Class ParserConstraint

  • All Implemented Interfaces:
    Serializable

    public class ParserConstraint
    extends Object
    implements Serializable
    A Constraint represents a restriction on possible parse trees to consider. It says that a parse cannot be postulated that would contradict having a constituent from position start to end, and that any constituent postulated with this span must match the regular expression given by state. Note, however, that it does not strictly guarantee that such a constituent exists in a returned parse.
    The words in constraint bounds are counted starting from 0. Furthermore, the end number is not included in the constraint. For example, a constraint covering the first two words of a sentence has a start of 0 and an end of 2.
    state must successfully match states used internal to the parser, not just the final states, so the pattern "NP" will not match anything. Better is "NP.*". Note that this does run the risk of matching states with the same prefix. For example, there is SBAR and SBARQ, so "SBAR.*" could have unexpected results. The states used internal to the parser extend the state name with non-alphanumeric characters, so a fancy expression such as "SBAR|SBAR[^a-zA-Z].*" would match SBAR but not SBARQ constituents.
    See Also:
    Serialized Form
    • Field Detail

      • start

        public final int start
      • end

        public final int end
      • state

        public final Pattern state
    • Constructor Detail

      • ParserConstraint

        public ParserConstraint​(int start,
                                int end,
                                String pattern)
      • ParserConstraint

        public ParserConstraint​(int start,
                                int end,
                                Pattern state)