% ============================================================ % Placement: Main Text (Section 2 - Taxonomy) % ============================================================ \begin{table*}[t] \centering \small \begin{tabular}{lc lll} \toprule \textbf{Category} & \textbf{Lint Alerts} & \textbf{C++ (clang-tidy)} & \textbf{Python (Pylint)} & \textbf{Java (PMD)} \\ \midrule Error-prone & 18,434 & \texttt{bugprone-*} & \texttt{error} & \texttt{Error Prone}, \texttt{Security} \\ Performance & 17,776 & \texttt{performance-*} & \texttt{performance}& \texttt{Performance}, \texttt{Multithreading} \\ Refactor & 17,977 & \texttt{modernize-*} & \texttt{refactor} & \texttt{Design} \\ Style & 115,458 & \texttt{readability-*} & \texttt{convention} & \texttt{Code Style}, \texttt{Best Practices} \\ Other & 2,017 & \texttt{other} & \texttt{other} & \texttt{other} \\ \bottomrule \end{tabular} \caption{Cross-language taxonomy of lint alert categories in LintEval. Counts show the number of alerts in each diagnostic category. Representative rule families illustrate how each category maps to language-specific linter conventions.} \label{tab:cross_language_mapping} \end{table*}