Basic Patterns

Given a rule, a pattern of PoS tags is a sequence of tags used to identify a specific dependency. A Pattern must fill the following requirements:

• It must contain, at least, those tags that are involved in the dependency: both the head and the dependent. Complex dependencies (HeadRelDep, DepRelHead, etc.) also need a third element: the relator.
• It may contain contextual tags, which will be enclosed in square brackets: [ ].
• Any contextual tag can be tunned with standard wildcards representing optionality, iteration, etc., that is, well known operators used by languages based on regular expressions.

The following examples are Patterns that fill the requirements of DepPattern:

ADJ NOUN

[DT] ADJ NOUN

DT [X]* NOUN

VERB [ADV]* [DT]* [ADJ]* NOUN

VERB [DT]+ NOUN

VERB [DT]? NOUN

NOUN PRP [DT]* [ADJ]* NOUN

- -[DT] ADV VERB

VERB NOUN -[PRP]

The first pattern describes an adjective immediately followed by a noun. Both tags are involved in a simple dependency. The second pattern represents the same situation, but in this case there is a contextual determiner which is not involved in the dependency. The third pattern stands for a simple dependency constituted by a determiner followed by a noun and, optionally, by an unlimited number of different tags between them. The inserted tags are not involved in the dependency: they build the context. Tag X is a shortcut defined in the configuration file (see Section 1.3). The fourth pattern represents a simple dependency between a verb and a noun with three optional tags beween them building the context. The context is constituted by 0 or more adverbs, derterminers and adjectives. The fifth pattern represents the same simple dependency but, in this case, the context is not optional: there must be one or more determiners between the verb and the noun. The sixth pattern is similar to the previous one. The difference is that the contextual determiner is optional: there must be 0 or 1 determiner. The seventh pattern represents a complex syntactic dependency between three elements: a noun, a preposition and another noun. There are two optional contextual tags between the preposition and the second noun: 0 or more determiners and 0 or more adjectives. So, wildcards such as *, +, or ? have their standard meaning in regular expressions. The two last patterns contain negative contexts. The 8th pattern introduces a negative context at the left side of the rule. It matches any ADV VERB combination if only if there is no DT to the left. The last one introduces a negative context at the right side. It matches any VERB NOUN combinaion followed of a tag different from PRP.