1010961474532073472.md

reply to https://twitter.com/_inesmontani/status/1010961474532073472

class Annotator(object):
    def __init__(self, nlp, **kwargs):
        # `nlp` is a spaCy Language object
        self.nlp = nlp

    def annotate(self, agr, rules):
        """ `agr` (short for Agreement) is a tree-like data structure.
        A node of the tree (class `Element`) has a `tag` (e.g. PREAMBLE, SECTION) and may have 
        `content` (e.g. textual data), `annotations` (string identifiers of `Rule`s)
        and one or many `children`.
        """
        """ `rules` is a dictionary with string keys and values of class `Rule`.
        `Rule`s contain regex-like `patterns` using Universal Dependencies and `applicable_tags`
        for targeting specific Agreement node types (the `tag` from `Element`)
        """

        # ... pseudo-code for possible implementation details
        # for each node in agr:
        #   foreach key, rule in rules:
        #     - SKIP UNLESS node.tag IN rule.applicable_tags
        #
        #     - make spaCy Doc object out of node.content
        #     - IF rule matches Doc object
        #       -- append key to node.annotations

        return agr