Commit conformity verification system
AssigneeSap SE
InventorsRocio Cabrera Lozoya, Antonino Sabetta, Michele Bezzi
Life of this patent
Estimating the term…
Abstract
Systems and methods are provided for training a machine learning model to generate a score indicating a level of discrepancy between a commit message and a corresponding code change. The computing system receives a commit comprising a given commit message and a given corresponding code change and analyzes, using the trained machine learning model, the given commit message and given corresponding code change to generate a score indicating the level of discrepancy between the given commit message and the given corresponding code change of the received commit.
Claims20
A computer-implemented method comprising:
- generating, by one or more processors of a computing system, training data comprising:
- a first set of pairs wherein each pair of the first set of pairs comprises a commit message and a corresponding code change in an existing project;
- and
- a second set of pairs wherein each pair of the second set of pairs comprises a commit message and a randomly selected code change that is different than the corresponding code change for the commit message;
- training, by the one or more processors of the computing system, a machine learning model using the generated training data to generate a trained machine learning model configured to generate a score indicating a level of discrepancy between a commit message and a corresponding code change;
- receiving, by the one or more processors of the computing system, a commit comprising a given commit message and a given corresponding code change;
- and
- analyzing, using the trained machine learning model, the given commit message and given corresponding code change to generate a score indicating the level of discrepancy between the given commit message and the given corresponding code change of the received commit.
A system comprising:
- a memory that stores instructions;
- and
- one or more processors configured by the instructions to perform operations comprising:
- generating training data comprising:
- a first set of pairs wherein each pair of the first set of pairs comprises a commit message and a corresponding code change in an existing project;
- and
- a second set of pairs wherein each pair of the second set of pairs comprises a commit message and a randomly selected code change that is different than the corresponding code change for the commit message;
- training a machine learning model, using the generated training data, to generate a trained machine learning model configured to generate a score indicating a level of discrepancy between a commit message and a corresponding code change;
- receiving a commit comprising a given commit message and a given corresponding code change;
- and
- analyzing, using the trained machine learning model, the given commit message and given corresponding code change to generate a score indicating the level of discrepancy between the given commit message and the given corresponding code change of the received commit.
A non-transitory computer-readable medium comprising instructions stored thereon that are executable by at least one processor to cause a computing device to perform operations comprising:
- generating training data comprising:
- a first set of pairs wherein each pair of the first set of pairs comprises a commit message and a corresponding code change in an existing project;
- and
- a second set of pairs wherein each pair of the second set of pairs comprises a commit message and a randomly selected code change that is different than the corresponding code change for the commit message;
- training a machine learning model, using the generated training data, to generate a trained machine learning model configured to generate a score indicating a level of discrepancy between a commit message and a corresponding code change;
- receiving a commit comprising a given commit message and a given corresponding code change;
- and
- analyzing, using the trained machine learning model, the given commit message and given corresponding code change to generate a score indicating the level of discrepancy between the given commit message and the given corresponding code change of the received commit.
Dependencies are inferred from the claim text.