Software Similarity Dataset

This dataset contains the post-processed data for software similarity learning. More information is given: SoftwareSim_Github

post_process: All embedded software with autoencoder to make sure each function is the same length (1024 bits), each final is the embedded graph representation of software.

final_data: All information obtained by Somef & Inspect4py as well as cleaning. Each file represents software in the format given --> Function_Name: [[Called Function], [Function Tokens]]

lean_simscore.csv: This file contains software pairs as well as the similarity metrics, format is given:

Property
Example

Graph_1
kakaobrain_helo_word

Graph_2
mblondel_soft-dtw

miniLM
0.4503

Sbert
0.7204

TSDAE
0.5714