I am using NLTK POS-Tagging to extract information from a text, in this example I am looking for an IBAN. For some texts the code returns more than one chunk, but I don't mind that, I will sort the correct one out later with a RegEx. Now here is my question: is there a prettier way to get the Strings of the chunk so I can work with them or save them?
Of course you could go the artifical way (i.e. iterate through all lines in ibanChunk, then ibanChunk.replace(..) etc.) but there must be a better way, at least that's what I'm hoping.
tagged_sents = list(corp.tagged_sents())
tagger = ClassifierBasedGermanTagger(train=tagged_sents)
tagged_sents = tagger.tag(filtered_sentence)
ibanChunkGram = r"""Chunk: {(<VMPP><CARD>*)|(<FM><CARD>+)}"""
chunkParser = nltk.RegexpParser(ibanChunkGram)
ibanChunk = chunkParser.parse(tagged_sents)
print(ibanChunkGram)
Right now the output of the line looks like this:
(Chunk DE01/FM 2345/CARD 6789/CARD 0000/CARD 0000/CARD 00/CARD)
and what I want to have is:
DE01 2345 6789 0000 0000 00
Edit: Here is a minimalExample:
This is a minimal example of POS-tagging. I want to extract an IBAN (DE01 2345 6789 0000 0000 00) and I hope The Machine 01 can find it quick.
And this is the output of my code:
(S
This/NE
is/FM
a/FM
minimal/FM
example/FM
of/FM
POS-tagging/FM
./$.
I/FM
want/FM
to/FM
extract/FM
IBAN/FM
(/$(
(Chunk DE01/FM 2345/CARD 6789/CARD 0000/CARD 0000/CARD 00/CARD)
)/$(
and/NE
I/NE
hope/VAFIN
The/NE
Machine/NE
01/CARD
can/XY
find/XY
it/XY
quick/XY
./$.)
Alright, I have figured it out myself by now. In case anyone is ever stumbling upon that problem too, here is my solution: ibanChunk, as it is called in my case, is a list of tuples, but a Chunk turned out to be a tree, not a tuple, so I used that as my advantage, here is my code:
for elem in ibanChunk:
if isinstance(elem, nltk.Tree):
ibanString = ""
for (text, tag) in elem:
ibanString += text
chunkList.append(ibanString)
And there you have the text of all chunks in a list as Strings.