-
Notifications
You must be signed in to change notification settings - Fork 9
DEP
Welcome to the wiki page about the Greek DEP Tagger.
First of all, what Dependency Parsing is?
A definition from this website:
Syntactic Parsing or Dependency Parsing is the task of recognizing a sentence and assigning a syntactic structure to it. The most widely used syntactic structure is the parse tree which can be generated using some parsing algorithms. These parse trees are useful in various applications like grammar checking or more importantly it plays a critical role in the semantic analysis stage.
Greek DEP tagger was trained on this dataset.
An interesting module regarding DEP tagging that provides visualization using NLTK can be found here.
The greek models support the following DEP tags:
-
nsubj: nominal subject.
Sentence: Clinton defeated Dole. Explanation: Clinton is the nominal subject of the verb defeated.
-
obj: object.
Sentence: She gave me a raise. Explanation: raise is the object of the verb gave.
-
iobj: indirect object
Sentence: She gave me a raise. Explanation: Me is the indirect object of the verb gave, while raise is the direct object.
-
csubj: clausal subject
Sentence: What she said makes sense. Explanation: said is the clausal subject of the verb makes.
-
ccomp: clausal complement
Sentence: He says that you like to swim. Explanation: says is the ccomp of the word like.
-
xcomp: open clausal complement
-
aux: auxiliary
Sentence: She has died. Explanation: has is the auxiliary to the token died.
-
cop: copula
Sentence: Bill is honest. Explanation: is is the cop of honest.
-
mark: marker. It is linker between a finite and an infinite clause.
Sentence: Forces engaged in fighting after insurgents attacked. Explanation: after is the marker of the verb attacked.
-
det: determiner.
Sentence: The man is here. Explanation: The is the det of man.
-
case.
-
punct: punctuation.
-
root. The root of the sentence. Usually, the verb, if exists.
-
orphan
Sentence: Marie won gold and Peter bronze. Explanation: Peter is the orphan of bronze.
-
parataxis: The parataxis relation (from Greek for “place side by side”) is a relation between the main verb of a clause and other sentential elements, such as a sentential parenthetical, two sentences placed side by side without any explicit coordination or subordination, or a clause after a “:” or a “·” (άνω τελεία), in case a sentence splitter has recognized two different sentences.
Sentence: «Το αεροσκάφος» ανέφερε ο πρωθυπουργός «δεν εξέπεμψε σήμα κινδύνου». Ανέφερε is parataxis to εξέπεμψε.
-
amod: adjectival modifier.
Sentence: He ate red meat. Explanation: red is amod to meat.
-
acl/acl-relcl: clausal modifier of noun.
Sentence: το γεγονός ότι έφυγε. Sentence: η είδηση πως νίκησε. Sentence: η ίδεα ότι έχασες. Explanation: acl relationship between γεγονός - έφυγε.
-
advlc: adverbial clause modifier. It seems like the 17 dep but it is a relationship between verbs - verbs / verbs - adverbs.
Sentence: Το ατύχημα συνέβη καθώς έπεφτε η νύχτα. Sentence: έφυγε θυμωμένος Explanation: advcl relationship between συνέβη - έπεφτε Explanation: advcl relationship between έφυγε - θυμωμένος
-
nmod: dependency between nouns
Sentence: Η καρέκλα του γραφείου. Sentence: Το θρανίο της αίθουσας.
-
appos: appositional modifier (explanatory)
Sentence: Sam, my brother, arrived. Sentence: Bill (John's cousin) arrived.
-
nummod: numeric modifier
Sentence: He ate 3 sheeps. Sentence: Έδωσα 40 δολάρια.
-
advmod: adverbial modifier.
Sentence: genetically modified food.
I guess that this list of supported DEP tags is a bit difficult to memorize.
The DEP tags that you may need to remember because they are widely used are the following:
nsubj: υποκείμενο (Ο Γιάννης έφαγε.)
obj: αντικείμενο (πήρε το στυλό.)
amod: adjective modifier (κόκκινο δάνειο.)
advmod: adverb modifier (τροποποιημένο κρέας.)
nmod: noun modifier (βάρος παραγωγής.)
Probably, the most common usage of DEP in spaCy is the noun chunks extraction task.
You can find the related module here.
For visualization of DEP tags check also our demo: https://nlpbuddy.io