Path-Based Function Embedding and its Application to Specification\n Mining

Abstract

Identifying the relationships among program elements is useful for program\nunderstanding, debugging, and analysis. One such relationship is synonymy.\nFunction synonyms are functions that play a similar role in code, e.g.\nfunctions that perform initialization for different device drivers, or\nfunctions that implement different symmetric-key encryption schemes. Function\nsynonyms are not necessarily semantically equivalent and can be syntactically\ndissimilar; consequently, approaches for identifying code clones or functional\nequivalence cannot be used to identify them. This paper presents func2vec, an\nalgorithm that maps each function to a vector in a vector space such that\nfunction synonyms are grouped together. We compute the function embedding by\ntraining a neural network on sentences generated from random walks over an\nencoding of the program as a labeled pushdown system (l-PDS). We demonstrate\nthat func2vec is effective at identifying function synonyms in the Linux\nkernel. Furthermore, we show how function synonyms enable mining error-handling\nspecifications with high support in Linux file systems and drivers.\n