Detecting Malware Activity Using Public Search Data

Software vulnerability detection method based on code attribute graph presentation and Bi-LSTM neural network extraction

“…In order to compare with the work of Lin et al [8], The fragility FFmpeg, LibTIFE, LibPNG, Pidgin, VLC Media player and Asterisk constructed by Lin et al [8] have been segmented and tagged by function. The tags of the data-level vulnerability function are tagged according to the data of the National Vulnerability Database [1] and the Public Vulnerability Database [13]. Vulnerability across multiple functions or files is discarded in this data set; Excluding the known vulnerability function, Treat the remaining functions as non-vulnerable functions.Vulnerability detectors can be built at the function level from these function-level vulnerability data, Thereby achieving finer-grained detection capability than in files.…”

“…With the increasing complexity of software systems, The scale is getting bigger and bigger, The frequency of vulnerability codes is also increasing. According to the data published by the National Vulnerability Database [1], The number of vulnerability codes published in 2018 was 1,616,511, 17,305 in 2019 and 18,352 in 2020. It can be seen from the data that the number of vulnerability codes has been increasing, so it is very important to find the vulnerability codes in software in time.…”