ClickHouse/tests/performance/classification.xml

<test>
    <settings>
        <allow_experimental_nlp_functions>1</allow_experimental_nlp_functions>
    </settings>

    <preconditions>
        <table_exists>hits_100m_single</table_exists>
    </preconditions>

    <query>SELECT detectLanguage(SearchPhrase) FROM hits_100m_single LIMIT 10000000 FORMAT Null</query>
    <query>SELECT detectLanguageMixed(SearchPhrase) FROM hits_100m_single LIMIT 10000000 FORMAT Null</query>
    <query>SELECT detectTonality(SearchPhrase) FROM hits_100m_single FORMAT Null</query>

    <!-- Input is not really correct for these functions,
    but at least it gives us some idea about their performance -->
    <query>SELECT detectProgrammingLanguage(SearchPhrase) FROM hits_100m_single LIMIT 10000000 FORMAT Null</query>
    <query>SELECT detectLanguageUnknown(SearchPhrase) FROM hits_100m_single LIMIT 500000 FORMAT Null</query>
    <query>SELECT detectCharset(SearchPhrase) FROM hits_100m_single LIMIT 500000 FORMAT Null</query>

</test>