Dependency Grammar Annotator (DGA) is a tool conceived in order to facilitate the syntactic annotation of texts (of a corpus) within the formal framework of Dependency Grammars.
According to EAGLES: "syntactic annotation is the practice of adding syntactic information to a corpus by incorporating into the text indicators of the syntactic structure: eg. labeled bracketing, or symbols indicating dependency relations between words". Although very useful in practice (testing various grammatical theories, automatic acquisition of grammars etc.), such corpora are very costly since the operation of syntactic annotation is a great consumer of time and effort on the part of those performing it. DGA has been designed in order to minimize the human effort necessary during the process of corpus creation.
DGA is an easy to use graphical interface which allows the efficient creation and manipulation of syntactic structures. Since the used formalism is that of Dependency Grammars, the syntactic structures in this case consist of the dependency relations formed within the words of a sentence, labeled with the corresponding parts of speech, and of the grammatical relations existing among these words. Traditionally, the dependency relations are indicated by means of arches which link the dependent word to that word which it determines, these arches being labeled with the name of the relation existing between the linked words. Such a graphical representation (being in conformity with the EAGLES recommendations) is used by DGA as the base of the annotation operation.
During the entire annotation process the user acts directly upon this graphical representation. As a consequence, besides the advantage of convenient usage, the accuracy of annotation increases, since the user receives an immediate graphical feedback regarding any changes performed in the syntactic structure. Operating upon the syntactic structure is extremely easy and intuitive: in order to create a dependency relation only two mouse clicks are necessary (on the two words involved in the relation which is being created). For labeling a word with the corresponding part of speech or for establishing the type of a dependency relation only one mouse click, followed by the selection of the label from an appropriate list, is needed. DGA therefore allows a quick text annotation.
In our view DGA responds to the requirements which Marcus et al. have identified as being significant regarding the annotation process:
Accuracy - since one operates directly with the graphical representation and because of the immediate graphical feedback which DGA offers to the user.
Speed - creation and handling of the dependency relations are very quick operations (by means of the mouse).
Consistency - the sets of parts of speech and dependency relations are established by the user after which, in order for them to be used, it is only necessary to select them from various lists.
|
|
| >Dependency Grammar Annotator |
