1UARMSOLVER(1) User Commands UARMSOLVER(1)
2
6 uARMSolver – universal Association Rule Mining Solver
7
9 uARMSolver [-v|-?] [-sSETUP_FILE|-s SETUP_FILE]
10
12 The framework is written fully in C++ and runs on all platforms. It
13 allows users to preprocess their data in a transaction database, to
14 make discretization of data, to search for association rules and to
15 guide a presentation/visualization of the best rules found using exter‐
16 nal tools. As opposed to the existing software packages or frameworks,
17 this also supports numerical and real-valued types of attributes be‐
18 sides the categorical ones. Mining the association rules is defined as
19 an optimization and solved using the nature-inspired algorithms that
20 can be incorporated easily. Because the algorithms normally discover a
21 huge amount of association rules, the framework enables a modular in‐
22 clusion of so-called visual guiders for extracting the knowledge hidden
23 in data, and visualize these using external tools.
24
26 -h, -? Show a help message and exit.
27 -sSETUP_FILE, -s SETUP_FILE
29 Path to a setup file (default arm.set). See the FILES section
30 for details.
31
33 This section describes how to describes how to configure a universal
34 ARM Solver (uARMSolver) using a setup file. See also the -s option.
35 The setup file consists of three sections, including:
36 • a problem definition,
38 • parameter setting of a selected algorithm for solving ARM, and
40 • parameter setting of a selected visualization method.
42 Lines starting with the % character are comments and are not checked
44 for syntax.
45 PROBLEM DEFINITION
47 The problem definition begins with the reserved word Problem, then
48 curly brackets enclosing a series of parameter definitions. Each pa‐
49 rameter definition is a line of the form:
50 parameter = value
51 The following parameters are supported:
52 Tdbase_name = file_name
54 path of the transaction database
55 Rule_name = file_name
57 path of an existing archive of mined association rules
58 Out_name = file_name
60 path where the archive of mined association rules will be
61 written
62 Period = integer_value
64 how many periods are captured by archive files
65 The Period parameter determines whether more transaction databases or
67 archive files are processed by the solver simultaneously. When its
68 value is more than one, the solver expects that files are named with
69 extensions according to their sequence numbers, e.g., is denotes as a
70 sequence number (e.g., .1, .2, ..., .k. When the Period is set to one,
71 a single input file representing the transaction database or ARM ar‐
72 chive is processed.
73 ALGORITHM SELECTION
75 A line of the form:
76 Algorithm = mnemonic
77 selects a particular ARM solver algorithm. For example, the Differen‐
78 tial Algorithm has mnemonic DE, Particle Swarm Optimization has mne‐
79 monic PSO, and so on.
80 When the algorithm is NONE, the solver does not perform optimization.
82 Instead, it expects an ARM archive produced by another traditional al‐
83 gorithm (such as Apriori) and focuses on the visualization section of
84 the process.
85 ALGORITHM-SPECIFIC PARAMETERS
87 Next are algorithm-specific parameter blocks. These begin with a line
88 of the form
89 mnemonic_PARAM
90 followed by curly brackets enclosing a series of parameter definitions
91 of the form
92 mnemonic_param = value
93 For example, the Differential Algorithm (DE) supports the following pa‐
94 rameters:
95 DE_NP = integer_value
97 population size of DE algorithm
98 DE_FES = integer_value
100 maximum number of fitness function evaluations
101 DE_RUNS = integer_value
103 maximum number of an independent DE runs
104 DE_F = float_value
106 scaling factor used by DE mutation strategy
107 DE_CR = float_value
109 crossover parameter controlling the DE mutation strategy
110 DE_STRATEGY = integer_value in [1,12]
112 specific DE mutation strategy
113 VISUALIZATION SELECTION
115 In the future, the solver will support a visualization method. A line
116 of the form:
117 Visualization = mnemonic
118 selects a particular method of preparing the data from the ARM archive
119 for visualization.
120 Two algorithms are planned:
122 FLOW River flow
124 METRO Metro map
126 Method NONE will disable visualization preparation.
128 VISUALIZATION PARAMETERS
130 Visualization methods will have method-specific parameter blocks simi‐
131 lar to the algorithm-specific parameter blocks for the ARM solver.
132
134 uARMSolver -s arm.set
135 Novermber 2021 UARMSOLVER(1)