Пример #1
0
def eval_mst(model_path, test_path, out_prefix, lowercase=True, tagger=None, force=False, result_strem=None):
    mp = MSTParser()

    # -------------------------------------------
    # Use the output prefix to create some new files.
    # -------------------------------------------
    eval_path = out_prefix + '_eval_tagged.txt'
    out_path  = out_prefix + '_out_tagged.txt'


    # -------------------------------------------
    # Rewrite the test file; POS tag the data
    # with the POS tags from our tagger,
    # and strip features.
    # -------------------------------------------
    if not os.path.exists(eval_path) or force:
        LOG.log(1000, "")
        cc = ConllCorpus.read(test_path, lowercase=True)
        if lowercase:
            cc.lower()
        cc.strip_tags()
        cc.strip_feats()
        if tagger is not None:
            LOG.log(1000, "POS Tagging evaluation ")
            cc.tag(StanfordPOSTagger(tagger))
        os.makedirs(os.path.dirname(eval_path), exist_ok=True)
        cc.write(eval_path)
    # -------------------------------------------


    mp.test(model_path, eval_path, out_path)
    eval_conll_paths(test_path, out_path)
Пример #2
0
            eval_path   = ef.get_eval_file(lang)
            out_prefix  = ef.get_out_prefix(lang, aln_method, pos_source)

            if not os.path.exists(eval_path) or not os.path.exists(out_prefix):

                # if USE_CONDOR:
                if False:
                    prefix, name = ef.get_condor_result(aln_method, pos_source, lang)
                    eval_script = os.path.join(intent_dir, 'intent/scripts/eval/dep_parser.py')
                    run_cmd([p3path, eval_script, 'test', '-p', parser_path, '-t', tagger_path, '--test', eval_path,
                             '-o', out_prefix], prefix, name, False, env='PYTHONPATH={}'.format(intent_dir))
                else:
                    # eval_mst(parser_path, eval_path, out_prefix, tagger=tagger_path)
                    out_path = os.path.join(out_prefix + '_out_tagged.txt')
                    if os.path.exists(out_path):
                        ce = eval_conll_paths(eval_path, out_path)
                        de_short.add(lang, aln_method, pos_source, ce.short_ul(), ce.short_ul_count(), ce.short_words())
                        de_long.add(lang, aln_method, pos_source, ce.long_ul(), ce.long_ul_count(), ce.long_words())



if USE_CONDOR:
    condor_wait_notify("Evaluation completed.", email_address, "CONDOR: Evaluation complete.")


de_short.print_stats(ARG_POS_PROJ)
de_long.print_stats(ARG_POS_PROJ)

de_short.print_stats(ARG_POS_CLASS)
de_long.print_stats(ARG_POS_CLASS)