Decision Logic Verification in DMN Decision Tables
The Decision Model and Notation (DMN) is an OMG standard for the representation of company decision logic. This tool can be used for the decision logic level verification of DMN tables. In this context, Smit et al. (2017) have recently proposed the so-called business rule management (BRM) capability framework. This framework identifies specific decision logic level verification capabilities, derived from qualitative re-search with industrial partners. Thus, the framework presented by those authors presents a comprehensive set of verification capabilities actually needed in practice. These capabilities are as follows:
- Identical rule verification. Detecting rules which have an identical input, i.e. are redundant.
- Equivalent rule verification. Detecting rules which are not identical, but still semantically equivalent. Here, our tool can verify if there exist multiple rules which use synonyms as inputs and are therefore equivalent, based on synonym relations via Wordnet.
- Subsumed rule verification. Detecting individual rules which are subsumed by other rules, i.e. they are not necessary. For example, rules containing wildcards often render more specific rules unnecessary due to subsumption.–
- Interdeterminism verification. Detecting rules which will always be activated together, but have differing or contradicting conclusions. For example, rules which will be activated together must not yield that a customer is both credit worthy, and not creditworthy, as this is logically inconsistent.
- Partial reduction verification. Checking wether ranges can be combined to simplify decision tables.
- Overlapping condition verification. Detecting whether there are any overlaps in rule conditions.
- Missing rule verification. Detecting whether there are any missing business rules, i.e. if there are rules missing for expected inputs.
Our tool implements the verification capabilities proposed by Smit et al. (2017). Please note that we did not implement "unnecessary fact verification" as this is geared towards analyzing case-dependent facts and is beyond the scope of this project. In the following, we present examples for the individual verification capabilities.
Identical rules verification
In this table, identical rules can be identified, e.g. rules 1 and 3, or rules 4 and 5.
Equivalent rules verification
In this table, there are equivalent rules. This can result e.g. from different modellers with different understandings or terminology for the same domain of interest. For example, "bill" and "invoice" are synonyms, thus the rules are equivalent due to the same implied semantics. Furthermore, rules 4 and 5, or rules 6 and 7 are equivalent and should be merged.
Subsumed rules verification
In this table, rule 3 is subsumed by rule 4.
In this table, rule 2 subsumes rules 3-7.
Interdeterminism verification
Interdeterminism can occur if multiple rules with different or contradictory conclusions could be activated together. In the example, there are overlapping or subsumed rule conditions (i.e. an input of 7 would trigger all 3 marked rules). Yet, the rules have different conclusions, thus it is unclear what should be inferred. This is cases such as in the example, where the outputs are logically inconsistent (the output for creditworthiness is true and false at the same time, so no inference can be made)
Partial reduction verification
The highlighted rules could be combined into a new range (10-30), which would simplify the DMN decision table.
Overlapping condition verification
In the example, the conditions of rules 2 and 3 are overlapping, i.e. an input of 16 would trigger both rules. This can lead to highly confusing rule tables, as it can not intuitively be seen which rules will be activated together. C.f. also a recent work on this topic by Batoulis and Weske (2018).
Missing rules verification
In the example, there is no rule defined for the input of x<2, or x=5. This could yield limitations to decision-making, if such input is encountered.
References
- Batoulis, K., & Weske, M. (2018, July). Disambiguation of DMN decision tables. In International Conference on Business Information Systems (pp. 236-249). Springer, Cham.
- Smit, K., & Zoet, M. (2018). Identifying Challenges in Business Rules Management Implementations Regarding the Governance Capability at Governmental Institutions.