1 Introduction
2 State of the art and state of research
2.1 The challenge of transferring expert knowledge into models
2.2 High and low level configuration systems
2.3 Previous approaches for the creation and validation of systems for product and process configuration
Generate inference models (1) | Validate input data (2) | Validate resulting model (3) | PVerify resulting model (4) | Predict BOM elements (5) | Predict BOM structure (6) | Predict Process elements (7) | Process structure (8) | Knowledge in form of rules is implementable(9) | Existing documents can be used (10) | Expert knowledge difficult to transfer or tacit (11) | Interpretable model (12) | ||
---|---|---|---|---|---|---|---|---|---|---|---|---|---|
Rule based approaches | [28] | ◐ | ○ | ○ | ○ | ◐ | ◐ | ◐ | ● | ◐ | ○ | ○ | ◐ |
[29] | ◐ | ○ | ○ | ○ | ◐ | ◐ | ◐ | ● | ◐ | ○ | ○ | ◐ | |
[27] | ◐ | ○ | ○ | ○ | ◐ | ◐ | ◐ | ● | ◐ | ○ | ○ | ◐ | |
[30] | ◐ | ○ | ○ | ○ | ◐ | ◐ | ◐ | ● | ◐ | ○ | ○ | ◐ | |
[23] | ◐ | ○ | ○ | ○ | ○ | ○ | ● | ● | ◐ | ○ | ○ | ● | |
[24] | ○ | ○ | ○ | ○ | ○ | ○ | ● | ● | ◐ | ○ | ○ | ● | |
[25] | ○ | ○ | ○ | ○ | ○ | ○ | ● | ● | ● | ○ | ○ | ● | |
[48] | ○ | ○ | ○ | ● | ○ | ○ | ● | ● | ○ | ○ | ○ | ○ | |
[47] | ○ | ○ | ○ | ● | ● | ● | ○ | ○ | ○ | ○ | ○ | ○ | |
[49] | ○ | ○ | ○ | ● | ● | ● | ○ | ○ | ○ | ○ | ○ | ○ | |
Data based approaches | [34] | ● | ○ | ○ | ○ | ○ | ○ | ● | ● | ○ | ● | ○ | ○ |
[35] | ● | ○ | ○ | ○ | ○ | ○ | ● | ● | ○ | ● | ○ | ◐ | |
[32] | ● | ○ | ○ | ○ | ○ | ○ | ● | ● | ○ | ● | ○ | ○ | |
[31] | ● | ○ | ○ | ○ | ○ | ○ | ● | ● | ○ | ● | ○ | ○ | |
[33] | ● | ○ | ○ | ○ | ○ | ○ | ● | ● | ○ | ● | ○ | ○ | |
[38] | ◐ | ◐ | ○ | ○ | ◐ | ◐ | ● | ● | ○ | ● | ○ | ● | |
[39] | ◐ | ○ | ○ | ○ | ◐ | ◐ | ● | ● | ○ | ● | ○ | ● | |
[37] | ◐ | ○ | ○ | ○ | ◐ | ◐ | ● | ● | ○ | ● | ○ | ● | |
[41] | ○ | ○ | ○ | ○ | ○ | ○ | ● | ○ | ○ | ● | ○ | ● | |
[40] | ○ | ○ | ○ | ○ | ○ | ○ | ● | ● | ○ | ● | ○ | ● | |
[43] | ○ | ○ | ○ | ○ | ○ | ○ | ● | ● | ○ | ◐ | ○ | ● | |
[42] | ● | ○ | ○ | ○ | ○ | ○ | ● | ● | ○ | ● | ○ | ● | |
[44] | ○ | ○ | ○ | ○ | ○ | ○ | ● | ● | ○ | ◐ | ○ | ● | |
[45] | ○ | ○ | ○ | ○ | ○ | ○ | ● | ● | ○ | ◐ | ○ | ● | |
[21] | ◐ | ○ | ○ | ○ | ● | ● | ○ | ○ | ○ | ◐ | ○ | ● | |
[22] | ◐ | ○ | ○ | ○ | ● | ● | ○ | ○ | ○ | ◐ | ○ | ● | |
[20] | ◐ | ○ | ○ | ○ | ● | ● | ○ | ○ | ○ | ◐ | ○ | ● | |
[19] | ◐ | ○ | ○ | ○ | ● | ● | ○ | ○ | ○ | ◐ | ○ | ● |
-
− For the creation of systems for product and process configuration that consider knowledge in form of rules, hardly transferable or tacit expert knowledge and existing documents, or
-
− For the systematic validation of such configuration systems in the sense described above as well as the validation of the data used to create such configuration systems.
3 Approach
3.1 Systematic selection of variants for the generation of the LLCS (module 1)
Active learning method | Query the sample… | Applicability |
---|---|---|
Uncertainty sampling | … for which the model has the highest uncertainty | Not applicable |
Query-by-committee | … for which two or more separately trained models disagree the most | Applicable |
Expected model change | … which leads to the biggest expected change in the model | Applicable |
Variance reduction/Expected error reduction | … which is expected to lead to highest variance or expected error reduction of the model | Not applicable |
Density weighted | … from the region with the highest density | Applicable |
Greedy sampling | … furthest away from all previously queried samples | Applicable |
3.2 Learning of super BOMs (module 2)
3.3 Learning super routings (module 3)
3.4 Learning of dependencies (module 4)
Row | x1 | x2 = 0 | x2 = 1 | x2 = 2 | x2 = 3 | x3 | … | x7 | yc1 | yc2 | … | yc33 | yo1 | … | yo4 | ys1 |
---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
1 | 0 | 0 | 1 | 0 | 0 | 0 | … | 0 | 1 | 1 | … | 0 | 1 | … | 1 | 0 |
2 | 0 | 0 | 1 | 0 | 0 | 0 | … | 1 | 1 | 1 | … | 0 | 1 | … | 1 | 0 |
… | … | … | … | … | … | … | … | … | … | … | … | … | … | … | … | … |
Supervised learning method | Short explanation | Applicability |
---|---|---|
Associative classification [54] | Considers how often certain items (here from x and y) occur together and how specific this correlation is | Applicable. Generates Boolean expressions |
Quine-McCluskey classification [53] | Learns the simplest Boolean expression that explains the known labels | Applicable. Generates Boolean expressions |
Greedy Quine-McCluskey classification [53] | Same approach as above but heuristic and therefore faster | Applicable. Generates Boolean expressions |
Decision tree classification | Learns a decision tree that explains the known labels | Applicable. Decision trees can be converted to Boolean expressions |
3.5 Checking the input data (module 5)
3.6 Checking the LLCS (module 6)
x1 | ¬x1 | (x2 = 0) | (x2 = 1) | (x2 = 2) | (x2 = 3) | ¬(x2 = 3) | x1 ∧ x4 | |
---|---|---|---|---|---|---|---|---|
yc1 | 0 | 0 | 0 | 0 | 0 | 0 | 0 | 1 |
yc2 | 0 | 1 | 0 | 0 | 0 | 0 | 0 | 0 |
yc3 | … | … | … | … | … | … | … | … |
yc4 | 0 | 0 | 0 | 0 | 0 | 0 | 0 | 0 |
yc5 | 0 | 0 | 0 | 0 | 0 | 1 | 0 | 0 |
yc6 | 0 | 1 | 0 | 0 | 0 | 0 | 0 | 0 |
yc7 | 0 | 0 | 0 | 0 | 0 | 0 | 0 | 0 |
… | … | … | … | … | … | … | … | … |
yc38 | 0 | 0 | 0 | 0 | 0 | 1 | 0 | 0 |
x1 | ¬x1 | (x2 = 0) | ¬(x2 = 0) | (x2 = 1) | ¬(x2 = 1) | (x2 = 2) | ¬(x2 = 2) | (x2 = 3) | ¬(x2 = 3) | x3 | ¬x3 | x4 | ¬x4 | x5 | ¬x5 | x6 | ¬x6 | x7 | ¬x7 | |
---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
x1 | 1 | 0 | 0 | 0 | 0 | 0 | 0 | 0 | 0 | 0 | 0 | 0 | 0 | 0 | 0 | 0 | 0 | 0 | 0 | 0 |
¬x1 | 0 | 1 | 0 | 0 | 0 | 0 | 0 | 0 | 0 | 0 | 0 | 0 | 0 | 0 | 0 | 0 | 0 | 0 | 0 | 0 |
(x2 = 0) | 0 | 0 | 1 | 0 | 0 | 0 | 0 | 0 | 0 | 0 | 0 | 0 | 0 | 0 | 0 | 0 | 0 | 0 | 0 | 0 |
(x2 = 1) | 0 | 0 | 0 | 0 | 1 | 0 | 0 | 0 | 0 | 0 | 0 | 0 | 0 | 0 | 0 | 0 | 0 | 0 | 0 | 0 |
(x2 = 2) | 0 | 0 | 0 | 0 | 0 | 0 | 1 | 0 | 0 | 0 | 0 | 0 | 0 | 0 | 0 | 0 | 0 | 0 | 0 | 0 |
(x2 = 3) | 0 | 0 | 0 | 0 | 0 | 0 | 0 | 0 | 1 | 0 | 0 | 0 | 0 | 0 | 0 | 0 | 0 | 0 | 0 | 0 |
¬(x2 = 3) | 0 | 0 | 0 | 0 | 0 | 0 | 0 | 0 | 0 | 1 | 0 | 0 | 0 | 0 | 0 | 0 | 0 | 0 | 0 | 0 |
x 1 ∧ x 4 | 1 | 0 | 0 | 0 | 0 | 0 | 0 | 0 | 0 | 0 | 0 | 0 | 1 | 0 | 0 | 0 | 0 | 0 | 0 | 0 |
4 Assessment of the required effort for applying the approach to general use cases
k | ||||
---|---|---|---|---|
2 (%) | 3 (%) | 4 (%) | ||
n | 5 | 100.0 | 97.9 | 75.0 |
6 | 99.3 | 97.2 | 98.2 | |
7 | 100.0 | 98.9 | 97.7 | |
8 | 99.7 | 98.6 | 98.0 | |
9 | 99.7 | 98.0 | 97.9 |