<div class="gmail_quote">2010/7/12 Norbert E. Fuchs <<a href="mailto:fuchs@ifi.uzh.ch">fuchs@ifi.uzh.ch</a>> <blockquote class="gmail_quote" style="margin: 0pt 0pt 0pt 0.8ex; border-left: 1px solid rgb(204, 204, 204); padding-left: 1ex;"> <div class="im"> On 12 Jul 2010, at 19:24, Gabriele Kahlout wrote: > I'm wondering why "There are 3 cards" is not paraphrased in any 'intelligent' way, eg. "There is a card X1. There is a card X2. There is a card X3." > In English I would then unambiguously refer to "the first card [of the 3 cards]" and then "the only red card in the 3 cards/them" and "a card of the 3 cards". My hope is that ACE could express (at least conceptually) every thing unambiguous english could. > > Is it because paraphrasing "there are 10000000 cards" is expensive? > If so, no cheap alternatives exist? > Could this be one: > If the parser encounters "the first card [of the 3 cards]" and it's the first encounter then the parser backtracks and adds "there is a card Xn". Then the first card could be paraphrased as The card X1 [...]. > > "the only red card in the 3 cards/them": > (Is the lack of support for only a matter of time/priority or a technical difficulty?) > There is a card Xn. The card Xn is a red card. > > "a card of the 3 cards": > There is a card Xn. The card Xn [...]. > Xn could then actually be Xn-1 for instance. </div><div class="im">On 12 Jul 2010, at 19:49, Jean-Marc Vanel wrote: > 2010/7/12 Gabriele Kahlout <<a href="mailto:gabriele@mysimpatico.com">gabriele@mysimpatico.com</a>> </div><div class="im">> Hello, > I'm wondering why "There are 3 cards" is not paraphrased in any 'intelligent' way, eg. "There is a card X1. There is a card X2. There is a card X3." > </div><div class="im">> First your proposed paraphrase is not semantically correct ACE (and FOL). > There is no garanty that the 3 variables are distinct. > </div><div class="im">> A possible ACE paraphrase would be : > ./ape.exe -guess -text 'p:X1 is a card. p:X2 is a card. p:X3 is a card. ' -solo paraphrase > X1 is a card. > X2 is a card. > X3 is a card. > > A proper noun ( p: prefix ) is the way to "instanciate" a unique object. </div>Some comments. Concerning Gabriele's statements. First, ACE is a syntactic language that leaves interpretations – for instance the meaning of words, the meaning of the copula , or the meaning of collective plurals – to the users, respectively to their programs that process the DRS generated by APE.</blockquote> <div>I assume paraphrasing is done by APE (described as Attempto system in the troubleshooting guide). If so then it seems to me that some 'interpretation' is done. Paraphrasing X1 is a card into: <pre class="resultParaphrase"> There is something X1. X1 is a card. </pre> does not seem just syntactic analysis, but interpretation. If so then my expectations above are founded. </div><blockquote class="gmail_quote" style="margin: 0pt 0pt 0pt 0.8ex; border-left: 1px solid rgb(204, 204, 204); padding-left: 1ex;"> Second, consider that "intelligent" paraphrasing depends on the expectations of the respective user. Your expectations may differ from mine.</blockquote><div>So my paraphrasing followed from the above one did, and I would argue that for the described case above (at least, the first, the second, etc..) is unambiguous natural language. I do agree that it could be done at a later stage, by RACE for instance. </div><blockquote class="gmail_quote" style="margin: 0pt 0pt 0pt 0.8ex; border-left: 1px solid rgb(204, 204, 204); padding-left: 1ex;"> Third, to provide a feedback to the users ACE's paraphrasing gives a *syntactic* variant of the input text, not a semantic one. </blockquote><div>I'm not sure I grasp the fine line between semantic analysis and paraphrasing. So if a user accepts a paraphrase then his expression and the paraphrase are semantically equivalent. So the accepted syntactic variant is also a semantic variant. Then my suggestion could be a syntatic variant too. </div><blockquote class="gmail_quote" style="margin: 0pt 0pt 0pt 0.8ex; border-left: 1px solid rgb(204, 204, 204); padding-left: 1ex;"> Concerning Jean-Marc's comments. Whether or two variables or two constants – expressed in ACE as proper nouns– refer to different or to the same object depends on the program that interprets the DRS. While RACE – like FOL – uses the unique name assumption, OWL – generated by the ACE -> OWL translation – does not. Regards. --- nef </blockquote></div> -- Regards, K. Gabriele --- unchanged since 25/1/10 --- P.S. Unless a notification (LON), please reply either with an answer OR with " ACK" appended to this subject within 48 hours. Otherwise, I might resend. In(LON, this) ∨ In(48h, TimeNow) ∨ ∃x. In(x, MyInbox) ∧ IsAnswerTo(x, this) ∨ (In(subject(this), subject(x)) ∧ In(ACK, subject(x)) ∧ ¬IsAnswerTo(x,this)) ⇒ ¬IResend(this). Also note that correspondence may be received only from specified a priori senders, or if the subject of this email ends with a code, eg. -LICHT01X, then also from senders whose reply contains it. ∀x. In(x, MyInbox) ⇒ In(senderAddress(x), MySafeSenderList) ∨ (∃y. In(y, subject(this) ) ∧ In(y,x) ∧ isCodeLike(y, -LICHT01X) ).