Pfennigschmidt, Bernhard · Figshare · 2025
We present the BioCube framework, a systematic computational approach for organizing and analyzing the genetic code through three-dimensional coordinate addressing. Agnostic to amino acid properties, using quaternary-to-decimal conversion (CA = 4×first + 16×middle + 1×third), all 64 codons map to coordinates 0-63, creating a 4×4×4 structural organization. This framework reveals systematic patterns in amino acid distribution and enables quantitative analysis of mutational relationships. Preliminary validation shows pathogenic variants exhibit significantly higher coordinate address changes (ΔCA = 17.3) compared to benign variants (ΔCA = 8.1), suggesting potential utility for computational variant analysis. The systematic organization facilitates pattern recognition, distance calculations, and provides a consistent mathematical foundation for genetic code research applications.
Cited by 0Open access
Enna Basic, Alberto Giaretta · ArXiv.org · 2024
Large Language Models (LLMs) have emerged as powerful tools for automating programming tasks, including security-related ones. However, they can also introduce vulnerabilities during code generation, fail to detect existing vulnerabilities, or report nonexistent ones. This systematic literature review investigates the security benefits and drawbacks of using LLMs for code-related tasks. In particular, it focuses on the types of vulnerabilities introduced by LLMs when generating code. Moreover, it analyzes the capabilities of LLMs to detect and fix vulnerabilities, and examines how prompting strategies impact these tasks. Finally, it examines how data poisoning attacks impact LLMs performance in the aforementioned tasks.
Cited by 1Open access
T. Aaron Gulliver, V.K. Bhargava · IEEE Transactions on Computers · 1993
A double error correcting systematic (16,8) quasi-cycle (QC) code that can detect all triple-adjacent errors within each 8-b byte is presented. This code is useful in computer memory applications where adjacent errors are more likely than random errors. As an alternative, a systematic (24,16) QC code is given which corrects single errors, and detects all double errors in addition to all triple-adjacent and quadruple-adjacent errors. A third code, a systematic (32,16) QC code, is presented which corrects all triple errors and detects all quadruple errors. Decoding can be achieved either by syndrome generation and table lookup or by majority logic.< <ETX xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">></ETX>
Cited by 21
Anonymous, Anonymous · Zenodo (CERN European Organization for Nuclear Research) · 2025
Anonymized review protocol for the paper: Evaluating Code Recommender Systems: A Review.
Cited by 0Open access
Lummerzheim, Camilla · Zenodo (CERN European Organization for Nuclear Research) · 2025
This software includes the code for the figures and visualizations presented in "Quality Characteristics for Software in HPC Environments: A Systematic Literature Review".
Cited by 0Open access
Anna Uzonyi, Ronit Nir, Ofir Shliefer et al. · Molecular Cell · 2021
Cited by 41Open access
Fábio do Rosario Santos, Ricardo Choren Noya · Information and Software Technology · 2025
Detecting code smells using Machine Learning presents inherent challenges due to the unbalanced nature of the problem and susceptibility to interpretation biases. It is a data-driven process for code quality assurance that aims to detect if a given piece of code presents a fundamental design principles violation that negatively impacts design quality. Researchers in the field have been advised to carefully analyze the internal mechanisms of forecasting models before interpreting the results generated by them. The review aims to summarize and synthesize studies that utilized Data Preprocessing techniques for Machine Learning-based code smell detection. And also, to investigate the relationship between Data Preprocessing and more advanced Machine Learning techniques , i.e., Ensemble Methods , Deep Learning , and Transfer Learning . To obtain insights into Data Preprocessing for Machine Learning-based code smell detection solutions, we employed a systematic approach, identifying and analyzing 69 studies published up to November 2023. In Data Preprocessing, Data Balancing techniques, Feature Selection techniques, and Filtering emerged as prominent strategies. SMOTE was the most frequently used Data Balancing technique, while Autoencoder , Chi-square, Gain Ratio, Information Gain , PCA , and CFS were notable choices for Feature Selection. Tokenization and Syntax Trees were commonly paired with Deep Learning or Transfer Learning methods. Normalization and Standardization were implemented for Data Scaling. Regarding Machine Learning techniques used with Data Preprocessing, 46% of the combinations occurred with at least one Ensemble Method. Deep Learning was employed in 37% of cases. Data Balancing techniques combined with Deep Learning (32%) or Ensemble Methods (19%) were used most. The findings of this SLR are an integrated and comprehensive source of information regarding data preparation practices, challenges, and solutions for Machine Learning-based code smell detection, emphasizing the continuous endeavor towards more resilient, contextually sensitive, and developer-informed strategies within this dynamic field. • Custom Data Balancing and Feature Selection enhance code smell detection. • Ensemble Methods boost accuracy but face complexity and resource challenges. • Capturing code semantics with DL and TL is promising in cross-language scenarios. • Balancing accuracy, scalability, and semantic understanding is key to robustness.
Cited by 2Open access
Rokon, Md. Omar Faruk · eScholarship (California Digital Library) · 2022
How can we find malware source code and establish the similarity, influence, and phylogeny of these malware? This question is motivated by a real need: there is a dearth of malware source code, which impedes various types of security research. Our work is driven by the following insight: public archives, like GitHub, have a surprising number of malware repositories. This thesis spans three interrelated problems in this space. First, we address the problem of scarcity of malware source code. We propose, SourceFinder, a supervised-learning approach to identify repositories of malware source code efficiently. We evaluate and apply our approach using 97K repositories from GitHub. Second, we propose Repo2Vec, a comprehensive embedding approach to represent a repository as a distributed vector by combining features from three types of information sources. As our key novelty, we consider three types of information: (a)metadata, (b) the structure of the repository, and (c) the source code. It enables ML techniques for similarity identification, clustering, and classification tasks. We evaluate our approach with 1013 java repositories to find similarities and clusters among them. Third, we propose PIMan, a systematic approach to quantify the influence among the repositories in a software archive by focusing on the social level interactions. We introduce the concept of Plausible Influence which considers three types of information: (a) repository level interactions, (b) author level interactions, and (c) temporal considerations. We evaluate and apply our method using 2089 malware repositories from GitHub spanning approximately 12 years. In our thesis, we use the data from GitHub and security forums. We show that our approach, SourceFinder identifies malware repositories with 89% precision and 86% recall using a labeled dataset. We use SourceFinder to identify 7504 malware source code repositories, which arguably constitute the largest malware source code database. Second, we show that our method outperforms previous methods in terms of precision (93%vs 78%), with nearly twice as many Strongly Similar repositories and 30% fewer False Positives. We show how Repo2Vec provides a solid basis for: (a) distinguishing between malware and benign repositories, and (b) identifying a meaningful hierarchical clustering. For example, we achieve 98% precision and 96% recall in distinguishing malware and benign repositories. We study the social level interaction between two repositories and establish a plausible influence network among them. We find that there is a significant collaboration and influence among the repositories in our dataset. We argue that our approach and our large repository of malware source code can be a catalyst for research studies, which are currently not possible.
Cited by 0Open access
Robin King, Barbara Hooper, Wendy Wood · Medical Teacher · 2011
Bibliographic tools can contribute to make the methods that researchers adopt for the phase following the initial literature search more transparent and systematic. The illustrative case description involves a systematic review of academic education in occupational therapy, but easily generalizes to systematic reviews in other health science professions.
Cited by 47
Fengji Zhang, Zexian Zhang, Jacky Keung et al. · Journal of Systems and Software · 2024
Cited by 10
Umama, Kamaluddeen Usman Danyaro, Maged Nasser et al. · IEEE Access · 2025
The rapid emergence of Large Language Models (LLMs) has significantly advanced the field of code generation, sparking growing research interest across both academia and industry. While existing reviews provide foundational insights into LLM-based code generation, they do not provide a comprehensive analysis of recent developments or adequately address the dual perspective of technical and demographic insights. This Systematic Literature Review (SLR) bridges critical gap in existing work by systematically analysing both technical and demographic dimensions of LLM-based code generation. Following a PRISMA-guided methodology 58 studies published between 2020 and 2025 were selected from four major digital libraries. Key findings reveal a sharp rise in evaluation studies and frameworkdriven contributions. Notably, HumanEval, MBPP and APPS emerged as dominant benchmarks. While functional-correctness metrics particularly pass@k are the most frequently adopted metrics. Moreover, the review uncovers a previously unexamined demographic landscape, highlighting, geographic distribution, institutional collaboration, team size, publishing trends, venue types, and the top 10 highly cited studies from the selected studies. Significantly, a high proportion of selected studies were published in conferences. Several persistent challenges found in existing reviews that hinder the reliability and applicability of LLM-based code generation. These include hallucinations, limited generalizability, security vulnerabilities, and a lack of interpretability. These limitations constrain the robustness, trustworthiness, and scalability of current systems. To address these issues, this reviewoffers actionable insights and strategic guidelines to guide future research, evaluation strategies, and practical implementations in LLM-based code generation.
Cited by 1Open access
Anonymous · Zenodo (CERN European Organization for Nuclear Research) · 2026
Cited by 0Open access
Mitchell T. Foster, Ailith Ewing, Margaret C. Frame et al. · Neuro-Oncology · 2023
Abstract AIMS Glioblastoma (GBM) is a heterogeneous and aggressive brain tumour that is invariably fatal despite maximal treatment. GBM-relevant ‘biomarkers’ (such as known driver-mutations or cell-states) could be used to stratify patients for treatments. However, pairing biomarkers with appropriate therapeutic ‘targets’ is challenging. If a gene is more essential in GBM cells which contain such a biomarker, inhibiting the gene product may be therapeutically useful in GBM tumours where that biomarker is present. METHOD We describe a novel computational tool which distils new translational insights from existing disparate large- scale biological data (genomics, transcriptomics and functional genomics data from 60 distinct patient-derived, IDH-wildtype, adult GBM cell lines) using Targeted Learning (a subfield of mathematics and causal machine learning) to systematically identify context-dependent gene essentiality in Glioblastoma (GBM-CoDE). RESULTS We derived multiple target-biomarker pair hypotheses, two of which (WWTR1 is more essential in EGFR- mutated/amplified GBM, and VRK1 is more essential in VRK2-methylated GBM) have recently been published by external groups, implying that our additional novel findings may be valid. We developed an interactive plat- form to make our findings accessible to other researchers to prioritise pairs for experimental validation and subsequent translation. CONCLUSIONS GBM-CoDE represents a potent new approach to solving the problem of developing patient-tailored therapeutics in GBM, made possible with recent innovations in computational biology. While the majority of these target- biomarker hypotheses have yet to be experimentally validated, our results may in time accelerate effective translation to biomarker stratified clinical trials in GBM. Our method is translatable to other cancers of unmet need.
Cited by 0Open access
Delano Oliveira, Reydne Santos, Fernanda Madeiral et al. · Greater South Information System · 2023
Context: Software programs can be written in different but functionally equivalent ways. Even though previous research has compared specific formatting elements to find out which alternatives affect code legibility, seeing the bigger picture of what makes code more or less legible is challenging. Goal: We aim to find which formatting elements have been investigated in empirical studies and which alternatives were found to be more legible for human subjects. Method: We conducted a systematic literature review and identified 15 papers containing human-centric studies that directly compared alternative formatting elements. We analyzed and organized these formatting elements using a card-sorting method. Results: We identified 13 formatting elements (e.g., indentation) and 33 levels of formatting elements (e.g., two-space indentation), which are about formatting styles, spacing, block delimiters, long or complex code lines, and word boundary styles. While some levels were found to be statistically better than other equivalent ones in terms of code legibility, e.g., appropriate use of indentation with blocks, others were not, e.g., formatting layout. For identifier style, we found divergent results, where one study found a significant difference in favor of camel case, while another study found a positive result in favor of snake case. Conclusion: The number of identified papers, some of which are outdated, and the many null and contradictory results emphasize the relative lack of work in this area and underline the importance of more research. There is much to be understood about how formatting elements influence code legibility before the creation of guidelines and automated aids to help developers make their code more legible.
Cited by 0Open access
Maen Alnuzha, Tanya Bloch · Journal of Information Technology in Construction · 2025
Building design must adhere to numerous codes, laws, and regulations. In practice, despite the available tools for Automated Code Checking (ACC) and the extensive research in the field, checking whether a given design complies with all relevant regulatory requirements remains a manual and time-consuming task. A comprehensive checking system that provides highly automated solutions for a wide range of regulations remains a distant goal. Recent studies have underscored the potential of engaging advanced technologies like Machine Learning (ML) and Natural Language Processing (NLP) to enhance ACC abilities. Hence, this work aims to review recent advancements in implementing ML technologies in code compliance checking, identify knowledge gaps, and suggest future research directions. By implementing a systematic literature review methodology, we identify three key research areas within the domain: processing regulatory text, processing design information, and an overall checking mechanism. Existing efforts explore each of these areas using a variety of ML algorithms to enhance their effectiveness. Despite the notable advancements, challenges persist due to the complexity of regulations, ambiguity of legal texts, and the scarcity of training data, all of which limit the scalability of the presented approaches. Additionally, while ML enhances flexibility by learning from data rather than relying on hardcoded rules, it introduces uncertainties in decision-making processes critical to building permitting. The review highlights the potential for hybrid approaches that combine the strengths of both rule-based systems and ML models to effectively address these challenges.
Cited by 8Open access
Anonymous · Zenodo (CERN European Organization for Nuclear Research) · 2026
Cited by 0Open access
Shivani Shukla, Himanshu Joshi, Romilla Syed · 2025
The rapid adoption of Large Language Models (LLMs) for code generation has transformed software development, yet little attention has been given to how security vulnerabilities evolve through iterative LLM feedback. This paper analyzes security degradation in AI-generated code through a controlled experiment with 400 code samples across 40 rounds of "improvements" using four distinct prompting strategies. Our findings show a 37.6% increase in critical vulnerabilities after just five iterations, with distinct vulnerability patterns emerging across different prompting approaches. This evidence challenges the assumption that iterative LLM refinement improves code security and highlights the essential role of human expertise in the loop. We propose practical guidelines for developers to mitigate these risks, emphasizing the need for robust human validation between LLM iterations to prevent the paradoxical introduction of new security issues during supposedly beneficial code "improvements."
Cited by 1
Shivani Shukla, Himanshu Joshi, Romilla Syed · ArXiv.org · 2025
The rapid adoption of Large Language Models(LLMs) for code generation has transformed software development, yet little attention has been given to how security vulnerabilities evolve through iterative LLM feedback. This paper analyzes security degradation in AI-generated code through a controlled experiment with 400 code samples across 40 rounds of "improvements" using four distinct prompting strategies. Our findings show a 37.6% increase in critical vulnerabilities after just five iterations, with distinct vulnerability patterns emerging across different prompting approaches. This evidence challenges the assumption that iterative LLM refinement improves code security and highlights the essential role of human expertise in the loop. We propose practical guidelines for developers to mitigate these risks, emphasizing the need for robust human validation between LLM iterations to prevent the paradoxical introduction of new security issues during supposedly beneficial code "improvements".
Cited by 1Open access
Chen-Chen Qian, Shengmei Zhao, Qianping Mao · 2016
The transmission distance of continuous variable QKD (CV QKD) protocol based on Gaussian modulation has been quite limited by the low efficient reconciliation protocol. On the other hand, noiseless linear amplifier (NLA) have greatly enhanced the ability of CV QKD. In this paper, we propose a Gaussian post-selection method simulating NAL in CV QKD, followed by a reconciliation process based on systematic Polar code. We distill the reliable data through the post-selection filter as the raw key and then encode them with systematic polar code. The numerical simulation show that the secure transmission distance can be enlarged more than thirty kilometers while keeping the advantages of Gaussian security evidences. The key error rate can close to zero when signal-to-noise ratio (SNR) is near 3 dB.
Cited by 1
Weaving, Dan, Rennie, Gordon, Chesson, Lucy et al. · OSF Preprints (OSF Preprints) · 2022
This review presents the existing research examining the effects of rule changes within football-code sports and summarises existing research in this area.
Cited by 0Open access