o –hRã@sxdZddlZddlZddlmZddlmZGdd„dƒZedkr:edd d gƒZ ej   ej   e ¡d ¡Ze  e¡dSdS) zß Baseline model for the existing lemmatizer which always predicts "be" and never "have" on the "'s" token. The BaselineModel class can be updated to any arbitrary token and predicton lemma, not just "be" on the "s" token. éN)Úevaluate_sequences)Úload_doc_from_conll_filec@s$eZdZdd„Zdd„Zdd„ZdS)Ú BaselineModelcCs||_||_||_dS©N)Útoken_to_lemmatizeÚprediction_lemmaÚprediction_upos)Úselfrrr©r úh/var/www/html/env_mimamsha/lib/python3.10/site-packages/stanza/models/lemma_classifier/baseline_model.pyÚ__init__s zBaselineModel.__init__cCs||jkr|jSdSr)rr)r Útokenr r r Úpredicts ÿzBaselineModel.predictcCs˜t|ƒ}gg}}|jD]3}gg}}|jD]}|j|jvr4|j|jkr4|j} |j} |  | ¡|  | ¡q|  |¡|  |¡q t ||ƒ\} } } | | fS)aQ Evaluates the baseline model against the test set defined in conll_path. Returns a map where the keys are each class and the values are another map including the precision, recall and f1 scores for that class. Also returns confusion matrix. Keys are gold tags and inner keys are predicted tags ) rÚ sentencesÚwordsÚuposrÚtextrrÚlemmaÚappendr)r Ú conll_pathÚdocÚgold_tag_sequencesÚpred_tag_sequencesÚsentenceÚ gold_tagsÚ pred_tagsÚwordÚpredÚgoldÚmulticlass_resultÚ confusion_mtxÚ weighted_f1r r r Úevaluates      €  zBaselineModel.evaluateN)Ú__name__Ú __module__Ú __qualname__r rr"r r r r r s rÚ__main__z'sÚbeÚAUXzen_gum-ud-train.conllu)Ú__doc__ÚstanzaÚosÚ.stanza.models.lemma_classifier.evaluate_modelsrÚ.stanza.models.lemma_classifier.prepare_datasetrrr#Úbl_modelÚpathÚjoinÚdirnameÚ__file__Ú coNLL_pathr"r r r r Ús  %ü