Replication package for A Unified Cross-Language Intermediate Representation for Infrastructure-as-Code Smell Detection
Year: 2026
Type: conference-paper
Abstract: General Description: This release includes the four IaC full dataset, the four IaC oracle dataset, the code for IRIS described in the paper, and the prompt used for the zero-shot LLM evaluation. All the results are described in our paper. All required artifacts are attached to this release. Users can download the project and run the main.py in the Code folder to test IaCIR on the IaC dataset. Users should read the README.md files in each folder to understand the purpose of each file within the folder. Note for the Tables & Figures: The complete IaC template for Figure 1 can be found in the "figure" folder of this package. For Tables 5 and 6 in the paper, the data in each cell represents [True Positive Count (Precision, Recall)]. For instance, 2 (1.00, 0.50) indicates that two true positive cases are being detected with a precision of 1.00 and a recall of 0.50. Package Structure: You can check the datasets under the "data" folder. You can check the code for IRIS under the "code" folder. You can check the complete motivation example in the "figure" folder. All our experimental results are publicly available on this page. For simplicity, we provide the following instructions for our artifacts: RQ1_ontology [code/evaluation/ontology/] - stores the ontology, IR construct set, and mapping result. RQ1_efficiency [code/evaluation/result/] - stores the evaluation result CSV files, which "parse_time" column shows how long the IR conversation takes. RQ2_oracle_dataset [data/oracle/] - stores the oracle dataset and the ground truth for the four IaC languages. RQ2_static_analysis_tools [code/evaluation/static_analysis_tools/] - stores the result for the static analysis tool evaluation on the oracle dataset. Please note the data used in RQ2 stores i RQ2_IRIS [data/oracle/result/] - stores the result for the IRIS run on oracle dataset. RQ2_opus4.6 [code/llm/llm_eval_outputs_xxx/] - stores the LLM results for each IaC language, where xxx is the placeholder for the IaC language name. RQ3_dataset [data/xxx_collected_template_new/] - stores the dataset for each IaC language, where xxx is the placeholder for the IaC language name. RQ3_result [code/evaluation/result/] - stores the smell detection results by IRIS on the full dataset for each IaC language (data/xxx_collected_template_new). Please note that the data used in RQ1_efficiency and RQ2_IRIS are taken from these files.
Author: Anonymous
Language: English
Subfield: Information Systems
Field: Computer Science
Domain: Physical Sciences
Open Access status: green
OpenAlex ID: W7141634991
FWCI:
Cites:
Cited by:
Related to: