Supervised Deep Features for Software Functional Clone Detection by Exploiting Lexical and Syntactical Information in Source Code

By Huihui Wei, Ming Li. 2017

316 citations recorded by the source corpus.

Subfields: Information Systems, Software, Program Analysis & Verification, Signal Processing

How researchers see this paper

The researchers whose trusted networks rank it highest, strongest advocate first. Every researcher gets a different order.

In Information Systems

In Software

In Program Analysis & Verification

In Signal Processing

Citations in the corpus

CodeXGLUE: A Machine Learning Benchmark Dataset for Code Understanding and Generation (2021, methodology): “CDLH [88] learns representations of code fragments via AST-based LSTM.”
DOBF: A Deobfuscation Pre-Training Objective for Programming Languages (2021, methodology): “Machine learning is frequently used on tasks involving programming languages, including code completion [28, 33, 35, 42], bug detection and code repair [15, 38, 46], code summarization [8, 25], clone detection [2, 47, 48], code search [13, 23] and code translation [14, 41].”
Learning-Based Recursive Aggregation of Abstract Syntax Trees for Code Clone Detection (2019, methodology): “Code Clone Detection The closest related work [16] also uses a Siamese Network of Recursive LSTM to learn a representation of code in a supervised way.”
Detecting Code Clones with Graph Neural Network and Flow-Augmented Abstract Syntax Tree (2020, methodology): “CDLH [3] uses binary Tree-LSTM [5] to encode ASTs, and a hash function to optimize the distance between the vector representation of AST pairs by hamming distance.”
Retrieval-based Neural Source Code Summarization (2020, methodology): “Recently neural models are incorporated to learn vector representations of source code and use these vectors to compute code similarity [56, 57, 65, 67], which can achieve better performance but need additional time-consuming training and labeled data.”
DeepSim: deep learning code functional similarity (2018, methodology): “The F1 score of DeepSim is 0.98, compared to 0.82 by CDLH. DeepSim does not achieve 1.0 F1 score on T1-ST3 clones (which should be guaranteed by our encoding approaches as discussed in Section 3), because the tool misses some data flow when generating DFG from source code, due to the approximation w”
Multi-task Learning based Pre-trained Language Model for Code Completion (2020, methodology): “nguage. Thus, statistical language models have been used for source code modeling [19, 42, 46], benefiting many software engineering tasks, including code summarization [23, 49], code clone detection [51, 52] program repair [15, 47], especially, in code completion [17, 19, 27, 46]. Code completion i”
Neural Detection of Semantic Code Clones Via Tree-Based Convolution (2019, methodology): “However, the CDLH approach [9] does not make sufficient use of the structural information in an AST, despite such information being crucial to the semantics of the corresponding code.”
A Novel Neural Source Code Representation Based on Abstract Syntax Tree (2019, methodology): “For example, CDLH [6] can only have the F1 value of 57% in one public benchmark for clone detection, and the studies in NLP [23], [41], [21] show that the tree size and depth do matter and have signiﬁcant impact on the performance.”
HELoC: Hierarchical Contrastive Learning of Source Code Representation (2022, methodology): “CDLH [58] learns code representations via AST-based LSTM and uses a hash function to optimize the func-”