arxiv: 2601.13334 · v2 · submitted 2026-01-19 · 💻 cs.SE

SEER: Spectral Entropy Encoding of Roles for Context-Aware Attention-Based Design Pattern Detection

Tarik Houichime , Younes El Amrani This is my paper

Pith reviewed 2026-05-16 12:52 UTC · model grok-4.3

classification 💻 cs.SE

keywords design pattern detectionspectral entropyrole encodingLaplacian spectrumattention-based classificationGoF patternscode context modeling

0 comments p. Extension

The pith

SEER encodes class roles from Laplacian spectra and weights calls by duration priors to raise design pattern detection accuracy.

A machine-rendered reading of the paper's core claim, the machinery that carries it, and where it could break.

The paper upgrades an earlier attention-based detector for Gang of Four patterns by adding two components that clarify who performs which action inside each class. A spectral-entropy encoder extracts role embeddings directly from the Laplacian spectrum of the class interaction graph, while a separate module assigns learned duration weights to different kinds of method calls such as constructors and virtual dispatches. Together these changes steer the Transformer toward role-consistent and temporally relevant signals without increasing model size. On the PyDesignNet collection the revised system records modest but consistent lifts in macro-F1 and accuracy while cutting false positives by nearly twenty percent. Readers interested in practical code analysis would value the gains because they come from tighter modeling of structure rather than from more data or parameters.

Core claim

SEER augments the prior Context Is All You Need pipeline with a spectral-entropy role encoder that produces per-member embeddings from the Laplacian spectrum of each class interaction graph and with a time-weighted calling context that applies empirically calibrated duration priors to method categories. These additions produce macro-F1 of 93.20 percent and accuracy of 93.98 percent on PyDesignNet while lowering false positives by nearly 20 percent, delivering interpretable symbol-level attributions that match canonical roles, and preserving portability across languages.

What carries the argument

Spectral-entropy role encoder that extracts per-member embeddings from the Laplacian spectrum of each class interaction graph, paired with time-weighted calling context that assigns calibrated duration priors to method categories.

If this is right

Macro-F1 rises from 92.47 percent to 93.20 percent and accuracy from 92.52 percent to 93.98 percent.
False-positive rate drops by nearly 20 percent.
Symbol-level attributions align with canonical design-pattern roles.
Performance remains stable under small perturbations of the interaction graphs.
The encoder and context modules integrate directly with existing Transformer sequence models.

Where Pith is reading between the lines

These are editorial extensions of the paper, not claims the author makes directly.

The same spectrum-derived role vectors could be reused as input features for related tasks such as defect prediction or automated refactoring.
If duration priors prove stable, similar temporal weighting may improve attention models that process other kinds of sequential software artifacts.
The approach suggests that graph spectra can serve as a lightweight, language-agnostic substitute for hand-engineered role features in code mining pipelines.

Load-bearing premise

The Laplacian spectrum of a class interaction graph supplies distinct and useful role signals, and the duration priors fitted on PyDesignNet transfer to new codebases.

What would settle it

Running SEER on an independent corpus of Java or C++ projects that contain the same GoF patterns and observing that the macro-F1 gain disappears or reverses would falsify the claim that the two additions produce reliable improvements.

read the original abstract

This paper presents SEER, an upgraded version of our prior method Context Is All You Need for detecting Gang of Four (GoF) design patterns from source code. The earlier approach modeled code as attention-ready sequences that blended lightweight structure with behavioral context; however, it lacked explicit role disambiguation within classes and treated call edges uniformly. SEER addresses these limitations with two principled additions: (i) a spectral-entropy role encoder that derives per-member role embeddings from the Laplacian spectrum of each class's interaction graph, and (ii) a time-weighted calling context that assigns empirically calibrated duration priors to method categories (e.g., constructors, getters/setters, static calls, virtual dispatch, cloning). Together, these components sharpen the model's notion of "who does what" and "how much it matters," while remaining portable across languages with minimal adaptation and fully compatible with Transformer-based sequence encoders. Importantly, SEER does not "force" a win by capacity or data; it nudges the classifier, steering attention toward role-consistent and temporally calibrated signals that matter most. We evaluate SEER on PyDesignNet (1,832 files, 35,000 sequences, 23 GoF patterns) and observe consistent gains over our previous system: macro-F1 increases from 92.47% to 93.20% and accuracy from 92.52% to 93.98%, with macro-precision 93.98% and macro-recall 92.52%. Beyond aggregate metrics, SEER reduces false positives by nearly 20%, a decisive improvement that strengthens its robustness and practical reliability. Moreover, SEER yields interpretable, symbol-level attributions aligned with canonical roles, exhibits robustness under small graph perturbations, and shows stable calibration.

Editorial analysis

A structured set of objections, weighed in public.

Desk editor's note, referee report, simulated authors' rebuttal, and a circularity audit. Tearing a paper down is the easy half of reading it; the pith above is the substance, this is the friction.

Desk Editor's Note private letter to a colleague

SEER adds a spectral-entropy role encoder from Laplacian spectra and time-weighted call priors to the authors' prior detector, with modest metric gains but thin evidence that the new parts drive the lift.

read the letter

The core update here is two additions to the authors' earlier Context Is All You Need system: a spectral-entropy encoder that pulls role embeddings from the Laplacian spectrum of each class interaction graph, and time-weighted calling context that uses duration priors for method categories such as constructors, getters, and virtual calls. These are presented as fixes for weak role disambiguation and uniform treatment of edges in the prior work. On PyDesignNet the numbers move from 92.47% macro-F1 to 93.20% and accuracy from 92.52% to 93.98%, with a claimed 20% false-positive drop. That is the positive side: the changes are described cleanly and the model stays compatible with standard Transformer encoders. The paper also notes some interpretability and stability under small perturbations. The soft spots sit right at the center of the claims. No ablations isolate what the spectral encoder or the priors actually contribute, there are no error bars or significance tests on the gains, and the duration priors are only called empirically calibrated without showing the data split used for calibration. If that calibration touched the same test files, part of the reported improvement could be fitting rather than the method. There is also no supporting check, such as eigenvector inspection or role clustering, that the Laplacian spectrum actually separates canonical GoF roles in a useful way. This is incremental engineering work aimed at people already doing automated design-pattern detection in code. A specialist in that niche might pick up the spectral idea or the weighting scheme and try them, but the paper does not look like a broad advance or a result that would change how most readers approach the task. I would send it for peer review so the authors can add the missing controls and make the experimental case tighter.

Referee Report

3 major / 1 minor

Summary. The paper presents SEER as an extension of a prior attention-based method for detecting Gang of Four design patterns in source code. It introduces two additions: a spectral-entropy role encoder that derives per-member role embeddings from the Laplacian spectrum of each class's interaction graph, and a time-weighted calling context that assigns empirically calibrated duration priors to method categories such as constructors and getters. Evaluated on PyDesignNet (1,832 files, 35,000 sequences, 23 patterns), SEER reports macro-F1 rising from 92.47% to 93.20%, accuracy from 92.52% to 93.98%, and a nearly 20% reduction in false positives, while claiming improved interpretability and robustness under graph perturbations.

Significance. If the metric gains prove robust and the spectral encoder demonstrably contributes role-specific information, the work could advance context-aware design pattern detection by improving disambiguation of roles within classes and providing temporally calibrated signals, with potential benefits for practical software engineering tools that require reliable and interpretable pattern detection.

major comments (3)

[Abstract] Abstract: the reported lifts (0.73 pp macro-F1, 1.46 pp accuracy, 20% FP reduction) are presented without error bars, statistical tests, ablation studies, or per-component breakdowns, so it is impossible to determine whether the gains are significant or attributable to the spectral-entropy encoder versus the duration priors.
[Evaluation] Evaluation: the duration priors are described as 'empirically calibrated' on PyDesignNet; without explicit confirmation that calibration was performed on held-out data or a separate validation split, the improvements remain compatible with dataset-specific tuning rather than the claimed principled role disambiguation.
[Method] Method: no analysis (eigenvector inspection, role-clustered embeddings, or controlled ablation removing the spectral-entropy encoder) is supplied to show that the Laplacian spectrum of the class interaction graph encodes information about distinct GoF roles beyond what the prior sequence model already captures.

minor comments (1)

[Abstract] Abstract: the statement that SEER is 'fully compatible with Transformer-based sequence encoders' would be strengthened by a short note on integration overhead or parameter count.

Simulated Author's Rebuttal

3 responses · 0 unresolved

We thank the referee for the constructive feedback. The comments highlight important areas for improving the rigor and clarity of our claims regarding the contributions of the spectral-entropy encoder and time-weighted context. We address each major comment below and will incorporate the requested analyses and details in the revised manuscript.

read point-by-point responses

Referee: [Abstract] Abstract: the reported lifts (0.73 pp macro-F1, 1.46 pp accuracy, 20% FP reduction) are presented without error bars, statistical tests, ablation studies, or per-component breakdowns, so it is impossible to determine whether the gains are significant or attributable to the spectral-entropy encoder versus the duration priors.

Authors: We agree that the presentation of results would be strengthened by additional statistical support and breakdowns. In the revised version, we will augment the abstract and results section with error bars computed over multiple random seeds, statistical significance tests (e.g., McNemar's test for paired comparisons), and explicit ablation studies that isolate the spectral-entropy role encoder from the time-weighted calling context. These additions will clarify the magnitude and attribution of the reported improvements. revision: yes
Referee: [Evaluation] Evaluation: the duration priors are described as 'empirically calibrated' on PyDesignNet; without explicit confirmation that calibration was performed on held-out data or a separate validation split, the improvements remain compatible with dataset-specific tuning rather than the claimed principled role disambiguation.

Authors: The current manuscript describes the priors as empirically calibrated on PyDesignNet without detailing the data split. We will revise the evaluation section to provide a full description of the calibration procedure, explicitly confirming the use of a held-out validation split (distinct from training and test sets) and reporting the calibration methodology. This will demonstrate that the priors reflect generalizable temporal signals rather than test-set tuning. revision: yes
Referee: [Method] Method: no analysis (eigenvector inspection, role-clustered embeddings, or controlled ablation removing the spectral-entropy encoder) is supplied to show that the Laplacian spectrum of the class interaction graph encodes information about distinct GoF roles beyond what the prior sequence model already captures.

Authors: We acknowledge that the manuscript does not currently include direct analyses of the spectral-entropy encoder's role-specific information. In the revision, we will add eigenvector inspection of the class interaction graph Laplacians, visualizations of role-clustered embeddings, and a controlled ablation that removes the spectral-entropy component while retaining the time-weighted context. These will quantify the encoder's contribution to role disambiguation beyond the baseline sequence model. revision: yes

Circularity Check

2 steps flagged

Duration priors reduce to training-set calibration; spectral role link asserted without derivation

specific steps

fitted input called prediction [Abstract]
"a time-weighted calling context that assigns empirically calibrated duration priors to method categories (e.g., constructors, getters/setters, static calls, virtual dispatch, cloning)"

The priors are calibrated empirically on PyDesignNet; the same splits are used for the reported macro-F1/accuracy gains, so the improvement is statistically forced by the calibration step rather than derived from first principles.
other [Abstract]
"a spectral-entropy role encoder that derives per-member role embeddings from the Laplacian spectrum of each class's interaction graph"

The derivation is asserted without an explicit mapping, clustering validation, or controlled ablation; the claimed role disambiguation therefore reduces to an unverified assumption rather than a demonstrated reduction from the Laplacian spectrum.

full rationale

The paper's central performance claim rests on two additions presented as principled. The time-weighted context component explicitly uses 'empirically calibrated duration priors' fitted to method categories on PyDesignNet; because the same dataset supplies both calibration and final metrics, the reported 0.73 pp macro-F1 lift is partly forced by that fit rather than an independent derivation. The spectral-entropy encoder is asserted to 'derive per-member role embeddings from the Laplacian spectrum' yet supplies no equation, eigenvector inspection, or ablation showing that the spectrum encodes GoF-distinct roles beyond what the prior sequence model already captured. These two reductions together produce partial circularity (score 6) while the remainder of the architecture remains non-circular.

Axiom & Free-Parameter Ledger

1 free parameters · 1 axioms · 0 invented entities

The central claim rests on two domain assumptions: that Laplacian spectra of class interaction graphs encode role distinctions, and that method-category duration priors can be calibrated once and reused. No free parameters are explicitly listed beyond the calibration step; no new physical or mathematical entities are introduced.

free parameters (1)

duration priors for method categories
Empirically calibrated values assigned to constructors, getters/setters, static calls, virtual dispatch, and cloning; these values directly influence the time-weighted context signal.

axioms (1)

domain assumption Laplacian spectrum of a class interaction graph encodes per-member role information
Invoked to justify the spectral-entropy role encoder; no proof or external validation supplied in the abstract.

pith-pipeline@v0.9.0 · 5623 in / 1599 out tokens · 41719 ms · 2026-05-16T12:52:21.812577+00:00 · methodology

discussion (0)

Lean theorems connected to this paper

Citations machine-checked in the Pith Canon. Every link opens the source theorem in the public Lean library.

IndisputableMonolith/Foundation/AbsoluteFloorClosure.lean reality_from_one_distinction unclear

?

unclear
Relation between the paper passage and the cited Recognition theorem.

spectral-entropy role encoder that derives per-member role embeddings from the Laplacian spectrum of each class's interaction graph ... H(G) = -∑ p_i log₂ p_i with p_i = λ_i / ∑ λ_j
IndisputableMonolith/Cost/FunctionalEquation.lean washburn_uniqueness_aczel unclear

?

unclear
Relation between the paper passage and the cited Recognition theorem.

time-weighted calling context that assigns empirically calibrated duration priors to method categories (constructors Σ=2.5τ, getters ϕ=0.25τ, ...)

What do these tags mean?

matches: The paper's claim is directly supported by a theorem in the formal canon.
supports: The theorem supports part of the paper's argument, but the paper may add assumptions or extra steps.
extends: The paper goes beyond the formal theorem; the theorem is a base layer rather than the whole result.
uses: The paper appears to rely on the theorem as machinery.
contradicts: The paper's claim conflicts with a theorem or certificate in the canon.
unclear: Pith found a possible connection, but the passage is too broad, indirect, or ambiguous to say the theorem truly supports the claim.

Reference graph

Works this paper leans on

74 extracted references · 74 canonical work pages · 7 internal anchors

[1]

Nelson, M. L. A Survey of Reverse Engineering and Program Comprehension (2005)

work page 2005
[2]

Ogheneovo, E. E. On the Relationship between Software Complexity and Maintenance Costs.Journal of Computer and Communications2, 1–16 (2014)

work page 2014
[3]

Dehaghani, S. M. H. & Hajrahimi, N. Which Factors Affect Software Projects Maintenance Cost More?Acta Informatica Medica21, 63–66 (2013)

work page 2013
[4]

& Hasheminejad, S

Yarahmadi, H. & Hasheminejad, S. M. H. Design pattern detection approaches: A systematic review of the literature.Artificial Intelligence Review53, 5789–5846 (2020)

work page 2020
[5]

& Rasool, G.Recovery of Mobile Game Design Patterns, 1–7 (2020)

Khan, M. & Rasool, G.Recovery of Mobile Game Design Patterns, 1–7 (2020)

work page 2020
[6]

Yu, D.et al.Efficiently detecting structural design pattern instances based on ordered sequences.Journal of Systems and Software142, 35–56 (2018)

work page 2018
[7]

& Gupta, M

Singh, J., Chowdhuri, S., Bethany, G. & Gupta, M. Detecting design patterns: A hybrid approach based on graph matching and static analysis.Information Technology and Management23(2022)

work page 2022
[8]

Nacef, A., Bahroun, S., Khalfallah, A. & Ben Ahmed, S.Automatic Detection of Implicit and Typical Implementation of Singleton Pattern Based on Super- vised Machine Learning:, 202–210 (SCITEPRESS - Science and Technology Publications, Lisbon, Portugal, 2023)

work page 2023
[9]

& Smith, J.Limitations of the unique-attribute representation for a learning system, 219–225 (1993)

Bayazitoglu, A., Johnson, T. & Smith, J.Limitations of the unique-attribute representation for a learning system, 219–225 (1993)

work page 1993
[10]

& Sanborn, A

Spicer, J. & Sanborn, A. N. What does the mind learn? A comparison of human and machine learning representations.Current Opinion in Neurobiology55, 97– 102 (2019)

work page 2019
[11]

& Amrani, Y

Houichime, T. & Amrani, Y. E. Introduction to Analytical Software Engineering Design Paradigm (2025). arXiv:2505.11979

work page arXiv 2025
[12]

& Zheng, Y

Nazar, N., Aleti, A. & Zheng, Y. Feature-based software design pattern detection. Journal of Systems and Software185, 111179 (2022)

work page 2022
[13]

& Egyed, A.Feature Maps: A Comprehensible Software Representation for Design Pattern Detection, 207–217 (2019)

Thaller, H., Linsbauer, L. & Egyed, A.Feature Maps: A Comprehensible Software Representation for Design Pattern Detection, 207–217 (2019)

work page 2019
[14]

J., Murty, M

Johnson, S. J., Murty, M. R. & Navakanth, I. A detailed review on word embed- ding techniques with emphasis on word2vec.Multimedia Tools and Applications 83, 37979–38007 (2024). 34

work page 2024
[15]

& Yahav, E

Alon, U., Zilberstein, M., Levy, O. & Yahav, E. Code2vec: Learning distributed representations of code.Implementation, data and a trained model for the code2vec paper3, 40:1–40:29 (2019)

work page 2019
[16]

Dissecting Contextual Word Embeddings: Architecture and Representation

Peters, M. E., Neumann, M., Zettlemoyer, L. & Yih, W.-t. Dissecting Contextual Word Embeddings: Architecture and Representation (2018). arXiv:1808.08949

work page internal anchor Pith review Pith/arXiv arXiv 2018
[17]

Understanding the Effective Receptive Field in Deep Convolutional Neural Networks

Luo, W., Li, Y., Urtasun, R. & Zemel, R. Understanding the Effective Receptive Field in Deep Convolutional Neural Networks (2017). arXiv:1701.04128

work page internal anchor Pith review Pith/arXiv arXiv 2017
[18]

Structural Representations of Music Performance.Proceedings of the Annual Meeting of the Cognitive Science Society11(1989)

Palmer, C. Structural Representations of Music Performance.Proceedings of the Annual Meeting of the Cognitive Science Society11(1989)

work page 1989
[19]

& Krumhansl, C

Palmer, C. & Krumhansl, C. L. Independent temporal and pitch structures in determination of musical phrases.Journal of Experimental Psychology. Human Perception and Performance13, 116–126 (1987)

work page 1987
[20]

R.et al.Perception of phrase structure in music.Human Brain Mapping24, 259–273 (2005)

Knösche, T. R.et al.Perception of phrase structure in music.Human Brain Mapping24, 259–273 (2005)

work page 2005
[21]

& El Amrani, Y

Houichime, T. & El Amrani, Y. Context is All You Need: A Hybrid Attention- Based Method for Detecting Code Design Patterns.IEEE Access1–1 (2025)

work page 2025
[22]

Chung, F. R. K.Spectral Graph Theory(American Mathematical Soc., 1997)

work page 1997
[23]

A.Spectral Graph Theory and its Applications, 29–38 (2007)

Spielman, D. A.Spectral Graph Theory and its Applications, 29–38 (2007)

work page 2007
[24]

& Biamonte, J

De Domenico, M. & Biamonte, J. Spectral Entropies as Information-Theoretic Tools for Complex Network Comparison.Physical Review X6, 041062 (2016)

work page 2016
[25]

https://www.oracle.com/java/technologies/whitepaper.html

The Java HotSpot Performance Engine Architecture. https://www.oracle.com/java/technologies/whitepaper.html

work page
[26]

https://www.oracle.com/technical-resources/articles/java/architect-evans- pt1.html

Understanding Java JIT Compilation with JITWatch, Part 1. https://www.oracle.com/technical-resources/articles/java/architect-evans- pt1.html

work page
[27]

https://wiki.openjdk.org/display/HotSpot/Inlining

Inlining-Inlining-OpenJDKWiki. https://wiki.openjdk.org/display/HotSpot/Inlining

work page
[28]

& Ungar, D

Hölzle, U., Chambers, C. & Ungar, D. America, P. (ed.)Optimizing dynamically- typed object-oriented languages with polymorphic inline caches. (ed.America, P.)ECOOP’91 European Conference on Object-Oriented Programming, 21–38 (Springer, Berlin, Heidelberg, 1991)

work page 1991
[29]

& Mössenböck, H.Partial Escape Analysis and Scalar Replacement for Java, CGO ’14, 165–174 (Association for Computing Machinery, New York, NY, USA, 2018)

Stadler, L., Würthinger, T. & Mössenböck, H.Partial Escape Analysis and Scalar Replacement for Java, CGO ’14, 165–174 (Association for Computing Machinery, New York, NY, USA, 2018). 35

work page 2018
[30]

& Mössenböck, H.Escape Analysis in the Context of Dynamic Compilation and Deoptimization(2005)

Kotzmann, T. & Mössenböck, H.Escape Analysis in the Context of Dynamic Compilation and Deoptimization(2005)

work page 2005
[31]

https://html.spec.whatwg.org/multipage/structured- data.html

Living Standard. https://html.spec.whatwg.org/multipage/structured- data.html

work page
[32]

https://developer.mozilla.org/en-US/docs/Web/API/Web_Workers_API/Structured_clone_algorithm (2025)

The structured clone algorithm - Web APIs | MDN. https://developer.mozilla.org/en-US/docs/Web/API/Web_Workers_API/Structured_clone_algorithm (2025)

work page 2025
[33]

Numerical Solution of Natural Frequencies and Mode Shapes

Koubova, L. Numerical Solution of Natural Frequencies and Mode Shapes. Procedia Structural Integrity63, 35–42 (2024)

work page 2024
[34]

S., Malik, A

Azam, M. S., Malik, A. H., Irshad, A., Iqbal, M. & Ahmad, I. Measurement of natural frequency and mechanical damping of thin brass diaphragm by pulsed laser generated vibrations.Journal of Vibroengineering24, 1226–1234 (2022)

work page 2022
[35]

McGraw, P. N. & Menzinger, M. Laplacian Spectra as a Diagnostic Tool for Network Structure and Dynamics.Physical Review E77, 031102 (2008)

work page 2008
[36]

I., Narang, S

Shuman, D. I., Narang, S. K., Frossard, P., Ortega, A. & Vandergheynst, P. The Emerging Field of Signal Processing on Graphs: Extending High-Dimensional Data Analysis to Networks and Other Irregular Domains.IEEE Signal Processing Magazine30, 83–98 (2013)

work page 2013
[37]

A construction of cospectral graphs for the normalized Laplacian

Butler, S. & Grout, J. A construction of cospectral graphs for the normalized Laplacian (2012). arXiv:1008.3646

work page internal anchor Pith review Pith/arXiv arXiv 2012
[38]

Are almost all graphs determined by their spectrum?Notices of the South African Mathematical Society47, 42–45 (2016)

Haemers, W. Are almost all graphs determined by their spectrum?Notices of the South African Mathematical Society47, 42–45 (2016)

work page 2016
[39]

Brimkov, B.et al.Graphs that are cospectral for the distance Laplacian.The Electronic Journal of Linear Algebra36, 334–351 (2020)

work page 2020
[40]

& EL Amrani, Y

Houichime, T. & EL Amrani, Y. PyDesignNet Dataset.Kaggle(2024)

work page 2024
[41]

Strang, G.Lecture Notes for Linear Algebra

work page
[42]

& Rose, C

Keller, M. & Rose, C. Gaussian upper bounds for heat kernels on graphs with unbounded geometry (2022). arXiv:2206.04690

work page arXiv 2022
[43]

The von Neumann entropy of networks

Passerini, F. & Severini, S.The von Neumann Entropy of Networks(2011). arXiv:0812.2597

work page internal anchor Pith review Pith/arXiv arXiv 2011
[44]

Sun, Y., Zhao, H., Liang, J. & Ma, X. Eigenvalue-based entropy in directed complex networks.PLOS ONE16, e0251993 (2021). 36

work page 2021
[45]

On the Von Neumann Entropy of Graphs

Minello, G., Rossi, L. & Torsello, A. On the Von Neumann Entropy of Graphs (2018). arXiv:1809.07533

work page internal anchor Pith review Pith/arXiv arXiv 2018
[46]

Liu, X., Fu, L. & Wang, X.Bridging the Gap between von Neumann Graph Entropy and Structural Information: Theory and Applications, WWW ’21, 3699– 3710 (Association for Computing Machinery, New York, NY, USA, 2021)

work page 2021
[47]

A continuity property of the entropy density for spin lattice systems

Fannes, M. A continuity property of the entropy density for spin lattice systems. Communications in Mathematical Physics31, 291–294 (1973)

work page 1973
[48]

Audenaert, K. M. R. A Sharp Fannes-type Inequality for the von Neumann Entropy.Journal of Physics A: Mathematical and Theoretical40, 8127–8136 (2007)

work page 2007
[49]

& Stella, F

Zanoni, M., Arcelli Fontana, F. & Stella, F. On applying machine learning techniques for design pattern detection.Journal of Systems and Software103, 102–117 (2015)

work page 2015
[50]

& Ziadi, T

Mzid, R., Rezgui, I. & Ziadi, T. Galster, M.et al.(eds)Attention-based method for design pattern detection. (eds Galster, M.et al.)Software Architecture, 86–101 (Springer Nature Switzerland, Cham, 2024)

work page 2024
[51]

& Halkidis, S

Tsantalis, N., Chatzigeorgiou, A., Stephanides, G. & Halkidis, S. T. Design Pattern Detection Using Similarity Scoring.IEEE Transactions on Software Engineering32, 896–909 (2006)

work page 2006
[52]

& Sartipi, K

Hu, L. & Sartipi, K. Dynamic Analysis and Design Pattern Detection in Java Programs.Proceedings of the Twentieth International Conference on Software Engineering & Knowledge Engineering (SEKE’2008), San Francisco, CA, USA, July 1-3, 2008846 (2008)

work page 2008
[53]

CodeBERT: A Pre-Trained Model for Programming and Natural Languages

Feng, Z.et al.CodeBERT: A Pre-Trained Model for Programming and Natural Languages (2020). arXiv:2002.08155

work page internal anchor Pith review Pith/arXiv arXiv 2020
[54]

GraphCodeBERT: Pre-training Code Representations with Data Flow

Guo, D.et al.GraphCodeBERT: Pre-training Code Representations with Data Flow (2021). arXiv:2009.08366

work page internal anchor Pith review Pith/arXiv arXiv 2021
[55]

& Chen, X

Wu, J., Zhao, Z., Sun, C., Yan, R. & Chen, X. Few-shot transfer learning for intelligent fault diagnosis of machine.Measurement166, 108202 (2020)

work page 2020
[56]

& O’Hare, N

Gupta, A., Thadani, K. & O’Hare, N. Scott, D., Bel, N. & Zong, C. (eds) Effective Few-Shot Classification with Transfer Learning. (eds Scott, D., Bel, N. & Zong, C.)Proceedings of the 28th International Conference on Computational Linguistics, 1061–1066 (International Committee on Computational Linguistics, Barcelona, Spain (Online), 2020). 37

work page 2020
[57]

& Shen, B

Li, X., Yuan, S., Gu, X., Chen, Y. & Shen, B. Few-shot code translation via task-adapted prompt learning.Journal of Systems and Software212, 112002 (2024)

work page 2024
[58]

Zhuang, F.et al.A Comprehensive Survey on Transfer Learning.Proceedings of the IEEE109, 43–76 (2021)

work page 2021
[59]

Hosna, A.et al.Transfer learning: A friendly introduction.Journal of Big Data 9, 102 (2022)

work page 2022
[60]

Sartipi, K. & Hu, L. BEHAVIOR DRIVEN DESIGN PATTERN RECOVERY. Computer Science(2008)

work page 2008
[61]

Alsobeh, A. M. R. & Clyde, S. Unified Conceptual Model for Joinpoints in Distributed Transactions.ICSEA 14(2014)

work page 2014
[62]

Microservice API Pattern Detection : Using Business Processes and Call Graphs.Tampere University(2022)

Bakhtin, A. Microservice API Pattern Detection : Using Business Processes and Call Graphs.Tampere University(2022)

work page 2022
[63]

A., Cerny, T

Bakhtin, A., Maruf, A. A., Cerny, T. & Taibi, D. Survey on Tools and Techniques Detecting Microservice API Patterns (2022). arXiv:2205.10133

work page arXiv 2022
[64]

AlSobeh, A. M. R. OSM: Leveraging model checking for observing dynamic behaviors in aspect-oriented applications.Online Journal of Communication and Media Technologies13, e202355 (2023)

work page 2023
[65]

AlSobeh, A. M. R. & Magableh, A. A. BlockASP: A Framework for AOP-Based Model Checking Blockchain System.IEEE Access11, 115062–115075 (2023)

work page 2023
[66]

& Risi, M

Lucia, A., Deufemia, V., Gravino, C. & Risi, M. Improving Behavioral Design Pattern Detection through Model Checking.Proceedings of the Euromicro Conference on Software Maintenance and Reengineering, CSMR185 (2010)

work page 2010
[67]

G., Petridis, M

Al-Obeidallah, M. G., Petridis, M. & Kapetanakis, S. A Structural Rule- Based Approach for Design Patterns Recovery.Software Engineering Research, Management and Applications107–124 (2018)

work page 2018
[68]

D., Deufemia, V., Gravino, C

Lucia, A. D., Deufemia, V., Gravino, C. & Risi, M. Design pattern recovery through visual language parsing and source code analysis.Journal of Systems and Software82, 1177–1193 (2009)

work page 2009
[69]

& Antoniol, G

Guéhéneuc, Y.-G. & Antoniol, G. DeMIMA: A Multilayered Approach for Design Pattern Identification.IEEE Transactions on Software Engineering34, 667–684 (2008)

work page 2008
[70]

Comparing Word-Based and AST-Based Models for Design Pat- tern Recognition, PROMISE 2023, 44–48 (Association for Computing Machinery, New York, NY, USA, 2023)

Chand, S.et al. Comparing Word-Based and AST-Based Models for Design Pat- tern Recognition, PROMISE 2023, 44–48 (Association for Computing Machinery, New York, NY, USA, 2023). 38

work page 2023
[71]

H., Sekar, V., Leclercq, E

Ngo, L. H., Sekar, V., Leclercq, E. & Rivalan, J.Exploring code2vec and ASTminer for Python Code Embeddings, 53–57 (2023)

work page 2023
[72]

& Koay, A.Embedding Java Classes with code2vec: Improvements from Variable Obfuscation, 243–253 (2020)

Compton, R., Frank, E., Patros, P. & Koay, A.Embedding Java Classes with code2vec: Improvements from Variable Obfuscation, 243–253 (2020)

work page 2020
[73]

& Vaidyanathan, P

Yoon, B.-J. & Vaidyanathan, P. Context-Sensitive Hidden Markov Models for Modeling Long-Range Dependencies in Symbol Sequences.IEEE Transactions on Signal Processing54, 4169–4184 (2006)

work page 2006
[74]

auth_init

Gamma, E., Johnson, R., Helm, R., Johnson, R. E. & Vlissides, J.Design Patterns: Elements of Reusable Object-Oriented Software(Pearson Deutschland GmbH, 1995). Appendix A Selected Class Implementations for Entropy Method Testing: 1classAuthManager: 2def__init__(self, user_store, logger): 3self.user_store = user_store 4self.session =None 5self._logger = lo...

work page 1995