Korea University Research and Business Foundation

Hakjoo Oh

Junhee Lee

Seongjoon Hong

Sooyoung Cha

Dowon Song

Wonseok Oh

Disclosed is a method for repairing a null pointer exception performed by a computing device including a processor, which includes: obtaining an original program; and obtaining, by using a classification model, a patch program from the original program by repairing an error of the original program; and, in which the classification model may be trained in advance by using a null processing pattern dataset obtained from at least one program.

Method for repairing null pointer exception

Disclosed is a device and method for automatically repairing type errors occurring in a dynamically typed language. A method of automatically repairing a type error refers to a method of automatically repairing a type error occurring in a dynamically typed language, performed by a computing device including a processor and includes acquiring a buggy program, a testcase, and error information; collecting a candidate variable in which a type error occurs; inferring a negative type that is a type of the candidate variable observed from a negative testcase and a positive type that is a type of the candidate variable observed from a positive testcase; estimating a fault location at which the type error occurs; generating a patch; and generating a fixed program by applying the patch to the buggy program, and the collecting of the candidate variable includes selecting a variable included in the error information as the candidate variable.

Device and method for program repair for type errors in dynamically typed language

Disclosed is a method for feedback of a program, which is performed by a computing device including at least one processor, which include: matching respective functions included in a first program received from an external device with respective functions included in a prestored correct answer program corresponding to the first program; identifying error functions different from the respective functions included in the correct answer program among the respective functions included in the first program; generating at least one repair template including at least one reference function extracted based on a calling-context for a function of a specific function in order to repair the error function among the prestored functions; generating a first edit script for repairing the error function to at least one reference function based on a first repair template among one or more repair templates; and transmitting a signal including the first edit script to the external device.

Method for feedback of program

A symbolic execution device, a symbolic execution method, and a probability distribution update method for symbolic execution are provided. The symbolic execution device includes a storage unit for storing at least one of a first probability distribution and a second probability distribution and a processor for sampling and obtaining a weight vector from the first probability distribution, for sampling and obtaining a removal rate from the second probability distribution, for adding the weight vector to at least one feature vector obtained from at least one candidate state to obtain a score for each of the at least one candidate state, and for selecting a state to be removed from among the at least one candidate state by using the score of each of the at least one candidate state and the removal rate.

Method for automatically pruning search space of symbolic execution via machine learning

A device for automatically repairing memory deallocation errors is disclosed. The device includes: a static analysis unit configured to generate status information for each one of the objects included in the source code of a program by way of a static analysis of the source code, where the status information includes position information, pointer information, and patch information, the position information associated with allocation sites of the objects, the pointer information associated with pointers pointing to the objects, the patch information associated with deallocation statements capable of deallocating the objects; a decision unit configured to choose patch candidates from the patch information and decide on a combination of the patch candidates capable of deallocating each of the objects only once; and a repair unit configured to repair the source code according to the combination of patch candidates.

Device and method for automatically repairing memory deallocation errors

Provided is a method for automatically generating a search heuristic that is optimal for a test subject program and a method of concolic testing that uses a parameterized search heuristic to yield a consistent test performance for any program.

Growing community of inventors

Seoul, South Korea

Hakjoo Oh