exemple.xml

<?xml version="1.0" encoding="UTF-8"?>
<?xml-model href="protogenie/schema.rng" schematypens="http://relaxng.org/ns/structure/1.0"?>
<config>
    <output column_marker="TAB">
        <header>
            <key></key>
        </header>
    </output>
    <memory path="memory.csv" />
    <default-header>
        <header type="order">
            <key map-to="token">0</key>
            <key map-to="lemma">1</key>
            <key map-to="POS">2</key>
            <key map-to="morph">3</key>
        </header>
    </default-header>
    <corpora>
        <corpus path="datasets/chrestien.tsv" column_marker="TAB">
            <splitter name="token_window" window="20"/>
            <header type="default" />
        </corpus>
        <corpus path="datasets/dotmarkers.tsv" column_marker="TAB">
            <splitter name="punctuation" sentence_markers=";?!."/>
            <header type="default" />
        </corpus>
        <corpus path="datasets/empty_line.tsv" column_marker="TAB">
            <splitter name="empty_line"/>
            <header type="default" />
        </corpus>
        <corpus path="datasets/flow.tsv" column_marker="TAB">
            <splitter name="file_split"/>
            <header type="default" />
        </corpus>
    </corpora>
</config>