Dataset Information

Image-based many-language programming language identification.

ABSTRACT: Programming language identification (PLI) is a common need in automatic program comprehension as well as a prerequisite for deeper forms of code understanding. Image-based approaches to PLI have recently emerged and are appealing due to their applicability to code screenshots and programming video tutorials. However, they remain limited to the recognition of a small amount of programming languages (up to 10 languages in the literature). We show that it is possible to perform image-based PLI on a large number of programming languages (up to 149 in our experiments) with high (92%) precision and recall, using convolutional neural networks (CNNs) and transfer learning, starting from readily-available pretrained CNNs. Results were obtained on a large real-world dataset of 300,000 code snippets extracted from popular GitHub repositories. By scrambling specific character classes and comparing identification performances we also show that the characters that contribute the most to the visual recognizability of programming languages are symbols (e.g., punctuation, mathematical operators and parentheses), followed by alphabetic characters, with digits and indentation having a negligible impact.

SUBMITTER: Del Bonifro F

PROVIDER: S-EPMC8592246 | biostudies-literature | 2021

REPOSITORIES: biostudies-literature

ACCESS DATA

Publications

Image-based many-language programming language identification.

Del Bonifro Francesca F Gabbrielli Maurizio M Lategano Antonio A Zacchiroli Stefano S

PeerJ. Computer science 20210723

Programming language identification (PLI) is a common need in automatic program comprehension as well as a prerequisite for deeper forms of code understanding. Image-based approaches to PLI have recently emerged and are appealing due to their applicability to code screenshots and programming video tutorials. However, they remain limited to the recognition of a small amount of programming languages (up to 10 languages in the literature). We show that it is possible to perform image-based PLI on a ...[more]

PMID: 34825053

Similar Datasets

Project description:IntroductionDeep brain stimulation of the internal globus pallidus is an effective treatment for dystonia. However, there is a large variability in clinical outcome with up to 25% non-responders even in highly selected primary dystonia patients. In a large cohort of patients we recently demonstrated that the variable clinical outcomes of pallidal DBS for dystonia may result to a large degree by the exact location and stimulation volume within the pallidal region. Here we test a novel approach of programing based on these insights: we first defined probabilistic maps of anti-dystonic effects by aggregating individual electrode locations and volumes of tissue activated of > 80 patients collected in a multicentre effort. We subsequently modified the algorithms to be able to test all possible stimulation settings of de novo patients in silico based on the expected clinical outcome and thus potentially predict the best possible stimulation parameters for the individual patients.MethodsWithin the framework of a BMBF-funded study, this concept of a computer-based prediction of optimal stimulation parameters for patients with dystonia will be tested in a randomized, controlled crossover study. The main parameter for clinical efficacy and primary endpoint is based on the blinded physician rating of dystonia severity reflected by Clinical Dystonia Rating Scales for both interventions (best clinical settings and model predicted settings) after 4 weeks of continuous stimulation. The primary endpoint is defined as "successful treatment with model predicted settings" (yes or no). The value is "yes" if the motor symptoms with model predicted settings are equal or better (tolerance 5% of absolute difference in percentages) to clinical settings. Secondary endpoints will include measures of quality of life, calculated energy consumption of the neurostimulation system and physician time for programming.PerspectiveWe envision, that computer-guided deep brain stimulation programming in silico might provide optimal stimulation settings for patients with dystonia without the burden of months of programming sessions. The study protocol is designed to evaluate which programming method is more effective in controlling motor symptom severity and improving quality of life in dystonia (best clinical settings and model predicted settings). Trial registration Registered with ClinicalTrials.gov on Oct 27, 2021 (NCT05097001).

Dataset Information

Image-based many-language programming language identification.

Publications

Image-based many-language programming language identification.

Similar Datasets

OmicsDI is part of the ELIXIR infrastructure

Tweets