Ai Dreams Forum
Artificial Intelligence => General AI Discussion => Topic started by: LOCKSUIT on July 14, 2018, 03:28:25 pm
-
I'm looking for a parser program that takes a sentence, and can find the smaller parts of the whole like this:
Cows are very cute and eat food.
/ \
Cows are very cute and eat food.
/ \ / \
Cows are very cute and eat food.
/ \ / \ \ / \
Cows are very cute and eat food.
Note that it should not output "are very", as shown above it should output that one as "very cute".
I've seen parse trees but I think they are made by a human expert in the background, not computer.
-
I'm looking for a parser program that takes a sentence, and can find the smaller parts of the whole like this:
Cows are very cute and eat food.
/ \
Cows are very cute and eat food.
/ \ / \
Cows are very cute and eat food.
/ \ / \ \ / \
Cows are very cute and eat food.
Note that it should not output "are very", as shown above it should output that one as "very cute".
I've seen parse trees but I think they are made by a human expert in the background, not computer.
I thought i showed you about the TRIE TREES..... https://github.com/spydaz/ClassTrie
its exactly what a trie tree can do! i have the example in my git hub just load it and experiment so that you can see how the data looks....
-
You can use popular Nearley.js (https://nearley.js.org/). It has a lot of plugins, including railroad diagram (https://en.wikipedia.org/wiki/Syntax_diagram) generator. It is being there since 2014 and it is being used in numerous projects such is Shrdlite (https://github.com/ChalmersGU-AI-course/shrdlite-course-project) that is an AI solution involving natural language processing. Of course, this applies if you are up to Javascript.
What do you need it for?
-
Pretty much any dependency parser.
Spacy: https://spacy.io/usage/linguistic-features
Stanford: https://nlp.stanford.edu/software/stanford-dependencies.shtml
-
Those are good resources but I'm not sure I perfectly explained my question.
I know these parsers from people / Google can parse it and find the dependencies but...that wouldn't be what I'm looking for...
I'm looking for something that finds the same pair no matter the sentence.
For example one day we feed the parser:
"Cows are animals that eat food."
and another day we feed it:
"I think in our rain-forest the animals that eat food would be the great lion"
In both cases it should output these words as a single feature made of 3 words:
"Cows are animals that eat food."
"I think in our rain-forest the animals that eat food would be the great lion"
Do these parsers do that if given separately those 2 sentences? It's not meant to remember the last sentence, just does the same output for em both!
The reason I'm looking for this is because I want a program for AGI that knows many sentences it reads plus the bits, that make it up ex.:
"I think in our rain-forest the animals that eat food would be the great lion"
....."I think" goes together to build bigger and is re-useable. So is "in our". "I think in our" is a bigger re-useable feature in language/the world. Next is "I think in our rain-forest", and so on...
Also if you know of this program, another thing it will be like is it must store millions of words and pairs of words, inside a non-ANN so each entry is perfectly stored...If you got this baby please show me the program please.
-
(read above)
Btw so spydaz, I read your git page there, and the picture shown, going down the left branch, has 3 optional things to say in a conversation right?? Either it'll say that, there, or this. Right? And if it were words stored in nodes it'd be like it can say either i want brown cookies, i want soft cookies, or i want red hams, right?
-
(read above)
Btw so spydaz, I read your git page there, and the picture shown, going down the left branch, has 3 optional things to say in a conversation right?? Either it'll say that, there, or this. Right? And if it were words stored in nodes it'd be like it can say either i want brown cookies, i want soft cookies, or i want red hams, right?
In addition to the "Word" trie i added Add iteratively
Add
The cat sat on the mat
Cat sat on the mat
sat on the mat
on the mat
the mat
mat
by "shifting along the sentence / Word / Paragraph" we get "All suffixes and Prefixes" as complete words/pathways as well as combining simular patterns
The dog sat on the mat
Would add "the dog"
the terms The dog / The mat / The cat <<<<<<< would all be in the same node!
but you need to load it to visualize it ..... Hence a treeControl .... you will need to download it and compile it in Visual studio to run the binary file (EXE).... the source code is there ... quite well documented in the code! mess about with it! give it to your coder! he should be able to convert it to python quite easy! ......
Trees are just lists containing sub-lists....
When you can see what is happening inside the data structure then you will see what your looking for ! compile it.... its pure vb "no references"
The class trie is not really for conversation(yet)
But it is such a usefull data structure "for addressing" as well as finding structure in sentences; Capturing simular structures in data .... just as you mention above....it groups the simular structures in nodes etc...Its used for string matching(although i wouldnt).....many uses.. the tutorials or lessons on using tries etc.... is very low class/small ideas/ highly underrated/underused
-
Errr....right now I'm trying to write the instructions up it'd becoming very alien-like for a simple idea and this would be a big time saver if your program is my program or better. But I can't tell! I doubt your program is doing what I think. But we must see! ... :D
Take this picture, see it first before reading.
As you can see, the branch going down and to the left is using TWO re-useeable snippets, and at the bottom I show these in deeper pair layers combining into a bigger, re-useable, snippet ! No word or word pair is stored twice ever.
HENCE, I will now ask, in your program, would "the old cat was" be made up out of ex. "the old" + "cat was"? And would ex. "the old" be made up out of "the" + "old"?
It's not supposed to shift along though...for example in a beginning hierarchy like shown at bottom would-not store "old cat" no, only "the old" and "cat was" should be stored and referenced to create "the old cat was".
So does it do that ?
But why can't you just sceenshoot it running for us all lol? Please?
-
Errr....right now I'm trying to write the instructions up it'd becoming very alien-like for a simple idea and this would be a big time saver if your program is my program or better. But I can't tell! I doubt your program is doing what I think. But we must see! ... :D
Take this picture, see it first before reading.
As you can see, the branch going down and to the left is using TWO re-useeable snippets, and at the bottom I show these in deeper pair layers combining into a bigger, re-useable, snippet ! No word or word pair is stored twice ever.
HENCE, I will now ask, in your program, would "the old cat was" be made up out of ex. "the old" + "cat was"? And would ex. "the old" be made up out of "the" + "old"?
It's not supposed to shift along though...for example in a beginning hierarchy like shown at bottom would-not store "old cat" no, only "the old" and "cat was" should be stored and referenced to create "the old cat was".
So does it do that ?
But why can't you just sceenshoot it running for us all lol? Please?
Obviously its all seeing and all doing!
-
Is there a visualization that my coder will awaken? If I can see by the visualization in like 5 minutes then it makes no sense why that can't be recorded footage. If there's no visualization, I can't understand code. Makesy, noey, sensey....
-
LockSuit, you can code a parser that does that, same way you can do many different ways, so why dont you god damn do it!
Weve been kind to you for a long time, but you have to be a developer to get real respect, otherwise your just tagging along, number due.
-
Is there a visualization that my coder will awaken? If I can see by the visualization in like 5 minutes then it makes no sense why that can't be recorded footage. If there's no visualization, I can't understand code. Makesy, noey, sensey....
i did say i designed an interface with visualisation too... just input words or sentences in your case
the cat was =
(http://)
-
only three word sentences .... but the tree contains grouped information.
The cat was
The dog eats
The cat eats
The cat sat
These were added By parsing forward...
A different but similar structure can be defined by parsing backwards....I wont do that experiment!(blocks your own learning journey)(you will have to do you own experiment because actually (only i am doing this project at the moment too)...
As the data structure is loaded it can clearly be seen the value of the tree, As information pertaining to the dog is contained the the dog node , respectively data pertaining to the cat is in the cat node....
Data capture does not reshape the data in any form it captures data with high integrity. providing linearity, the data can be clustered or sequenced.
Obviously this can be further advanced ... or diversified ... as its a technique that has high possibilities for DNA searching as well as finding structures strings such as PI... as a it must be impossible for a number not to be followed by another number such as 1-9 multiple times... how many times is a particular group of number occurring? defining a sentence without grammatical means can be done with such a technique as using an insertion process.... yet with the trie tree structure the shape of the structure is dynamic to the data stored in a structure therefore ... its possible to use the structure to "normalise" components in unstructured sentences structures...
Again possibilities are endless its not only imagination which can drive you "LOCKSUIT" but you need to experiment as well build for your self ... people often give you answers but its just ideas for you but no potential for action because of methodology.. so in reality you cant know if something can really work? How could you? unless you do the test
you-tube is often well faked! so you will be led in to belief structure which although lovely and seem logical ... the the real knowledge we know it to be SCI FI! (those who know god would know the same problem with you-tube) lots of misleading stuff...Follow the right path to glory! (MAYBE)
PS: its taken me this long to figure out how to add an attachment on here....tonight was that night
ALSO: i only generally "Draw diagrams when im on the train ... or in class" usually i generate one or use an application as it also helps you understand requirements previously unthought of!
-
oh hey spydaz cool, wait on the forum gonna ask you a question in 1 minute
-
Ok so you got: see attachement
But why the duplicate the dog jumps and the cat was whyyyyy duplicatess noooo lol why
-
But more importantly, does your program prune away uncommon features like "then were" in "they ate and then were all happy"?
-
the numbers refer to the node level(Depth) ... if i expand those nodes you will see the "STOPCHAR" this is for the end of sentence marker.... therefore from the root node to the "StopChar" node is a complete sentence. or ( "SUFFIX" all other prefixes are contain within its own suffix.....)
-
But more importantly, does your program prune away uncommon features like "then were" in "they ate and then were all happy"?
If a node STopChar is at level 1/2 then obviously its just a term which has been mention in some string...
But if a node has 2/3 node then it must contain some structural data... if a branch has many sub branches then that branches data all must be structurally related... And so forth
as with the cat node containing all the different actions the cat took ... and dog the dog took... the single nodes had no sub-nodes so they must be some subject which is being referenced by an object (note the single item deteminer(the) Contains dog and Cat nodes) (noun phrase detection)
-
Can it connect the top node "the cat was" to one far below by a link...? Like this:
Else it can't re-use them to make 1 bigger thing
-
But more importantly, does your program prune away uncommon features like "then were" in "they ate and then were all happy"?
If a node STopChar is at level 1/2 then obviously its just a term which has been mention in some string...
But if a node has 2/3 node then it must contain some structural data... if a branch has many sub branches then that branches data all must be structurally related... And so forth
as with the cat node containing all the different actions the cat took ... and dog the dog took... the single nodes had no sub-nodes so they must be some subject which is being referenced by an object (note the single item deteminer(the) Contains dog and Cat nodes) (noun phrase detection)
All actions under the ALL determiner would group under all... this is why its useful for "Merging sentences" so now a paragraph of text can have defined structure and meaningful ngrams not brute force ngram creation... remember
we are building techniques which AI can utilise for unsupervised learning. (without training) ..... But also maybe can correct itself (clean bad data) (Success / Failure (reinforcement learning) from self built historical data.... <<<<<<<<<<< See the intelligent algorithm.. collects(unsupervised) / (success / failures (probabilistic / Statistical methods)/ Bayesian methods ( cleans) / refines collection / (learns) <<<<<Self supervised learning? Is that algorithm aware maybe of its own errors and failures and attempts to correct itself thereby learning and improving itself based on experience collected and anyalized (thinking)....Deciding .... Hmm very intelligent for a simple Process? maybe?
-
Yeah I learnt that yesterday, that nodes with deeper structure means it isn't stuff like "were then".....it can't combine it to make "were then is like"! but wait what if it makes "were then all happy"? Oh oh, then it thinks the lower smaller snippet "then were" is a good thing!...
I mean if it was "were then the gold" in "they ate but were then the gold winner and they didn't want to be" then it'd link together 2 bad uncommon pieces forever...and can it go past 2/3 nodes and be a nuisance? "were then the gold winner and"....yes?....
-
Can it connect the top node "the cat was" to one far below by a link...? Like this:
Else it can't re-use them to make 1 bigger thing
now when designing an Algorithm you have to set your RULE for the algorithm ;... when presented by your case data it is tested for MATCHES based on your STRATEGY then a selected action can be taken (old time game strategy)
The Inference engine
if that what your rules define as a "Legal" move then yes you can .... Construct a New branch by adding the Sub node to the other sub node at the desired point nodes can always be copied or moved.... if that what your rules denote then you can write a function to perform your action
-
oh just a fix on my pic of your pic:
-
But does *your* program combine 2 distant snippets like the red line I drew shows??
-
But does *your* program combine 2 distant snippets like the red line I drew shows??
All seeing all doing remember!
Don't forget to read the posts.... diagrams cant help you .... right now ....
-
So what do you do about if your program reads:
"We were then all tired and sat in the old lobby at the old school and then were all too cold to move."
and saves
"lobby at the old school and"
like this in your format (see image):
-
(see above)
in that image, it does have more than 2/3 nodes, a structure, yet, it is a bad take from a sentence, it'll never be re-used, and yes it got re-used by the word parts that followed but, that's all it'll ever get! It'll ass tons of clutter and false real world knowledge....slow searching...etc!! For a AGI..
-
oh and, is your tree saving them like on the left or the right?
I'd say the one on the left is really bad but hmm wait, thinking...if it scans many sentences and learns structure like that, um, it may learn "lobby at the", "at the old", "the old school" in its lifetime, and, and, no this is horrible clutter...
-
in that image, it does have more than 2/3 nodes, a structure, yet, it is a bad take from a sentence, it'll never be re-used, and yes it got re-used by the word parts that followed but, that's all it'll ever get! It'll ass tons of clutter and false real world knowledge....slow searching...etc!! For a AGI..
Also, That's also bad for re-use, again, it'll never be reusing those features...
- because it's doing the left side example, not the one on the right i show.
-
So what do you do about if your program reads:
"We were then all tired and sat in the old lobby at the old school and then were all too cold to move."
and saves
"lobby at the old school and"
like this in your format (see image):
There are a few valid shapes in there but the sentence is an irregular sentence and would need pre-processing such as changing the "WE" to a valid noun or object/subject as well as the and being replaced by full stops... a i say "bad data in bad data out" there is no real data in that sentence only snippets... perhaps pre-processing would give better results....even this irregular sentence has "missing data" so its contents wold not be pre-processed without the related sentences from the paragraph.... this is why a lexer / tokenizer / rule-set is required/// to root out the bad data and block data collection... intelligence.... smart code!
-
for example to my above posts
it'd store:
"lobby at the old school and"
"lobby at the old school"
"lobby at the old"
"lobby at the"
"lobby at"
"lobby"
.......these are strange uncommon cuts from some sentence, plus, as it adds ONE word at a time, it's sure to add clutter and non-reuseable features.
-
in that image, it does have more than 2/3 nodes, a structure, yet, it is a bad take from a sentence, it'll never be re-used, and yes it got re-used by the word parts that followed but, that's all it'll ever get! It'll ass tons of clutter and false real world knowledge....slow searching...etc!! For a AGI..
Also, That's also bad for re-use, again, it'll never be reusing those features...
- because it's doing the left side example, not the one on the right i show.
Again for a three word sentence ? what ngram actually contains the subject or the predicate or the object ? .... the answer is in the question as usual...
a) 1
b) 2
c) 3
d) 4
Everything depends on YOUR level of understanding....
-
oh and I mean that that is taken from a larger sentence like this
.....................[...............]...........
. = word
it won't know what to combine together from a sentence....it'd do all combos lol
-
for example to my above posts
it'd store:
"lobby at the old school and"
"lobby at the old school"
"lobby at the old"
"lobby at the"
"lobby at"
"lobby"
.......these are strange uncommon cuts from some sentence, plus, as it adds ONE word at a time, it's sure to add clutter and non-reuseable features.
This is easily done ? its backward parsing.... Recursively....i did hint at this in the diagram (ADDED ITERATIVLY) <<< Forward parsing....
Not really needed but should be a function to be added... but the first suffix contains all your sub suffixes (they are prefixes of the first suffix) ?
Good to ask.... but these i mentioned in my early text messages
-
oh and I mean that that is taken from a larger sentence like this
.....................[...............]...........
. = word
it won't know what to combine together from a sentence....it'd do all combos lol
A term document matrix to the applied Suffixes will show you the most valuable suffixes.(using node depths)....essentially creating a probabilistic Bag of suffixes<<<< again the algorythm mentioned before<<<<
I know your wondering how to use the outputs as inputs to your Neural network... this should be a forum project for you..... using the bag of word model for the neural network.... (its out there)
I asked one of the lecturers why they always set project which had been done before.... She said "im sending them on a learning journey"
-
You mean a frequency matrix?
-
You mean a frequency matrix?
A term document matrix to the applied Suffixes will show you the most valuable suffixes.(using node depths)
A trie tree cannot hold the same suffix twice?? O0
It may occur twice if that be the case you may choose to record its "position in the node" <You could use this as a "RULE" <<<
When designing your requirements for your developer its best to specify your desires as rules which can the be applied to your insertion strategy or evaluation Strategy or pre-processing strategy... this is why its good to brainstorm an idea.. to define your rules and requirements... when actual development takes place compromises may have to be made or even upgraded data structures maybe required to accommodate rules or strategies
there are many rules which we could define to determine the value of a suffix! as it would then be the value of the suffix that would be the important outcome. Which could then be selected based on suffix length/suffix position/ suffix contents for sentence generation or construction .... Statistically
Right now we should be using an inference engine based algorithm and a natural language processing methodology based with machine learning / statistical prediction .... using rules and strategies to determine if patterns are relevant or irrelevant taking actions based on statistical and probabilistic historical models... ... we have generated and processed some random sentence into some valuable data for use in output functions from learned input functions..... Based on some input T and some learning factor C a new output is generated E <<<<<<<<< sounds like AI
-
you say depth = it is not silly pairs of words ?
but as shown in this image I drew, it has great depth but bad snippet from a larger sentence:
-
also, what if it hears a word or word pair ex. "complex AGI" 1 or 3 times in the frequency measure, that doesn't mean to delete it, it doesn't mean it is common or not, or does it hmm, it may see that "we are" appears many times (448 times), while "are all" appears fewer times (30 times), and "lobby at the old" appears once um, not sure if that only keeps the common pair features of sentences...
-
This is you tree added
The branches with more than 1 level contain data ... the most data is contained in the branches with the highest nodes....
A pattern after testing multiple sentence shapes and adding into single trees may show that 3ngram sentences generally contain Determiner,Noun phrase... etc... such pattern would be learned over time if that was one of the objectives that the strategies were trying to fulfil.
The strategies bring you closer to your objective... even strategies used can also be measured and strategies can be deemed irrelevant as well ranking the strategies by their application can reduce code execution speed therefore increasing the speed of "Thinking time"..... A potential objective of the algorithm to improve itself...refining it processes.... this may be applied to document processing learning or total learning process. (refinement)... reinforcement learning.... choosing winning strategies for recognised opponents ...applying a basic wining strategies or learning process to learn the opponents strategies and create a combination of strategies from existing rulest to win
-
hmm you say frequency measure and depth nodes, well, that may mean "the" appears 100 times, "the cat" appears 50 times, "the cat sat" appears 20 times, and "the cat sat but" appears 2 times. But this happens for ALL bigger features like any darn sentence ex. "I work on AI all night but went skydiving" only appears once or so in meaning-wise, same for that weird sentence built up from a larger sentence "lobby etc etc", and no not for short things ex. "the" will be common and even the stupid uncommon "are then" will appear more yet less, maybe we should compare it to its peers ex. "the object" appears 444 times while "are then" appears 35 times, not sure that proves its good to trash though. So, frequency+depth nodes um, maybe combine the 2 peer reviews!?:
the object - 400 times
lobby at - 40 times
+
the object is so cool - 40 times
lobby at the old - 2 times
=
440
42
-
That's cool spydaz. Yes I realized the first dotted line going down is a new start at each word ex. you get "at the old school". Hmm, but this is saving all the possible combos...sure it'll end up saying:
lobby at the old - 1 times
lobby at the - 5 times
lobby at - 7 times
lobby - 22 times
at the old - 20 times
at the - 300 times
at - 400 times
the old - 40 times
the - 400 times
old - 100 times
but um, hmm let's try adding the sum of frequencies in these depth nodes....calculating/thinking.........
Ok so the algorithm could see that old and the old and at the old and lobby at the old add up to:
161 times
features containing the:
766 times
at:
733 times
lobby:
35 times
the old:
61 times
at the:
326 times
at the old:
21 times
lobby at:
13 times
lobby at the:
6 times
lobby at the old:
1 times
-
This is you tree added
The branches with more than 1 level contain data ... the most data is contained in the branches with the highest nodes....
A pattern after testing multiple sentence shapes and adding into single trees may show that 3ngram sentences generally contain Determiner,Noun phrase... etc... such pattern would be learned over time if that was one of the objectives that the strategies were trying to fulfil.
The strategies bring you closer to your objective... even strategies used can also be measured and strategies can be deemed irrelevant as well ranking the strategies by their application can reduce code execution speed therefore increasing the speed of "Thinking time"..... A potential objective of the algorithm to improve itself...refining it processes.... this may be applied to document processing learning or total learning process. (refinement)... reinforcement learning.... choosing winning strategies for recognised opponents ...applying a basic wining strategies or learning process to learn the opponents strategies and create a combination of strategies from existing rulest to win
Therefore only a jedi can defeat another jedi....Only an AI could defeat another AI
Using the same strategy essentially finding a strategies to defeat itself.... em AI WAR
-
see my edit above XD
ok so then re-arranging those numbers in order of highest to smallest frequency we get:
the:
766 times
at:
733 times
at the:
326 times
old:
161 times
the old:
61 times
lobby:
35 times
at the old:
21 times
lobby at:
13 times
lobby at the:
6 times
lobby at the old:
1 times
-
and if we look again at:
lobby at the old - 1 times
lobby at the - 5 times
lobby at - 7 times
lobby - 22 times
at the old - 20 times
at the - 300 times
at - 400 times
the old - 40 times
the - 400 times
old - 100 times
we can see that in comparison to the add-ups, well,:
at the old:
21 times
lobby at:
13 times
lobby at the:
6 times
lobby at the old:
1 times
are pretty low common frequency
Hmm maybe I did this wrong..........................the single words are good ok but I have the in more pairs than lobby.....
I think we can conclude though that.......um............
.........But wait no I said earlier to compare small pair to small pair and big pair to big pair....obviously if we did that we'd see similar number of lobby and hat or tiger, but less of lobby at the old and more of the red person.
Hence some things are just not common, and but may be good, algorithm no worky...
-
I'll draw a sketch of what your algotithm would look like of that pic in my format...
-
So here we save all the possible pairs linearly up on an angle i mean lol and um, 3 sentences like this, which is, bad, it should be triangles made of triangles like shown before and not every shift over like "and the cat" "the cat ate" "cat me but".
-
So here we save all the possible pairs linearly up on an angle i mean lol and um, 3 sentences like this, which is, bad, it should be triangles made of triangles like shown before and not every shift over like "and the cat" "the cat ate" "cat me but".
Answered!
-
ur ruining his design spyda!!!!
-
lol ranch. Honestly I don't think his program is what I need. That'd be hilariously easy and time saving and powerful if it was.
Again, why does it learn "word word [word {word] word} word" (the [ ] or the { }?) over one or the other order? Is it more common/useful? (Either "the lobby at the school" or "lobby at the school was").
It can't store both possible shifts. And when it does store 1, it must re-use it for sentences containing it.
It'll always be quick to point out/see it apart in any sentence (or build using it) ex. "Ya so that girl was at the [lobby at the school] but was tired" and ex. "The hat was at the [lobby at the school] but was missing all week long"
-
ur ruining his design spyda!!!!
You cant force good knowledge into a head as the say! -
La-La Land is a real easy place to live -
"I just felt like running! -(Forrest Gump)"
I was explaining a whole strategy... for designing an algorithm...
And how to mould products such a the "Simple Trie Tree" to your will... but being a non coder he asked so many questions i was up all night ... trying Explain basic Functions / Strategies / Objectives / Rules / Matches .... one of the first level AI ... techniques...
We even discover a super intelligent algorithm process for understanding that data collected needs to be cleaned then categorised before farming information and pre-processed.
its not instant ! there is no solve all and fix all .... you have to Taylor everything to your needs... or your just using peoples projects as "BLOCKS" in your AI process. hoping for a big discovery!
-
Lol I never asked you to explain Ai techniques, I just wanted to know how your program works / if it did what mine requires...
Well obviously I didn't understand what you said then...
-
Lol I never asked you to explain Ai techniques, I just wanted to know how your program works / if it did what mine requires...
Well obviously I didn't understand what you said then...
Obviously ; the question was answered; Perhaps your just taking the michael... in future we shall know not to explain anything to you... as the answer was presented but you couldn't take it on board....Perhaps even now you don't get it!....More research for you is required.... we shall keep you to the basics in future!
Pmlamo (peeing myself laughing my ass off) Lol (laughing out loud) <<<<<<<<<<< Breaking it down
Now we know your not serious!
-
On expedition now Later guys.... Chat soon!
-
Ok.
Which post is the answer in?
And anyone care to shrink it into a 15 word sentence?
-
Lock, you have to learn to learn. It can be helpful occasionally, trust me. Sometimes it just pays off to spend an hour or more in learning what other people intended to show you. Often, you can base your own discoveries on what you've learned. There will always be some pointless stuff for you, but generally, it pays off at the end.
There is some nice stuff out there in outer world. Use it. Expand it, or optimize it. And then ask us to learn from you.
There is no such thing as a self made man.
-
As the thread went along I had read all of his posts.
Ya but ivan, there's 54 posts now...Did any of yous get my desire or his answer? I think we BOTH had a language barrier.
Hopefully yous understand the program/his answer, I need to ask some better questions...
-
Another way to put it simply is I want it to learn by lots of text (or some universal verb/object/predicate structure) that "are is" is BAD and "is soft" is GOOD, i.e. it outputs back out your sentence like this "Another thing they are is soft like cats.".
I have not heard anyone mention/reply back on this subject !
-
Another way to put it simply is I want it to learn by lots of text (or some universal verb/object/predicate structure) that "are is" is BAD and "is soft" is GOOD, i.e. it outputs back out your sentence like this "Another thing they are is soft like cats.".
I have not heard anyone mention/reply back on this subject !
There are grammarc natural language rules rules that say whether a sentence is well formed. A part of such grammar rules may be written in BNF notation (https://en.wikipedia.org/wiki/Backus%E2%80%93Naur_form), but if you want a real deal you'll need a link grammar notation (https://en.wikipedia.org/wiki/Link_grammar). There are a lot of programs that use either a kind of BNF or link grammar to parse texts, but you have to provide them a valid grammar. That grammar could be anything, from grammar for parsing math expressions, over grammars for parsing programming languages, to grammars for parsing natural language sentences.
Some parsers like Stanford Parser come bundeled with a natural language grammar (see references from Don Patrick (http://aidreams.co.uk/forum/index.php?topic=13251.msg54045#msg54045)), but generally, natural language parsing problem stays unsolved problem due to ambiguities that raise by overlapping natural language grammar rules. An example of this is:
Last night I shoot an elephant in my pajamas.
The question here is whether the elephant wears a pajama or the shooter wears pajama. However, there are hints to solve such problems, like `Last night` and `my pajams` parts of the sentence that hint that the shooter wore a pajama, but this is still not a guarantee. Ambiguity resolving could also be done by comparing near sentences from the same paragraph, or by statistical analysis, but it still remains open question in linguistics, which is hoped to be solved when AI would be invented. There are even some contests (like winograd schema challenge (https://en.wikipedia.org/wiki/Winograd_Schema_Challenge)) in which competitors (their programs in fact) try to resolve as much ambiguities as they can. I think no one ever made it with 100% correctness. Questions posed in such competitions are like:
A cat didn't enter a box because it was too big. What was big, a cat or a box?
This is a big problem, to answer this question, isn't it? Well, it gives programmers a lots of headaches.
To analyze natural language sentences, some efforts have been made by parsing a sentence corpus by humans, and packaging it in treebanks (https://en.wikipedia.org/wiki/Treebank) freely available for download in various versions in various languages. Treebanks could be used as references for checking correct solutions in solving a natural language parsing problem by a machine.
Does this answer your question?
-
Did you hear about Parsey McParseFace? The ambiguity problem is OVER. Human accuracy dude!
https://ai.googleblog.com/2016/05/announcing-syntaxnet-worlds-most.html
give it a test:
https://deepai.org/machine-learning-model/parseymcparseface
I found a link grammer tester looky! Also AbiWord I mayyy have to try it out too then...Now how does it know though!!!!???
http://www.link.cs.cmu.edu/cgi-bin/link/construct-page-4.cgi#submit
Did the programmer dude label each freaking word in English whether it's a noun, verb, etc? That'd take a long time, including how it knows WHICH can fit together....how does it know "I am" "soft like" don't fit together yet "I am" the cat" fit together for a rough example?
If I try AbiWord and it underlines wrong word pairs, this tells me it KNOWS which words can go beside each other.....
https://www.abisource.com/download/
sorry not available anymore wtf you kidding me abiword!!! LOL pft
-
It looks like Abiword's source code is right there: https://www.abisource.com/downloads/abiword/3.0.2/source/
Parsey doesn't handle ambiguity at human level though. It only does well on "well formed text", and the score is not quite what it seems: individual dependencies between words includes all the times it gets "and" and "the" right. Its score will be drastically lower when you look at entire sentences: 95% accuracy means 1 out of 20 words will be wrong, and with that it will get one out of every two sentences wrong. It uses statistical probability to handle ambiguity.
No sole programmer needs to label nouns manually, they can just use existing dictionaries. Parsers produce parse trees, and dependency parsers not only tell you which words are nouns, but also which group of words belong together as a noun phrase like "the black cat". Simply put it looks for valid combinations of e.g. determiner-adjective-noun, predetermined sequences. It also marks the main verb so you know to keep that separate. However, which parts of the parse tree you select to output is up to you to program.
there are hints to solve such problems, like `Last night` and `my pajams` parts of the sentence that hint that the shooter wore a pajama
I'm curious, In what way do you consider "last night" a hint? I handle the elephant/shooter in pajamas ambiguity using knowledge of their sizes, among other things. The top score in the Winograd Schema Challenge was 58%, though people have come to claim up to 70% success since, but that is of course while already knowing the questions.
-
there are hints to solve such problems, like `Last night` and `my pajams` parts of the sentence that hint that the shooter wore a pajama
I'm curious, In what way do you consider "last night" a hint? I handle the elephant/shooter in pajamas ambiguity using knowledge of their sizes, among other things.
Last night? People usually wear pajamas when they sleep at night (besides that they generally wear pajamas, sometimes during day), while elephants are not wearing pajamas neither during day, neither during night, unless someone is trying to make a fun out of a poor elephant, which doesn't depend on day time. So, human wearing pajamas depending on night versus elephant wearing pajamas not depending on night is what may increase a probability that it is a man who wears pajama. At least, that is set in my NN fuzzy mind, I'm not sure about the real probability proof. I could be wrong, of course.
As for NN parsing, I'd bet on NN picking the right combination, considering the whole context, part of which is knowledge of our world. But I'm curious of a crucial thing: is NN capable of hosting what is considered to be loops in algorithms? I believe everything could be described in terms of semantic tables, and I believe NN handles these well, but what if there is a loop inside a semantic table, a recursive reference in a cell that points to a parent semantic table? I wonder if NN is able to describe such a thing. If it can, I'd dare to say that NN language is a Turing complete language, and I'd raise my bets in NN handling correct parsing.
[Edit]
Fixed-point combinators (https://en.wikipedia.org/wiki/Fixed-point_combinator) make me believe that NN is Turing complete. Y combinator in particular can be used to represent a recursive functions in languages that don't support recursion at general level.
[Edit2]
But I think that NN is a brute force thing, there could pass centuries until it clues up the right combination. It would be interesting to see how the thing behaves after years or decades of unsupervised learning. A learning curve may raise exponentially, considering our species.
-
Does this do what abiword's link grammar does in showing you which is incorrect? http://www.link.cs.cmu.edu/cgi-bin/link/construct-page-4.cgi#submit
see attachements FIRST...it looks like it may idk !
Else > Does anyone have a Unix system here? I looked at building AbiWord tar file on Windows and it's a long list of steps, hence I'd rather like it if someone could screen-capture a test in AbiWord for me.
I want you to take several sentences I'd write and see what it underlines in green underlining with link grammar turned on.
I seen an image on Wiki showing 2 short sentences but I need to see more examples!
I need abiword XD.............the thing in it......
-
Ok now I need to know why, see attachment,
1) why does it return 20 possible linkages for this sentence and not 1 correct linkage? That defeats the goal, I could do that too lol!
2) why does, in my screenshot, the bottom linkage combine so so so many words on the left as 1 big long piece and only at the right end does it BUILD bigger? While the top linkage actually is building with the little atomic parts of the sentence? Is it in actuality building at EVERY + symbol sign? Or only when it goes up a level with the + sign ABOVE the below? That would fatally wound my goal if it thought those first 10 words were 1 part/utility/tool/fact at the smallest level and not "he said" or "he" !
-
It seems to me you do not understand the basic concept of hierarchy in parse trees. Do some reading.
https://en.wikipedia.org/wiki/Parse_tree
Ivan: I would think that if an elephant were to wear pajamas, it would also do so at night because that's what pajamas are for. If a program already has the knowledge that humans wear pajamas, or that elephants don't, then the time of wearing doesn't add anything in my view. Still, it's interesting that a specification of time could be a hint in other scenarios.
NNs typically do not loop, they just branch from input to output. However, Google set one up to feed the output back to the input in their Deep Dream project and this led to "hallucinations", where initially faint patterns are enhanced. http://www.clowndotpy.com/deepdream
This is not recommendable however, as it leads the machine to see things where there aren't. It's like you have a statistical method that's 90% correct, and then you pass that result through the same NN, then you get 90% of 90% correct, which is 81%, etc, etc. NN loops become less accurate as far as I understand.
-
So parse tree hierarchies works by rules, which evolve around/on the given sentence and doesn't apply for all sentences that contain "we were" in them? For example, two very structurally different sentences contain "we are", and as the parse rules say ah Ss ah ts ah those two make Mv and so, it might see the other sentence very differently and won't always single out "we are" as a smaller atomic part in both sentences? In sentence 1 it may have the smallest part comes out to be "and we are here" while in sentence 2 (separate sentence) the smallest part may come out as "we are next"?
-
It seems to me you do not understand the basic concept of hierarchy in parse trees. Do some reading.
https://en.wikipedia.org/wiki/Parse_tree
Ivan: I would think that if an elephant were to wear pajamas, it would also do so at night because that's what pajamas are for. If a program already has the knowledge that humans wear pajamas, or that elephants don't, then the time of wearing doesn't add anything in my view. Still, it's interesting that a specification of time could be a hint in other scenarios.
NNs typically do not loop, they just branch from input to output. However, Google set one up to feed the output back to the input in their Deep Dream project and this led to "hallucinations", where initially faint patterns are enhanced. http://www.clowndotpy.com/deepdream
This is not recommendable however, as it leads the machine to see things where there aren't. It's like you have a statistical method that's 90% correct, and then you pass that result through the same NN, then you get 90% of 90% correct, which is 81%, etc, etc. NN loops become less accurate as far as I understand.
I try to think of it as the "Subject/Object" >> is "Action" << "Subject/Object" in this way it matters not if the subject is an elephant and the object is pyjamas.... the first aim is to collect the components and structure.... If an Elephant can or cant wear pyjamas can be handled by a later component which deals with the logic of "Can an elephant wear pyjamas" as unless both of the "noun" objects/Subject have entanglement information collected....it must be TRUE, until a contradictory entanglement presents itself for probablisitc reasoning to begin... or even another confirmation of the FACT elephant is wearing pyjamas ?....
The important factor is the Truths which can be determined from the sentence ; and the extracted subjects/objects/actions Which enable higher level scripts to process the collected data.... as Parsing a sentence etc ... if all components are not discovered... it may not even be "Saved" as its incomplete record. but it may still be processed to return a response.. or extract some meaning.
-
Post #65 - am I right?
Yes I realized myself that before a sentence can match a 'node', it must first cleans the data, like "alice whom was dieting drove a car and saw bob" or "Last night I shot an elephant in pajamas." needs re-arranging or known facts like elephants/animals don't wear pajamas, they could! but don't, last night and humans wear pajamas matches humans wear pajamas at night. The aim though next is to match a memory like the very sentence itself, the understanding part is already done.
-
Here is what I was thinking...
http://aihax.com/analyze/
-
A cat didn't enter a box because it was too big.
Conditional Sentence(Propositional Logic)
Implied IF / BECAUSE
A cat didn't enter a box < Premise>
(a cat<Subject>) Did not enter (Predicate) (a box<Object>)
because
it was too big. <Conclusion>
It)- Subject (was too big) (Predicate)
What was too big, a cat or a box?
(Subject = A cat,It)
Potentially !
The important factor is in the normalising of the sentence as we know, the First named subject, Enables for the Identification of the following Identification as "She/He/They/It" .... these factors in a joined sentence...(complex sentence rules).... for grammatically correct sentences.
When saving the data collected from the sentence ... the data base should not have references to he or she which should be replaced by the named object/subject... usually confirm able as a noun/ noun phrase etc
-
So parse tree hierarchies works by rules, which evolve around/on the given sentence and doesn't apply for all sentences that contain "we were" in them?
Yes. Sentences with different structures will produce parse trees with different parts. The smallest parts are always single words. You'd probably want the medium parts like main verb (VP) and noun phrases (NP), but I don't believe you've yet mentioned what you need any of this for, at least not in language that I understand.
"A cat didn't enter a box because it was too big."
My program presumes that the cat did not enter because it was intimidated by the large size of the box. It would be a different matter if the cat could not enter the box, as opposed to did not. Anyway, no need to explain to me why syntax parsing is an important preprocessing phase, but sometimes the syntax is ambiguous itself and requires disambiguation during parsing.
-
Thanks for confirming that Don! So wait, it shows the structural BIG parts, the smallest parts (single words), but not some of them like "blah blah"? Ex. it outputs "blah blah blah blah blah" and of course we know the single words "blah" but doesn't show the second next most smallest parts "blah blah"?
I need this because I need my AI to know the smallesssst parts to even know what a "blah blah" means, then it needs to know what "blah blah" and "blah blah" mean to know what "blah blah blah blah" means, it can't just understand one big 16 word long sentence! And the bigger stuff is made out of the smaller stuff and doesn't store "I am" twice. It has to know the fundamental features of the real world, like water drops, gravity pulls, I am, not "swimming I" or "but water" that can be found in bigger sentences! Ex. After swimming I left the room. I could enter them in manually but that's laborious. Right now I have 4 valid ways to do the same thing: manual, link grammar (yep still 'works'), a backprop method, or a universal discovery method.
Yeah I'm using Parsey McParseFace hopefully before it reads sentences. Once text is re-arranged or clarified that I was wearing pajamas it'll answer back knowingly, it can then parse, and find the real parts of the sentences like "I am" but not "swimming I".
-
"A cat didn't enter a box because it was too big."
My program presumes that the cat did not enter because it was intimidated by the large size of the box. It would be a different matter if the cat could not enter the box, as opposed to did not. Anyway, no need to explain to me why syntax parsing is an important preprocessing phase, but sometimes the syntax is ambiguous itself and requires disambiguation during parsing.
I am currently researching the usages of the Could/Would/Should Relationships for gaining greater sentence understanding/meaning.... (modals)... i'm glad the phrase did not contain that added complexity...as of yet i have no handler for those sentences implemented yet either...
-
Here is what I was thinking...
http://aihax.com/analyze/
Results of analysis:
A_cat_didn't_enter_a_box_because_it = too_big.
What would be the next logical step in the algorithm?
Results of analysis:
What_would = the_next_logical_step_in_the_algorithm?
-
Here is what I was thinking...
http://aihax.com/analyze/
Results of analysis:
A_cat_didn't_enter_a_box_because_it = too_big.
What would be the next logical step in the algorithm?
Results of analysis:
What_would = the_next_logical_step_in_the_algorithm?
(Not all of the algorithms return perfect information (hence the need for multiple algorithms for different functions)) ... Incomplete..... or not valid for current interrogation? (a Check would be required to test the captured results to validate the algorithm is capturing valid data)
I was thinking about this also ; as there is still information not taken from the sentence; Probably as relational Subject / Object inferences.....
A_cat_didn't_enter_a_box_because_it = too_big.
A cat <is bigger than> A box....................
The problem with information extraction is to understand if the knowledge is temporal knowledge or long-term knowledge also another step.... because the predicate that the cat is "bigger" than the box is temporal knowledge and Should not be saved as a universal truth or long-term knowledge. Its hard to identify that this is a temporal statement. and yet by the Loose nature of the statement IE: He said / She said / It said clearly denotes that this knowledge is probably "conversational in nature" Specific to the HE/SHE... OR Event/Scenario also temporal.... (after reading i would hope that a sub function captured the fact(long Term) The Cat is Bigger than "Some" boxes) <<<< always the case (we notice even more entanglements) <<< Isnt English so complicated?....Very musical (great for composition)
Information becomes universal if it holds true for all possibilities for the even/Scenario Also hard to recognise. (from the sentence or paragraph structure)... yet again it also goes towards recalling the event or scenario? it becomes a case based reasoning argument. now sessions become "cases"to be stored as atomic argument related to their temporal moment in contrary to a factual statement to be incorporated in to the Data warehouse/ knowledge base. (other functions may store Full uniformed data, whereas another set of functions serve as temporal functions, such as temporary knowledge trees or Queries)
Also the posed question ; is regarding the "case" and not stored factual information ; as cases can be "fictitious scenarios"....
There is often always more information to be retrieved from a sentence... Now when executing a script the time it takes repeats each new function execute speed. now its time for some Parallel Programming. the same sentence may be parsed many times by many different functions each retriving different structures of information (although some items may be the same or repeated)
-
Hey spydaz, is this what your idea does/is? See attachment. It hears a sentence, and, when it hears another sentence with some similar stuff - it says hey these two got the same beginning I know where to splice them up at. ?
-
Hey spydaz, is this what your idea does/is? See attachment. It hears a sentence, and, when it hears another sentence with some similar stuff - it says hey these two got the same beginning I know where to splice them up at. ?
The reality is that you can choose the rule-set you require; for the task that you need. As in your diagram ....lol... The tree could split sentences every second word... why not .... it still would produce some interesting sets of data into some kind of interesting structure.
-
I pretty much verified my standing in some sketches in the past few days that your program (or related ideas) cannot do what I'm seeking for.
Your program/idea cannot learn to make a splice here "The thing they are | is cats.". Tell me why your program would avoid the splices here: "The thing they | are is | cats.". :P
-
I pretty much verified my standing in some sketches in the past few days that your program (or related ideas) cannot do what I'm seeking for.
Your program/idea cannot learn to make a splice here "The thing they are | is cats.". Tell me why your program would avoid the splices here: "The thing they | are is | cats.". :P
NonSensical
-
I'll try to be clearer.
I want a program that sees/says hey/realizes that the following sentence should have a splice/gap/separation where shown:
"The thing they are is cats."
But the program must never ever think or say or learn that these are the atomic parts:
"The thing they are is cats."
Does/can your program do that.
Keep in mind it's not about the structure of the moment/given sentence, it's about the features of any text. That example sentence could be 5 times long and make no difference to where splices/separation occurs.