« Return to Thread: Comma's in

Re: Comma's in <that>

by R. Vince :: Rate this Message:

Reply to Author | View in Thread

Hi Noel,

I am using AIML essentially as a front end, to an NLP system. This
simplifies things greatly, as I can take, say, categories like "Well *" and
just pass them along as "*".

But, ultimately, I am not using AIML to return the result to the user, just
to clean up input. Aside from that, I am trying to implement a full-blown
NLP system with seemingly good success thus far. However, I have been at
this a few years, and, to-date, can only answer queries that are of the
assertion type, e.g. yes/no, true/false. etc. The Wh-type I have not even
gotten to.

It is very difficult however, frought with hundreds of language complexities
which I have been seeking to overcome (pronoun resolution, multiple-names
for the same individual, etc. etc.). I DO find Aiml, however, to be superb
as a front-end filter for this type of system.

-Ralph



----- Original Message -----
From: "Noel Bush" <noel@...>
To: "Alicebot and AIML General Discussion"
<alicebot-general@...>
Sent: Tuesday, January 30, 2007 12:32 PM
Subject: Re: [alicebot-general] Comma's in <that>


> It would be interesting to hear more about how useful you find AIML in
> the way you're using it.  It does seem to have this "general purpose"
> capability, albeit with some gaps, but I don't know whether anyone has
> really kept up an effort to implement something like you're describing
> using AIML.  And it would be interesting to see AIML evaluated against
> other tools for the same kind of front-end purposes.
>
> R. Vince wrote:
>> Absolutely, but, the clarification (ambiguity reduction) seems to be at
>> about the square root of the effort, which is fortunate. That is, if I do
>> 4
>> units of work, I reduve 2 units of ambiguity. If 16 units of work,. w
>> units
>> of ambiguity. In other words, since I know we'll never get 100% absence
>> of
>> ambiguity (in fact, human beings themselves dont get this), if we can get
>> a
>> good chunk of ambiguity resolved, it's worhtwhile I find.
>>
>> But you;re right, it's monstrous to do even that! -Ralph
>>
>> ----- Original Message -----
>> From: "Noel Bush" <noel@...>
>> To: "Alicebot and AIML General Discussion"
>> <alicebot-general@...>
>> Sent: Tuesday, January 30, 2007 10:25 AM
>> Subject: Re: [alicebot-general] Comma's in <that>
>>
>>
>>> And I think that a linguist would tell you that it has consumed
>>> uncountable person-decades of many people's time.  ;-)
>>>
>>> R. Vince wrote:
>>>> Yes! There are POS taggers, but, it is the resolution of ambiguity via
>>>> context after the initial tagging, to correctly tag, that is the
>>>> challenge,
>>>> and has consumed uncountable hours of my time! -Ralph Vince
>>>>
>>>> ----- Original Message -----
>>>> From: "Noel Bush" <noel@...>
>>>> To: "Alicebot and AIML General Discussion"
>>>> <alicebot-general@...>
>>>> Sent: Tuesday, January 30, 2007 9:19 AM
>>>> Subject: Re: [alicebot-general] Comma's in <that>
>>>>
>>>>
>>>>> Now if it were only possible to accurately identify (or even define,
>>>>> in
>>>>> some cases) "parts of speech"..... :-)
>>>>>
>>>>> R. Vince wrote:
>>>>>> Though not part of the AIML spec (and I am mentioneing this solely
>>>>>> for
>>>>>> the
>>>>>> sake of some future coder faced with this problem, and how I resolved
>>>>>> it), I
>>>>>> tag the tokens with Part of Speech Tags as a pre-stage as part of the
>>>>>> Normalizing process, before Normalizing. If the POS's on either side
>>>>>> of
>>>>>> the
>>>>>> comma are the same (e.g. Jack/NP ,/CC Jill/NP) I convert the comma
>>>>>> into
>>>>>> the
>>>>>> coordingating conjunction 'and' (thus producing Jack/NP and/CC
>>>>>> Jill/NP).
>>>>>>
>>>>>> On the other hand, if the POS's on either side of the comma are NOT
>>>>>> the
>>>>>> same, I remove the comma (e.g. "In/CD the/DT end/NN ,/CC no/DT one/NN
>>>>>> cares/VBZ" thus becomes ("in/CD the/DT end/NN no/DT one/NN
>>>>>> cares/VBZ").
>>>>>>
>>>>>> Essentially, if you wanted, a semicolon could be treated the same
>>>>>> way.
>>>>>> But
>>>>>> again, this is off-spec. I simply have incorporated it because I am
>>>>>> working
>>>>>> on parsing text via AIML as a front end to a larger system.
>>>>>>
>>>>>> -Ralph Vince
>>>>>>
>>>>>> ----- Original Message -----
>>>>>> From: "mehri" <foreverlinux@...>
>>>>>> To: <alicebot-general@...>
>>>>>> Sent: Monday, January 29, 2007 11:20 PM
>>>>>> Subject: [alicebot-general] Comma's in <that>
>>>>>>
>>>>>>
>>>>>>> I ran into a particular issue.
>>>>>>>
>>>>>>> Some AIML files are written with comma's in <that> and
>>>>>>> some are not.
>>>>>>>
>>>>>>> For example,
>>>>>>>
>>>>>>> <that>PRESS 1 TO KNOW WHAT THE MOUSE EATS, AND TO
>>>>>>> DISCOVER ITS COLOUR</that>
>>>>>>>
>>>>>>> and some are:
>>>>>>>
>>>>>>> <that>PRESS 1 TO KNOW WHAT THE MOUSE EATS AND TO
>>>>>>> DISCOVER ITS COLOUR</that>
>>>>>>>
>>>>>>> I do believe that <that>'s shouldn't contain comma's
>>>>>>> or other punctuation.
>>>>>>>
>>>>>>> Is that true?
>>>>>>>
>>>>>>> I didn't see anywhere in the spec pointing to this
>>>>>>> specifically, unless I missed something again in the
>>>>>>> specification :-P
>>>>>>>
>>>>>>>
>>>>>>>
>>>>>>>
>>>>>>> __________________________________________________
>>>>>>> Do You Yahoo!?
>>>>>>> Tired of spam?  Yahoo! Mail has the best spam protection around
>>>>>>> http://mail.yahoo.com
>>>>>>> _______________________________________________
>>>>>>> This is the alicebot-general mailing list
>>>>>>> Reply to alicebot-general@...
>>>>>>> Unsubscribe and change preferences at
>>>>>>> http://list.alicebot.org/mailman/listinfo/alicebot-general
>>>>>>> Learn netiquette at http://www.dtcc.edu/cs/rfc1855.html
>>>>>>> Learn to read at http://www.literacy.org/
>>>>>> _______________________________________________
>>>>>> This is the alicebot-general mailing list
>>>>>> Reply to alicebot-general@...
>>>>>> Unsubscribe and change preferences at
>>>>>> http://list.alicebot.org/mailman/listinfo/alicebot-general
>>>>>> Learn netiquette at http://www.dtcc.edu/cs/rfc1855.html
>>>>>> Learn to read at http://www.literacy.org/
>>>>> _______________________________________________
>>>>> This is the alicebot-general mailing list
>>>>> Reply to alicebot-general@...
>>>>> Unsubscribe and change preferences at
>>>>> http://list.alicebot.org/mailman/listinfo/alicebot-general
>>>>> Learn netiquette at http://www.dtcc.edu/cs/rfc1855.html
>>>>> Learn to read at http://www.literacy.org/
>>>> _______________________________________________
>>>> This is the alicebot-general mailing list
>>>> Reply to alicebot-general@...
>>>> Unsubscribe and change preferences at
>>>> http://list.alicebot.org/mailman/listinfo/alicebot-general
>>>> Learn netiquette at http://www.dtcc.edu/cs/rfc1855.html
>>>> Learn to read at http://www.literacy.org/
>>> _______________________________________________
>>> This is the alicebot-general mailing list
>>> Reply to alicebot-general@...
>>> Unsubscribe and change preferences at
>>> http://list.alicebot.org/mailman/listinfo/alicebot-general
>>> Learn netiquette at http://www.dtcc.edu/cs/rfc1855.html
>>> Learn to read at http://www.literacy.org/
>>
>> _______________________________________________
>> This is the alicebot-general mailing list
>> Reply to alicebot-general@...
>> Unsubscribe and change preferences at
>> http://list.alicebot.org/mailman/listinfo/alicebot-general
>> Learn netiquette at http://www.dtcc.edu/cs/rfc1855.html
>> Learn to read at http://www.literacy.org/
> _______________________________________________
> This is the alicebot-general mailing list
> Reply to alicebot-general@...
> Unsubscribe and change preferences at
> http://list.alicebot.org/mailman/listinfo/alicebot-general
> Learn netiquette at http://www.dtcc.edu/cs/rfc1855.html
> Learn to read at http://www.literacy.org/ 

_______________________________________________
This is the alicebot-general mailing list
Reply to alicebot-general@...
Unsubscribe and change preferences at http://list.alicebot.org/mailman/listinfo/alicebot-general
Learn netiquette at http://www.dtcc.edu/cs/rfc1855.html
Learn to read at http://www.literacy.org/

 « Return to Thread: Comma's in