def obl_modifier(dep_graph: DependencyGraph, oia_graph: OIAGraph,
                 context: UD2OIAContext):
    """
    the adv before the verb should be processed by verb_phrase
    this converter should process the adv after the verb
    :param sentence:
    :return:
    """

    pattern = DependencyGraph()

    # TODO: it seems that in UD labeling, adv is used instead of adj for noun
    modified_node = DependencyGraphNode()
    modifier_node = DependencyGraphNode()

    pattern.add_nodes([modified_node, modifier_node])

    pattern.add_dependency(modified_node, modifier_node, r'\bobl')

    for match in dep_graph.match(pattern):

        dep_modified_node = match[modified_node]
        dep_modifier_node = match[modifier_node]

        if oia_graph.has_relation(dep_modified_node,
                                  dep_modifier_node,
                                  direct_link=False):
            continue

        oia_modified_node = oia_graph.add_words(dep_modified_node.position)
        oia_modifier_node = oia_graph.add_words(dep_modifier_node.position)

        oia_graph.add_mod(oia_modifier_node, oia_modified_node)
def adj_modifier(dep_graph: DependencyGraph, oia_graph: OIAGraph,
                 context: UD2OIAContext):
    """
    adj previous to noun is coped with by noun phrase
    this process the case that adj is behind the noun
    #################### a pretty little boy ########################
    :param dep_graph:
    :param oia_graph:
    :return:
    """

    pattern = DependencyGraph()

    noun_node = pattern.create_node()  # UPOS="NOUN|PRON|PROPN")
    adj_node = pattern.create_node()  # UPOS="ADJ|NOUN")

    pattern.add_dependency(noun_node, adj_node, r'amod')

    for match in dep_graph.match(pattern):
        dep_noun_node = match[noun_node]
        dep_adj_node = match[adj_node]

        oia_noun_node = oia_graph.add_words(dep_noun_node.position)

        oia_adj_node = oia_graph.add_words(dep_adj_node.position)
        logger.debug("adj_modifier: ")
        logger.debug(dep_noun_node.position)
        logger.debug(oia_noun_node)
        logger.debug(dep_adj_node.position)
        logger.debug(oia_adj_node)

        oia_graph.add_mod(oia_adj_node, oia_noun_node)
def acl_mod_verb(dep_graph: DependencyGraph, oia_graph: OIAGraph,
                 context: UD2OIAContext):
    """
    this is called after adnominal_clause_mark, which means there is no mark
    :param dep_graph:
    :param oia_graph:
    :return:
    """

    pattern = DependencyGraph()

    noun_node = pattern.create_node(UPOS="NOUN|PRON|PROPN|ADJ|ADV|NUM")
    # ADJ is for the cases that "many/some" are abbrv of many X/some X, representing NOUN
    # ADV is for the case of "here" for "i am here thinking xxx"
    verb_node = pattern.create_node(UPOS="VERB|AUX")
    # aux is for can, have which ommits the true verb

    pattern.add_nodes([noun_node, verb_node])

    pattern.add_dependency(noun_node, verb_node, r'acl')

    for match in dep_graph.match(pattern):

        dep_noun_node = match[noun_node]
        dep_verb_node = match[verb_node]

        if context.is_processed(dep_noun_node, dep_verb_node):
            continue

        if oia_graph.has_relation(dep_noun_node,
                                  dep_verb_node,
                                  direct_link=False):
            continue

        oia_verb_node = oia_graph.add_words(dep_verb_node.position)
        oia_noun_node = oia_graph.add_words(dep_noun_node.position)

        dep = dep_graph.get_dependency(dep_noun_node, dep_verb_node)
        labels = [x for x in dep.rels if x.startswith("acl:")]

        pred = None

        if labels:
            assert len(labels) == 1
            label = labels[0]
            pred = label.split(":")[1]
            if pred == "relcl":
                pred = None

        # if pred:
        #     # there is no mark, but we add it because it may be because of not being shared in conjunction
        #
        #     oia_pred_node = oia_graph.add_aux(pred)
        #     oia_graph.add_argument(oia_pred_node, oia_noun_node, 1, mod=True)
        #     oia_graph.add_argument(oia_pred_node, oia_verb_node, 2)
        # else:

        oia_graph.add_mod(oia_verb_node, oia_noun_node)
Ejemplo n.º 4
0
def adverbial_clause(dep_graph: DependencyGraph, oia_graph: OIAGraph,
                     context: UD2OIAContext):
    """
    Adverbial Clause
##### run in order to catch it. advcl with mark (in order to) #####
##### he worked hard, replacing his feud. advcl without mark #####

    :param dep_graph:
    :param oia_graph:
    :return:
    """
    pattern = DependencyGraph()
    verb_node = pattern.create_node()
    modifier_node = pattern.create_node()

    pattern.add_dependency(verb_node, modifier_node, "advcl")

    for match in list(dep_graph.match(pattern)):

        dep_verb_node = match[verb_node]
        dep_modifier_node = match[modifier_node]

        if context.is_processed(dep_verb_node, dep_modifier_node):
            continue

        oia_verb_node = oia_graph.add_words(dep_verb_node.position)
        oia_modifier_node = oia_graph.add_words(dep_modifier_node.position)

        logger.debug("adverbial clause: verb={0}, modifier={1}".format(
            dep_verb_node.position, dep_modifier_node.position))

        if oia_graph.has_relation(oia_verb_node, oia_modifier_node):
            continue

        mark = list(
            dep_graph.children(dep_modifier_node,
                               filter=lambda n, rel: "mark" in rel))

        if mark:
            mark, rel = mark[0]
            pred_node = oia_graph.add_words(mark.position)
            if pred_node is None:
                continue

            if mark.LEMMA in CONJUNCTION_WORDS[language]:
                continue

            oia_graph.add_argument(pred_node, oia_verb_node, 1, mod=True)
            oia_graph.add_argument(pred_node, oia_modifier_node, 2)
        else:

            oia_graph.add_mod(oia_modifier_node, oia_verb_node)
def det_predet(dep_graph: DependencyGraph, oia_graph: OIAGraph,
               context: UD2OIAContext):
    """

    :param dep_graph:
    :param oia_graph:
    :return:
    """

    for n1, n2, dep in dep_graph.dependencies():

        if "det:predet" in dep:
            oia_n1 = oia_graph.add_words(n1.position)
            oia_n2 = oia_graph.add_words(n2.position)
            oia_graph.add_mod(oia_n2, oia_n1)
def oblique_without_prep(dep_graph: DependencyGraph, oia_graph: OIAGraph,
                         context: UD2OIAContext):
    """

    :param dep_graph:
    :param oia_graph:
    :return:
    """

    # cut X by a knife
    pattern = DependencyGraph()
    verb_node = DependencyGraphNode(UPOS="VERB|NOUN|ADJ|PROPN|PRON")
    oblique_node = DependencyGraphNode()
    pattern.add_node(verb_node)
    pattern.add_node(oblique_node)
    pattern.add_dependency(verb_node, oblique_node, r'obl:tmod|obl:npmod|obl')

    for match in dep_graph.match(pattern):

        dep_verb_node = match[verb_node]
        dep_oblique_node = match[oblique_node]

        if oia_graph.has_relation(dep_verb_node,
                                  dep_oblique_node,
                                  direct_link=False):
            continue

        oblique_edge = dep_graph.get_dependency(dep_verb_node,
                                                dep_oblique_node)
        oblique_types = oblique_edge.values()

        if "tmod" in oblique_types:

            oia_pred_node = oia_graph.add_aux("TIME_IN")

            arg1_node = oia_graph.add_words(dep_verb_node.position)
            arg2_node = oia_graph.add_words(dep_oblique_node.position)

            oia_graph.add_argument(oia_pred_node, arg1_node, 1, mod=True)
            oia_graph.add_argument(oia_pred_node, arg2_node, 2)

        else:  # "npmod" in oblique_types and others

            oia_verb_node = oia_graph.add_words(dep_verb_node.position)
            obl_node = oia_graph.add_words(dep_oblique_node.position)

            oia_graph.add_mod(obl_node, oia_verb_node)
def nmod_without_case(dep_graph: DependencyGraph, oia_graph: OIAGraph,
                      context: UD2OIAContext):
    """

    #################### nmod:x ########################

    :param sentence:
    :return:
    """

    pattern = DependencyGraph()

    center_node = pattern.create_node()
    modifier_node = pattern.create_node()

    pattern.add_dependency(center_node, modifier_node, r'\w*nmod\w*')

    for match in dep_graph.match(pattern):

        dep_center_node = match[center_node]
        dep_modifier_node = match[modifier_node]

        rels = dep_graph.get_dependency(dep_center_node, dep_modifier_node)

        if "nmod:poss" in rels and dep_center_node in set(
                dep_graph.offsprings(dep_modifier_node)):
            # whose in there
            continue

        if oia_graph.has_relation(dep_center_node,
                                  dep_modifier_node,
                                  direct_link=False):
            continue

        oia_center_node = oia_graph.add_words(dep_center_node.position)
        oia_modifier_node = oia_graph.add_words(dep_modifier_node.position)

        oia_graph.add_mod(oia_modifier_node, oia_center_node)
def adv_verb_modifier(dep_graph: DependencyGraph, oia_graph: OIAGraph,
                      context: UD2OIAContext):
    """
    the adv before the verb should be processed by verb_phrase
    this converter should process the adv after the verb
    verb1 in order to verb2
    :param sentence:
    :return:
    """

    pattern = DependencyGraph()

    # TODO: it seems that in UD labeling, adv is used instead of adj for noun
    verb_node = DependencyGraphNode(
        UPOS="VERB|NOUN|PROPN|AUX|PRON")  # aux is for be word
    adv_node = DependencyGraphNode(UPOS="ADV|X|NOUN|ADJ|VERB")

    pattern.add_nodes([verb_node, adv_node])

    pattern.add_dependency(verb_node, adv_node, r'advmod')

    for match in dep_graph.match(pattern):

        dep_verb_node = match[verb_node]
        dep_adv_node = match[adv_node]

        if context.is_processed(dep_verb_node, dep_adv_node):
            continue

        if oia_graph.has_relation(dep_verb_node, dep_adv_node):
            continue

        obl_children = [
            x for x, l in dep_graph.children(
                dep_adv_node, filter=lambda n, l: l.startswith("obl"))
        ]

        obl_node = None
        obl_has_case = False
        if len(obl_children) == 1:

            obl_node = obl_children[0]

            case_nodes = list(n for n, l in dep_graph.children(
                obl_node, filter=lambda n, l: "case" in l))

            if case_nodes:
                # if obl with case, let the oblique to process it
                obl_has_case = True

        mark_children = [
            x for x, l in dep_graph.children(
                dep_adv_node, filter=lambda n, l: l.startswith("mark"))
        ]

        oia_verb_node = oia_graph.add_words(dep_verb_node.position)
        oia_adv_node = oia_graph.add_words(dep_adv_node.position)

        if obl_node and not obl_has_case:
            # arg_nodes = list(dep_graph.offsprings(obl_node))
            # arg_nodes.sort(key=lambda x: x.LOC)
            # arg_words = [x.ID for x in arg_nodes]
            # head = obl_node.ID

            oia_arg_node = oia_graph.add_words(obl_node.position)

            oia_graph.add_argument(oia_adv_node, oia_verb_node, 1, mod=True)
            oia_graph.add_argument(oia_adv_node, oia_arg_node, 2)
        else:
            if mark_children:
                mark_node = mark_children[0]
                oia_pred_node = oia_graph.add_words(mark_node.position)

                oia_graph.add_argument(oia_pred_node,
                                       oia_verb_node,
                                       1,
                                       mod=True)
                oia_graph.add_argument(oia_pred_node, oia_adv_node, 2)

            else:
                oia_graph.add_mod(oia_adv_node, oia_verb_node)
Ejemplo n.º 9
0
def object_relative_clause(dep_graph: DependencyGraph, oia_graph: OIAGraph,
                           context: UD2OIAContext):
    """
    ##### Object-extracted/referred relative clause #####
    ##### the person that Andy knows #####
    :param sentence:
    :return:
    """

    pattern = DependencyGraph()
    verb_node = DependencyGraphNode()
    entity_node = DependencyGraphNode()
    subj_node = DependencyGraphNode()

    pattern.add_nodes([verb_node, entity_node, subj_node])

    pattern.add_dependency(verb_node, subj_node, r'\w*subj\w*')
    pattern.add_dependency(entity_node, verb_node, r'\w*acl:relcl\w*')

    for match in dep_graph.match(pattern):

        dep_entity_node = match[entity_node]
        dep_subj_node = match[subj_node]
        dep_verb_node = match[verb_node]

        if dep_subj_node.LEMMA in {"what", "who", "which", "that"}:
            continue

        logger.debug("we found a objective relative clause")
        logger.debug("entity: {0}".format(dep_entity_node))
        logger.debug("subject: {0}".format(dep_subj_node))
        logger.debug("verb: {0}".format(dep_verb_node))

        if context.is_processed(dep_entity_node, dep_verb_node):
            logger.debug("processed")
            continue

        context.processed(dep_verb_node, dep_subj_node)
        context.processed(dep_entity_node, dep_verb_node)

        oia_entity_node = oia_graph.add_words(dep_entity_node.position)
        oia_verb_node = oia_graph.add_words(dep_verb_node.position)
        oia_subj_node = oia_graph.add_words(dep_subj_node.position)

        if oia_graph.has_relation(oia_entity_node, oia_verb_node):
            logger.debug("has relation between entity and verb")
            continue

        oia_graph.add_argument(oia_verb_node, oia_subj_node, 1)

        def __valid_ref(n, l):
            return l == "ref" and dep_entity_node.LOC < n.LOC < dep_verb_node.LOC

        ref_nodes = list(n for n, l in dep_graph.children(dep_entity_node,
                                                          filter=__valid_ref))
        ref_nodes.sort(key=lambda x: x.LOC)

        if ref_nodes:
            ref_node = ref_nodes[-1]

            oia_ref_node = oia_graph.add_words(ref_node.position)

            oia_graph.add_ref(oia_entity_node, oia_ref_node)

            logger.debug("we are coping with ref between:")
            logger.debug(dep_verb_node)
            logger.debug(ref_node)

            ref_relation = dep_graph.get_dependency(dep_verb_node, ref_node)

            case_nodes = list(n for n, l in dep_graph.children(
                ref_node, filter=lambda n, l: "case" in l))
            case_nodes.sort(key=lambda x: x.LOC)

            if ref_relation:
                if case_nodes:
                    # with which xxxx, the with will become the root pred
                    case_node = case_nodes[-1]
                    oia_case_node = oia_graph.add_words(case_node.position)

                    oia_graph.add_argument(oia_case_node,
                                           oia_verb_node,
                                           1,
                                           mod=True)
                    oia_graph.add_argument(oia_case_node, oia_ref_node, 2)
                    oia_graph.add_mod(oia_verb_node, oia_entity_node)
                else:

                    if "obj" in ref_relation:
                        oia_graph.add_argument(oia_verb_node, oia_ref_node, 2)
                    elif ref_relation == "advmod":
                        oia_graph.add_mod(oia_ref_node, oia_verb_node)
                    else:
                        raise Exception(
                            "unknown relation: {}".format(ref_relation))
                    # oia_graph.add_argument(oia_verb_node, oia_entity_node, 2, mod=True)

        oia_graph.add_argument(oia_verb_node, oia_subj_node, 1)
        oia_graph.add_argument(oia_verb_node, oia_entity_node, 2, mod=True)

        rels = dep_graph.get_dependency(dep_entity_node, dep_verb_node)

        #if rels.endswith("obj"):
        for node, l in dep_graph.children(dep_verb_node):
            if l == "ccomp":
                oia_ccomp_node = oia_graph.add_words(node.position)
                oia_graph.add_argument(oia_verb_node, oia_ccomp_node, 3)