diff --git a/spacy/lang/id/tag_map.py b/spacy/lang/id/tag_map.py new file mode 100644 index 000000000..93d6cadcc --- /dev/null +++ b/spacy/lang/id/tag_map.py @@ -0,0 +1,31 @@ +# coding: utf8 + +''' +POS explanations for indonesian available from https://www.aclweb.org/anthology/Y12-1014 +''' + +TAG_MAP = { + "NSD" : {POS: NOUN}, + "Z–" : {POS: PUNCT}, + "VSA" : {POS: VERB}, + "CC-" : {POS: NUM}, + "R–" : {POS: ADP}, + "D–" : {POS: ADV}, + "ASP": {POS: ADJ}, + "S–" : {POS: SCONJ}, + "VSP" : {POS: VERB}, + "H–" : {POS: CCONJ}, + "F–" : {POS: X}, + "B–" : {POS: DET}, + "CO-" : {POS: NUM}, + "G–" : {POS: ADV}, + "PS3" : {POS: PRON}, + "W–" : {POS: ADV}, + "O–" : {POS: AUX}, + "PP1" : {POS: PRON}, + "ASS" : {POS: ADJ}, + "PS1" : {POS: PRON}, + "APP" : {POS: ADJ}, + "CD-" : {POS: NUM}, + "VPA" : {POS: VERB}, + "VPP" : {POS: VERB}}