Abstract dependency trees

An abstract dependency is a directed acyclic graph that models the grammatical relations between lexical items and categories built from lexical items. The generator creates realizations for abstract dependency trees that describe at least one possible sentence. An abstract dependency tree can consist of four node types:

Edges between nodes have a dependency label, where hd indicates the head of a grammatical relation. Figure [dtree] shows a dependency tree for the sentence "De boeken kosten ons een klein fortuin". For interior nodes, the dependency label is shown as the first element, and the category as the second.

Possible dependency labels are listed in [deplabels]. In the following sections, the node types will be described in more detail.

All nodes are associated with a category. All possible categories are listed in table [categories]. Note that this is different from Alpino dependency structures, in which only interior nodes are associated with a category.

Lexical nodes are leaf nodes that provide an abstracted representation for (surface) words. As a minimum a lexical node should specify:

Part of speech tags and possible attributes are discussed in more detail below.

Indexed nodes are nodes associated with a specific index so that the node can be referred to by a reference node. For instance, in the sentence Ik heb de trein gemist both ik-heb and ik-gemist have a subject-head relation, while gemist is the head of a vc of heb. To allow for such representations, co-indexation is required.

Reference nodes only have an index, and no additional content nor sub-structure. The additional content can be found at the Index node with the same index.

The current assumption is, that no punctuation is specified in the ADT. The generator will add a minimum amount of obligatory punctuation for a given ADT.

To allow for an abstract description of verbs that have a separable particle, it is allowed to omit particles with the svp relation in an ADT. If such particles are not included as nodes in the ADT, the generator will attempt both to generate sentences with a separate particle as in ik bel hem op and a sentences in which the particle is not separated as in omdat ik hem opbel.

An ADT can be stored as a Prolog term that consists of recursive tree terms. The basic format is:

Where Relation is a relation term, and Daughters is a list of daughter nodes, or the empty list for leaf nodes. A relation has the following form:

Type indicates the type of relation, such as su, obj1, or mod. The label comes in four types: interior nodes, leaf nodes, index nodes and reference nodes. An interior node uses a p/1 term, for instance,

is a node of the category ppart with a vc relation.

Lexical nodes use a adt_lex/5 term as their label:

Here the category, root, sense, POS tag, and attributes of a lexical item are noted. The sense of a lexical item can contain additional information about a lexical item to select for a specific reading. For instance, for words with fixed parts, the fixed parts are often listed in the sense. E.g. the sense of rood aanlopen is rood-loop_aan. The sense can be omitted by replacing it with a variable (e.g., _).

This is an example of a lexical node:

This describes a noun with the root trein and an attribute (rnum=sg) with the head (hd) relation.

An index node is represented as:

where Label is an interior node or a leaf node.

Reference nodes are represented by

For instance, we can refer to

with:

Combined, we can construct ADTs as Prolog terms for all grammatical sentences. E.g., the ADT term for the sample discussed above is:

ADTs can also be represented as XML documents to allow for easy querying and manipulation outside the Alpino environment.

The dependency tree is represented in XML as a recursive structure of node elements. Each node has an identifier (id) and relation (rel). Category nodes have a cat attribute that specifies the category. For instance:

Every lexical node has the root, sense, and postag attributes for respectively describing the word root, sense, and part of speech tag. E.g.:

Lexical nodes have other attributes, as described in the previous sections. For instace, here the num and gen attributes are also listed for the noun number and gender.

A node can be co-indexed by adding the index attribute to a lexical node:

The referring node also contains an index attribute, but no other information specific to a lexical node. For example:

As an example of a full ADT, we include an ADT for the same sentence used in the Prolog ADT:

The Alpino option end_hook=adt_prolog will generate an ADT in Prolog format for every parse.

Likewise, the Alpino option end_hook=adt_xml will generate an ADT in XML format.

As a special case, the option end_hook=gen_suite(best_score) can be used to generate an ADT for the parse which resembles most the gold standard parse for the input sentence (obviously this assumes that we are parsing sentences for which the gold standard parse is available). The ADT uses the Prolog format in this case.

This is currently not functioning properly. Perhaps this is not such a good idea anyway.