[Attempto] Use of Prefixes with the Large Lexicon

[Norbert E. Fuchs]

On 10 Aug 2010, at 17:31, Chris Spencer wrote:

> I'm using the large lexicon, but I'm noticing that APE still requires
> I explicitly tag most nouns and verbs with a prefix.
> 
> For example, this fails to understand that "ostrich" is a noun, even
> though the lexicon only lists it as a noun.
> 
> ./ape.exe -text "An ostrich does not fly." -cparaphrase
> 
>    <message
> 	importance="error"
> 	type="word"
> 	sentence="1"
> 	token=""
> 	value="ostrich"
> 	repair="Use the prefix n:, v:, a: or p:."/>

This I do not understand. Here is my result

NEF-MacBook-Pro:ape fuchs$ ./ape.exe -text "An ostrich does not fly." -cparaphrase
<?xml version="1.0" encoding="UTF-8"?>

<apeResult>
  <duration tokenizer="0.009" parser="0.000" refres="0.000"/>
  <paraphrase>There is an ostrich X1.
It is false that the ostrich X1 flies.</paraphrase>
  <messages/>
</apeResult>

> And of course this runs fine:
> 
> ./ape.exe -text "An n:ostrich does not fly." -cparaphrase
> 
>  <paraphrase>There is an n:ostrich X1.
> It is false that the n:ostrich X1 flies.</paraphrase>

I too get this.

> Why is APE requiring I specify a prefix for certain words, even though
> they unambiguously exist in the lexicon?

???

I understand that you are running APE in Windows. I am running it in Mac OS X using SWI-Prolog 5.11.2.

But that does not explain the different results.

   --- nef