As long as the Excel App is compatible with either Open Office, or Office 2k3, I’ll be more than happy to help. Any newer version is beyond my ability to make use of, though.

Chuck Bolin - Oct 3, 2010:

Hi,
Those are some interesting percentages about ‘imagination’ versus ‘functional ability’.  I like this point about age vs. learning a language. That is akin to ‘teaching old dogs new tricks’.

For the past few years I’ve wondered if the apparent ability of children to learn new languages quickly is due to actual differences in brain function, or to their increased willingness to “learn like a child”. That is, the process of language learning that involves months of only being able to eek out simple concepts, being condescended to and corrected regularly, and being completely immersed in the new language. Any and all of these things quickly drive adults to give up or at least become very frustrated. But a child has no choice in the matter, doesn’t often communicate complex concepts (at least at a young age), and at any rate is used to being corrected.

Chuck Bolin - Oct 3, 2010:

This suggests that a chat bot design should be aimed at creating ‘child-like’ bot framework…as opposed to a ‘mature’ bot. Not sure what that means but it is certainly something worth thinking about.

Until we figure out the real reason kids learn languages well, we won’t know how to mimic the effect. What does “child-like” even mean for a robot? Depends alot on the architecture I imagine.

Chuck Bolin - Oct 3, 2010:

Right now I’m going to work on a multi-chat capability within XBot. If I get this working I’ll need a volunteer or two to help me test. I can provide the Excel program.

That is very cool! I would be interested in testing it. I have Excel 2003, but if it requires something more up-to-date, then I’ll need you to provide the program. (Is there a free version of an “Excel reader”? I know this exists for Powerpoint…)

Hi,
I’ve got two issues to resolve before I can support multi-chat. The first is to learn how to do ‘port forwarding’ on my router. This would allow folks outside our home network to connect to the bot server app.  Second is the inability to create an ‘array’ of winsock objects within VBA. It’s easy to do in VB…but not supported in VBA.

Dave and CR,
I save my Excel projects in the 97-2003 version. So this will work nicely. Thanks!!

CR,
A child has no choice…  Good point!!  Have raised 5 children I should have thought of that. =)

*Update*
I’ve been on the grammar parser tonight. I’ve chosen to use the “Penn Tree Tag Set”.  Here’s the current output of the data log. It’s important to note that the word “closing” may serve as a verb or noun. I had to add the ability to define these easily in an Excel worksheet.

NN = noun singular
IN = preposition
DT = determiner
VBG = verb - gerund (...ing ending)

Regarding DOCUMENTATION, I’ve started to write up the status of the various problems that have been solved along with a description of the solution. I decided to do this in order to track research results from the various coding experiments.  I’ll post this soon.

Regards,
Chuck

For the past few years I’ve wondered if the apparent ability of children to learn new languages quickly is due to actual differences in brain function, or to their increased willingness to “learn like a child”.

If I remember clearly, an infant’s brain is not fully grown yet when born. Couldn’t it be that the brain simply grows to accommodate the mother-language. Which also explains why it’s so much harder for adults: their brain might be changeable to a degree, but doesn’t grow any more, so it can no longer create the required structures to handle this new data as efficiently as the infant.

Chuck Bolin - Oct 4, 2010:

NN = noun singular
IN = preposition
DT = determiner
VBG = verb - gerund (...ing ending)

Human: closing of the plant
Bot: [NN] [IN] [DT] [NN]
Human: plant closing in the town
Bot: [NN] [NN] [IN] [DT] [NN]
Human: closing the plant
Bot: [VBG] [DT] [NN] 

Chuck

Actually this is not exactly correct.
The word “closing” in “closing of the plant” is a “VBG”.
Also “closing” in “plant closing in the town” is also a “VBG”.

Yes, they are technically nouns also (gerund = verb playing role of noun), but more specifically they are gerunds.

And in “closing the plant”, the word “closing” is not a VBG but instead simply a verb.

“closing the plant” is a predicate, ‘closing’ being its verb.  Put a noun like ‘we’ in front and you get subject+predicate = sentence “we are closing the plant”

In Chuck’s defence: Couldn’t you also say: I went to the closing of the plant, as in ‘a happening’. That would be a noun, wouldn’t it?

Yes, it would be a noun.

But again, more accurately, “closing” in that case is a gerund.

And “closing of the plant” is a gerund phrase which is playing the role of the direct object.

So yes, it is a noun, but it is more specifically described as being a gerund.  Gerund is a noun , and a verb at the same time.

Is there anything particularly dangerous about just denoting gerunds as nouns? I’ve been doing this from the beginning and functionally, it is quite convenient. As far as the bot recognizing that the noun in question relates to a specific verb, that type of processing (along with recognizing relationships between adjectives/adverbs/etc. forms of words) will be left to a separate interpretation stage that may or may not be necessary for parsing, but probably will be useful to reply meaningfully.

Keep up the good work, Chuck!

Is there anything particularly dangerous about just denoting gerunds as nouns? I’ve been doing this from the beginning and functionally, it is quite convenient.

I don’t know if there are any dangers involved when denoting gerunds as nouns. I am doing the same though. From a semantical point of view, I think they are different. Ex:
I went eating.
I went to the eating.
It’s almost the same, but not quite, is it?

It is important at the parsing stage to know a word is gerund and not simply a noun. 

The reason being is that a gerund can take a direct compliment whereas a noun cannot.

Example,

“I like chatbots”—direct object of predicate verb “like” is noun “chatbots”

“His job is painting houses” 

  direct object of predicate verb “is” is the gerund “painting”

Since it is a gerund and not simply a noun, it has its own direct object of “houses”

So, yes, it actually is important to know a word is a gerund and not simply a noun at the parsing stage even.

Victor Shulist - Oct 4, 2010:

It is important at the parsing stage to know a word is gerund and not simply a noun. 

The reason being is that a gerund can take a direct compliment whereas a noun cannot.

Example,

“I like chatbots”—direct object of predicate verb “like” is noun “chatbots”

“His job is painting houses” 

  direct object of predicate verb “is” is the gerund “painting”

Since it is a gerund and not simply a noun, it has its own direct object of “houses”

So, yes, it actually is important to know a word is a gerund and not simply a noun at the parsing stage even.

What you are saying is of course true (notice the spelling !! ). Just 1 small thing to note: in your example, it’s not exactly a noun, but a verb (at least, that’s how I would parse it).

Jan Bogaerts - Oct 4, 2010:

Just 1 small thing to note: in your example, it’s not exactly a noun, but a verb (at least, that’s how I would parse it).

No, in my example it is both a noun and a verb (at the same time); that is what a gerund is

Hi,
The discrepancy is that Vic is using Canadian English grammar rules, I’m using the A-mar-i-can grammar rule set. =)

I guess the main point for the moment as a coder is that I now have the ability to define somewhat more sophisticated grammatical rules.  This is an improvement over my previous approach where I just had a list of words.

[IN]
in
out
under

So, if any of the words above appeared in text then I would substitute the expression [IN] (preposition) for the word.

Now, instead of having only a list of words I can include the tag set with the word. E.g. ‘out [DT]’ instead of just the word ‘out’.  So now I might do this to define a preposition.

[IN]
in [DT]
out [DT]
under [DT]

or I can put the tag set before the word such as ‘[DT] beginning].

There are several other peculiarities that I want to allow for.

* Detect if a word is in the beginning or end of a sentence. E.g. “English Literature” vs. “The English Language”.  English is different POS (part of speech) in the two cases. Stranded prepositions may be in the last position of the end of a sentence.

* Include flag to indicate that a second word should exist.  E.g. ‘either’ in an ‘either-or’
sentence is a different POS then a sentence with ‘either’ but not ‘or’.

I’m looking through a grammar book and tag set now to find strange rules that I can try to implement a coding mechanism.

The goal, is to allow the bot trainer the ability to define somewhat complex grammar rules more easily.  Just a goal.

Jan,

It’s almost the same, but not quite, is it?

I’m not convinced that a bot needs to understand every nuance of a grammar.  The bot could interpret with its limited grammar…and then ask a clarifying or a confirmation question or wait to be corrected by the human. 

Regards,
Chuck

Chuck Bolin - Oct 4, 2010:

It’s almost the same, but not quite, is it?

I’m not convinced that a bot needs to understand every nuance of a grammar.  The bot could interpret with its limited grammar…and then ask a clarifying or a confirmation question or wait to be corrected by the human. 

Regards,
Chuck

Agreed.  The bot needs to know proper grammar and improper grammar, and there will always be cases were it won’t get it right, and simply have to ask.  Hell, humans have to do it all the time!

Victor Shulist - Oct 4, 2010:

It is important at the parsing stage to know a word is gerund and not simply a noun. 

The reason being is that a gerund can take a direct compliment whereas a noun cannot.

Example,

“I like chatbots”—direct object of predicate verb “like” is noun “chatbots”

“His job is painting houses” 

  direct object of predicate verb “is” is the gerund “painting”

Since it is a gerund and not simply a noun, it has its own direct object of “houses”

So, yes, it actually is important to know a word is a gerund and not simply a noun at the parsing stage even.

My parser would interpret “painting_houses” as a noun. (Well, it would also try calling “is_painting” a (progressive) verb and the two interpretations would battle it out.) I don’t know how important it is for parsing for the program to understand that “houses” is an object of “painting”. I’m planning on another set of tools altogether that relates nouns to verbs and adjectives to adverbs and so forth to determine if two words have similar meaning. That would be the level of interpretation that would find significance in “painting” also being a verb.

I’m not convinced that a bot needs to understand every nuance of a grammar.  The bot could interpret with its limited grammar…and then ask a clarifying or a confirmation question or wait to be corrected by the human.

Important point, Chuck!