Methods and arrangements to process comments
AssigneeCapital One Services, LLC
InventorsReza FARIVAR, Kenneth TAYLOR, Austin Grant WALTERS, Fardin ABDI TAGHI ABAD, Anh TRUONG, Vincent PHAM, Jeremy Edward GOODSITT
Life of this patent
Estimating the term…
Abstract
Described herein are embodiments for managing comments in a program code file. A system may select program code and compile it to an intermediary code. The system may compare the intermediary code to a library of intermediary code snippets associated with comments. Based on the comparison, a system may recognize the code to be obsolete. In some embodiments, a system may generate one or more recommendations to update a code. Based on received feedback regarding a recommendation, a system may accordingly update a code.
Claims20
A computer implemented method to process comments, comprising:
- identifying a program code segment with a comment;
- determining an intermediate code segment during compiling of the program code segment;
- identifying a library code segment corresponding to the intermediate code segment;
- identifying a set of one or more comments based on an association between the library code segment and the set of one or more comments;
- comparing the comment with the set of one or more comments to determine that a deviation between the comment and the set of one or more comments exceeds a deviation threshold;
- and
- outputting an indication that the comment might be obsolete.
An apparatus, comprising:
- memory;
- and
- logic circuitry coupled with the memory to:
- determine a program code segment;
- parse the program code segment to generate a training code segment, the training code segment includes at least one comment;
- compile the training code segment to generate an intermediate code segment associated with the training code segment and the at least one comment of the training code segment;
- and
- train a machine learning model with the intermediate code segment to detect similarities between input comments and comments in a library.
A computer-implemented method, comprising:
- determining a program code segment;
- parsing the program code segment to generate a training code segment, the training code segment includes at least one comment;
- compiling the training code segment to generate an intermediate code segment associated with the training code segment and the at least one comment of the training code segment;
- and
- training a machine learning model with the intermediate code segment to detect similarities between input comments and comments in a library.
Dependencies are inferred from the claim text.