The analysis tool carries out structural analyses of the sentences introduced in the source text window. If the text consists of several sentences, the corresponding analyses are represented following the sequence of the sentences.
Depending on the options set the target window displays either the result of the syntactic analysis or more abstract and compact representations which are derived from these and focus on the semantically relevant relations and informations. Both types of analysis may be displayed in formal or informal notation. In addition, syntactic representations may be displayed in a format which follows the CoNLL standard.
Formal notation means that the structures are displayed as they are computed by the system. Please contact us if you are interested in APIs for integrating our analysis components in your software systems.
In informal notation the technical terms used for describing the relations between the nodes and the morphological, syntactic and semantic classifications are spelled out in terms of the surface language of the tool-window for better understanding.
The CoNLL option maps the syntactic representations to the structural
format that has been suggested by the Conference on Natural Language
Learning (CoNLL) for its series of shared tasks and beyond this as a
kind of standard for syntactic annotations of corpora
With respect to node descriptions and labels we use our own terms
(as there are variants with respect to this anyway).
Please contact us if you are interested in obtaining annotations for larger texts or corpora.
The syntactic analysis structures are slot grammar representations
of the corresponding sentences. Slot grammar is a unification-based
dependency grammar (cf. McCord 1989,
It is used in Lingenio's Machine Translation systems and, for example,
also in deep parsing of IBM Watson.
The structure displayed in the window represents a slot grammar tree rotated by 90 degrees to the left (such that the top is left and the leaf nodes, the lexical elements, are on the right). The visualization is structured into three columns where the first represents the tree structure as such with the names of the edges connecting the items, the second represents the stems of the words recognized and the pointers to the corresponding lexical entries (complemented by role filling information) and the third column shows the morpho-syntactic annotation of the words.
The representations of this level reduce the slot grammar analyses to the semantically relevant nodes and decorate these by information from the semantic lexicon. In addition they are the result of some structural standardizations which are suitable with respect to deep semantic analysis in later steps of processing. We call these structures slot grammar-based abstract dependency trees: SGaDTs).