diff --git a/.gitignore b/.gitignore index 404e2f2..c203e4b 100644 --- a/.gitignore +++ b/.gitignore @@ -5,3 +5,4 @@ __pycache__/ results/ data/ config.ini +configs/ diff --git a/README.md b/README.md index f5f6276..bc70735 100644 --- a/README.md +++ b/README.md @@ -24,5 +24,5 @@ python3 dependency-parsetree.py --config_file= Example: ```bash -python3 dependency-parsetree.py --config_file=config_template.ini +python3 dependency-parsetree.py --config_file=official_config.ini ``` diff --git a/config_template.ini b/config_template.ini deleted file mode 100644 index 0ff7abc..0000000 --- a/config_template.ini +++ /dev/null @@ -1,23 +0,0 @@ -[settings] -input = /media/luka/Portable Disk/Datasets/dependency_treeparse/ssj500k.conllu/sl_ssj-ud_v2.4.conllu -output = results/out.tsv -internal_saves = ./internal_saves -;ngrams = 0 -;tree_size = 3 -;lines_threshold = 10000 -;frequency_threshold = 1 -association_measures = no -print_root = no -node_order = free -tree_type = all -dependency_type = unlabeled -nodes_number = yes -;label_whitelist = nsubj|obj|obl -;root_whitelist = lemma=leto -;root_whitelist = upos=NOUN&Case=Nom|upos=ADJ&Degree=Sup -;root_whitelist = lemma=mati&Case=Acc|lemma=lep&Degree=Sup -; analyze_type options: form, deprel, lemma, upos, xpos -;node_type = form -cpu_cores = 12 -node_type = form -query = upos=NOUN > _ diff --git a/official_config.ini b/official_config.ini new file mode 100644 index 0000000..9df06cc --- /dev/null +++ b/official_config.ini @@ -0,0 +1,28 @@ +[settings] + +;___GENERAL SETTINGS___ +input = data/sl_ssj-ud_v2.4.conllu +output = results/out_official.tsv +internal_saves = ./internal_saves +cpu_cores = 12 + +;___TREE SPECIFICATIONS___ +tree_size = 2-4 +tree_type = complete +dependency_type = labeled +node_order = free +node_type = upos + +;___TREE RESTRICTIONS___ +;label_whitelist = nsubj|obj|obl +;root_whitelist = lemma=mati&Case=Acc|lemma=lep&Degree=Sup + +;___SEARCH BY QUERY___ +;query = _ > _ + +;___OUTPUT SETTINGS___ +;lines_threshold = 10000 +;frequency_threshold = 0 +association_measures = no +print_root = yes +nodes_number = yes