Validation of models and data for compliance with laws
The present disclosure provides computing systems and techniques for validating a decision model against a canon of regulation. A server can deconstruct a decision model into a number of branching decisions and also generate a Markov chain comprising a number of sequences from a canon of regulation. The server can compare the branching decisions to the sequences and can validate the decision model with the canon of regulation based on the comparison.
1. An apparatus, comprising:
a processor; and
a memory coupled to the processor, the memory comprising instructions that when executed by the processor cause the processor to:
receive a decision model;
deconstruct the decision model into a plurality of branching decisions;
receive an information element comprising indications of a canon of regulation, the canon of regulation comprising a plurality of requirements and a plurality of prohibitions;
recognize text and syntax from the canon of regulation;
input the recognized text and syntax to a recursive neural network (RNN) to generate a Markov chain comprising a plurality of sequences, wherein ones of the plurality of sequences of the Markov chain correspond to the plurality of requirements and other ones of the plurality of sequences of the Markov chain correspond to the plurality of prohibitions; and
determine whether the decision model complies with the canon of regulation based on the plurality of branching decisions and the plurality of sequences of the Markov chain.
2. The apparatus of claim 1 , the memory further comprising instructions that when executed by the processor cause the processor to:
determine whether the plurality of branching decisions match ones of the plurality of sequences of the Markov chain; and
determine that the decision model infringes the canon or regulation based on a determination that the plurality of branching decisions do not match ones of the plurality of sequences of the Markov chain.
3. The apparatus of claim 2 , the memory further comprising instructions that when executed by the processor cause the processor to determine, for each of the plurality of branching decisions, whether the branching decision matches at least one of the plurality of sequences of the Markov chain.
4. The apparatus of claim 3 , the memory further comprising instructions that when executed by the processor cause the processor to match the branching decision to at least one of the plurality of sequences of the Markov chain based on a fuzzy logic algorithm.
5. The apparatus of claim 3 , the memory further comprising instructions that when executed by the processor cause the processor to match the branching decision to at least one of the plurality of sequences of the Markov chain based on a neural network.
6. The apparatus of claim 3 , the decision model to generate an output based on a set of input data, the memory further comprising instructions that when executed by the processor cause the processor to:
identify a lineage of the set of input data;
match the lineage of the set of input data to at least one of the plurality of sequences of the Markov chain; and
determine that the decision model infringes the canon or regulation based on a determination that the lineage of the set of input data does not match at least one of the plurality of sequences of the Markov chain.
7. The apparatus of claim 1 , wherein the decision model is a customer engagement model, a targeted advertising model, a business forecasting model, or a loan scoring model.
8. At least one machine-readable storage medium comprising instructions that when executed by a processor at a computing platform, cause the processor to:
receive a decision model;
deconstruct the decision model into a plurality of branching decisions;
receive an information element comprising indications of a canon of regulation, the canon of regulation comprising a plurality of requirements and a plurality of prohibitions;
recognize text and syntax from the canon of regulation;
input the recognized text and syntax to a recursive neural network (RNN) to generate a Markov chain comprising a plurality of sequences, wherein ones of the plurality of sequences of the Markov chain correspond to the plurality of requirements and other ones of the plurality of sequences of the Markov chain correspond to the plurality of prohibitions; and
determine whether the decision model complies with the canon of regulation based on the plurality of branching decisions to the plurality of sequences of the Markov chain.
9. The at least one machine-readable storage medium of claim 8 , comprising instructions that further cause the processor to:
determine whether the plurality of branching decisions match ones of the plurality of sequences of the Markov chain; and
determine that the decision model infringes the canon or regulation based on a determination that the plurality of branching decisions do not match ones of the plurality of sequences of the Markov chain.
10. The at least one machine-readable storage medium of claim 9 , comprising instructions that further cause the processor to determine, for each of the plurality of branching decisions, whether the branching decision matches at least one of the plurality of sequences of the Markov chain.
11. The at least one machine-readable storage medium of claim 10 , comprising instructions that further cause the processor to match the branching decision to at least one of the plurality of sequences of the Markov chain based on a fuzzy logic algorithm or a neural network.
12. The at least one machine-readable storage medium of claim 10 , the decision model to generate an output based on a set of input data, the medium comprising instructions that further cause the processor to:
identify a lineage of the set of input data;
match the lineage of the set of input data to at least one of the plurality of sequences of the Markov chain; and
determine that the decision model infringes the canon or regulation based on a determination that the lineage of the set of input data does not match at least one of the plurality of sequences of the Markov chain.
13. The at least one machine-readable storage medium of claim 8 , wherein the decision model is a customer engagement model, a targeted advertising model, a business forecasting model, or a loan scoring model.
14. A method comprising:
receiving a decision model;
deconstructing the decision model into a plurality of branching decisions;
receiving an information element comprising indications of a canon of regulation, the canon of regulation comprising a plurality of requirements and a plurality of prohibitions;
recognizing text and syntax from the canon of regulation;
inputting the recognized text and syntax to a recursive neural network (RNN) to generate a Markov chain comprising a plurality of sequences, wherein ones of the plurality of sequences of the Markov chain correspond to the plurality of requirements and other ones of the plurality of sequences of the Markov chain correspond to the plurality of prohibitions; and
determining whether the decision model complies with the canon of regulation based on the plurality of branching decisions to the plurality of sequences of the Markov chain.
15. The method of claim 14 , comprising:
determining whether the plurality of branching decisions match ones of the plurality of sequences of the Markov chain; and
determining that the decision model infringes the canon or regulation based on a determination that the plurality of branching decisions do not match ones of the plurality of sequences of the Markov chain.
16. The method of claim 15 , comprising determining, for each of the plurality of branching decisions, whether the branching decision matches at least one of the plurality of sequences of the Markov chain.
17. The method of claim 16 , comprising matching the branching decision to at least one of the plurality of sequences of the Markov chain based on a fuzzy logic algorithm or a neural network.
18. The method of claim 16 , the decision model to generate an output based on a set of input data, the method comprising:
identifying a lineage of the set of input data;
matching the lineage of the set of input data to at least one of the plurality of sequences of the Markov chain; and
determining that the decision model infringes the canon or regulation based on a determination that the lineage of the set of input data does not match at least one of the plurality of sequences of the Markov chain.
19. The method of claim 14 , wherein the decision model is a customer engagement model, a targeted advertising model, a business forecasting model, or a loan scoring model.