operator ?c

To really interrogate a similarity COS variable by sourcing the variables it depends on, precede the COS variable with the operator ??c and
this will cause sourcing the dependendies variables and may also result in updates in the sourced variables.

To view the archived similarities of a COS variable, precede the variable name with the operator @c

To view the variables that a COS variable depend on, precede the COS variable name with the operator @@c

ASPL set variables are stored in datasets symbol table (or simply called variables-symbol-table), while ASPL similarity variables are stored in COS-symbol-table. To display the variables-symbol-table use the command v or v sorted and to view the COS-symbol-table use the command c or c sorted

Remember that ASPL stores the similarity variables in its COS symbol table that you can display its contents anytime by typing any of the following commands at the ASPL prompt: c C cos COS

● Tickable Predicates for similarity

The similarity is tickable with predicates that are shown in the following table:

a yellow note

**-T- Tab. 9.1.1 [Tickable Predicates for similarity][Tickable Predicates for similarity]**
similarity`_	similarity on the aggregated element containment names
similarity`ffl	similarity on the aggregated element containment names
similarity`f	similarity on the aggregated element names
similarity`c	similarity on the aggregated checksums of the named elements
similarity`z	similarity on the aggregated zsums of the named elements
similarity`fflz	similarity on the aggregated element containment names AND their zsums
similarity`dosi	similarity on the aggregated document similarities of the named elements
similarity`h	similarity on the aggregated entropies of the named elements
similarity`c	similarity on the aggregated checksums of the named elements
similarity`fc	similarity on the aggregated named elements AND their checksums
similarity`fflc	similarity on the aggregated element containment names AND their checksums
similarity`fflmtm	similarity on the aggregated element containment names AND their make time
similarity`mtm	similarity on the aggregated make times of the named elements
similarity`fmtm	similarity on the aggregated named elements AND their make times
similarity`d	similarity on the aggregated containment subgroups
similarity`dmtm	similarity on the aggregated containment subgroups AND the make times of the subgroups