Hi Nigel
I just tried your text in GATE (using the standard pre-processing tools in 
ANNIE) and it correctly identified the sentences. 
The sentence splitter is rule-based so it’s also possible to adjust it if you 
find it makes errors of this sort (I can’t guarantee it will work on all tricky 
sentence-finding cases.)

We’re also doing legal text processing with GATE currently, incidentally.

If you want to test some sentences with ANNIE, you could try our demo here:

https://cloud.gate.ac.uk/shopfront/displayItem/annie-named-entity-recognizer

Hope that helps
Diana





> On 21 Oct 2019, at 11:22, njbruce <njbr...@hku.hk> wrote:
> 
> We are having problems with our NLP program [StanfordCoreNLP], as we are 
> finding it unable to discriminate between true fullstops and any abbreviation 
> points. [e.g. C.B.S. as below]. This is important to us as we want to show 
> student users of our conconcordancer 3 consecutive sentences, based around a 
> search term result sentence [bolded on the webpage, but not in the example 
> below], to offer them a fuller, more meaningful co-text/context than is 
> usually yielded by concordancers. The sentence-defining function is crucial 
> to this ambition.
> 
> E.g. in a contract law case on undue influence [1985 OSullivan], the original 
> text reads:
> "The first plaintiff claimed as against C.B.S. that the inducement letter did 
> not create any binding contractual obligation between himself and C.B.S. 
> because no consideration was given for it. The judge erred in upholding the 
> claim because it was in consideration for the first plaintiff's promise in 
> the inducement letter that C.B.S. entered into the agreement with the first 
> defendants."
> 
> - StanfordCoreNLP mistakes the final stop in "C.B.S." as sentence-ending, and 
> so the concordanced 3-sentence display it yields is grammatically incoherent:
> 
> “that the inducement letter did not create any binding contractual obligation 
> between himself and C.B.S. because no consideration was given for it. The 
> judge erred in upholding the claim because it was in consideration for the 
> first plaintiff's promise in the inducement letter that C.B.S."
> 
> We wonder if anyone is using a more sophisticated NLP program that produces 
> more reliable sentence-identifying results.
> Any help will be most welcome.
> 
> Nigel Bruce
> Law & Business Dept,
> Hong Kong Shue Yan University
> 
> _______________________________________________
> UNSUBSCRIBE from this page: http://mailman.uib.no/options/corpora
> Corpora mailing list
> Corpora@uib.no
> https://mailman.uib.no/listinfo/corpora

_______________________________________________
UNSUBSCRIBE from this page: http://mailman.uib.no/options/corpora
Corpora mailing list
Corpora@uib.no
https://mailman.uib.no/listinfo/corpora

Reply via email to