def get_coref_infos()

in modeling/coval/arrau/reader.py [0:0]


def get_coref_infos(key_directory,
        sys_directory,
        keep_singletons,
        keep_non_referring,
        use_MIN,
        print_debug=False):

    key_docs = get_all_docs(key_directory)
    sys_docs = get_all_docs(sys_directory)

    doc_coref_infos = {}
    doc_non_referrig_infos = {}

    for doc in key_docs:

        if doc not in sys_docs:
            print('The document ', doc,
                    ' does not exist in the system output.')
            continue

        key_clusters = get_doc_markables(doc, key_docs[doc], use_MIN)
        sys_clusters = get_doc_markables(doc, sys_docs[doc], False)

        (key_clusters, key_non_referrings, key_removed_non_referring,
                key_removed_singletons) = process_clusters(
                key_clusters, keep_singletons, keep_non_referring)
        (sys_clusters, sys_non_referrings, sys_removed_non_referring,
                sys_removed_singletons) = process_clusters(
                sys_clusters, keep_singletons, keep_non_referring)

        sys_mention_key_cluster = get_markable_assignments(
                sys_clusters, key_clusters)
        key_mention_sys_cluster = get_markable_assignments(
                key_clusters, sys_clusters)

        doc_coref_infos[doc] = (key_clusters, sys_clusters,
                    key_mention_sys_cluster, sys_mention_key_cluster)
        doc_non_referrig_infos[doc] = (key_non_referrings, sys_non_referrings)

        if print_debug and not keep_non_referring:
            print('%s and %s non-referring markables are removed from the '
                    'evaluations of the key and system files, respectively.'
                    % (key_removed_non_referring, sys_removed_non_referring))

        if print_debug and not keep_singletons:
            print('%s and %s singletons are removed from the evaluations of '
                    'the key and system files, respectively.'
                    % (key_removed_singletons, sys_removed_singletons))

    return doc_coref_infos, doc_non_referrig_infos