convert_openie_to_conll
Extraction#
Extraction = namedtuple(
"Extraction", # Open IE extraction
[
"sent", # Sentence in which this ...
Element#
Element = namedtuple(
"Element", # An element (predicate or argument) in an Open IE extraction
[
...
main#
def main(inp_fn: str, domain: str, out_fn: str) -> None
inp_fn: str
Path to file from which to read Open IE extractions in Open IE4's format.
domain: str
Domain to be used when writing CoNLL format.
out_fn: str
Path to file to which to write the CoNLL format Open IE extractions.
safe_zip#
def safe_zip(*args)
Zip which ensures all lists are of same size.
char_to_word_index#
def char_to_word_index(char_ind: int, sent: str) -> int
Convert a character index to word index in the given sentence.
element_from_span#
def element_from_span(span: List[Token], span_type: str) -> Element
Return an Element from span (list of spaCy tokens).
split_predicate#
def split_predicate(ex: Extraction) -> Extraction
Ensure single word predicate by adding "before-predicate" and "after-predicate" arguments.
extraction_to_conll#
def extraction_to_conll(ex: Extraction) -> List[str]
Return a CoNLL representation of a given input Extraction.
interpret_span#
def interpret_span(text_spans: str) -> Tuple[int, int]
Return an integer tuple from textual representation of closed/open spans.
interpret_element#
def interpret_element(
element_type: str,
text: str,
span: str
) -> Element
Construct an Element instance from regexp groups.
parse_element#
def parse_element(raw_element: str) -> List[Element]
Parse a raw element into text and indices (integers).
read#
def read(fn: str) -> Iterable[List[Extraction]]
convert_sent_to_conll#
def convert_sent_to_conll(
sent_ls: List[Extraction]
) -> Iterable[Tuple[str, ...]]
Given a list of extractions for a single sentence, converts it to CoNLL representation.
pad_line_to_ontonotes#
def pad_line_to_ontonotes(
line: Tuple[str, ...],
domain: str
) -> List[str]
Pad line to conform to OntoNotes representation.
convert_sent_to_conll_str#
def convert_sent_to_conll_str(
sent_ls: List[Extraction],
domain: str
) -> str
Given a dictionary from sentence -> extractions, return a corresponding CoNLL representation.