Friday, January 22, 2016

IBM Watson - AlchemyLanguage API and Clive Crook

The IBM AlchemyLanguage API is a sophisticated analysis tool.  It analyzes natural language content.  Clive Crook is a professional investment specialist.  He speaks English.  The way he speaks it mystifies me.  I look at it an conclude it is pompous, arrogant and bombastic.  That is a harsh personal judgment based on feeling.  I am a reason and logic based thinking guy.  I do have some feelings like most guys.  Sometimes guys even let their feminine mind express itself.  Sometimes I give into my feelings and let them lead my thinking and guide my actions instead of logic and reason.  Because....love?  Totally irrational by design?

Anyhow, I have these feelings about Clive Crook.  Sorry Clive.  I just feel that way.  Perhaps I should not.  Why do I feel that way?  I need to consult with someone that does not have a feeling side.  Simply and purely logical and rational.  Sounds like a machine to me.  Maybe some perfectly macho guy perhaps.

Perhaps IBM's Watson is the guru I seek? Watson uses the AlchemyLanguage API to analyze Natural Language.  Watson is smart.  It can learn about us from its ability to process natural language then use what it learns to give us an analytical look at what we say.  A different way of looking at what we say and telling us things about what we say, the way we say it to extract meaning of what we say and present it in an analytical fashion the way Watson's machine brain looks at it. 

I doubt that Watson would cry if given the most tear wrenching story.  Wastson would however rate it in terms of negative/positive from what it could infer from processing what we say.

Watson.  Siri's smarter (as far as machines go) Big Brother.  Ha, ha!  Sorry Watson that is a joke if you did not get it.  Say "Ha, ha" if you did.  Twice if you thought it was really funny. Four imoji smiley faces if you are ROTFLMAO.

Enough fun at Watson's expense.  Time to have fun at Clive Crook's expense.

The AlchemyLanguage (AL) does a good job at parsing our Natural Language and telling us something about it.  Try the API.  Give it a test drive with your own input.  Check it out.

AL does not understand what Clive Crook says.  Maybe it is because we are two countries divided by a common language? 

I read what Clive writes.  It looks like English to me.  All I get out of it is a feeling and some cloudy idea that he is either a genius talking way over my head or he does not know what he is talking about because his mind mixes up all the logical, rational conceptual relationships of the words he uses.  What he writes therefore comes out looking like he really has a grasp of the higher level conceptual structure he is talking about weaving together low level conceptual nouns and verbs.

Nouns and Verbs are low level conceptual expressions that Watson eats, digests and turns into machine intelligence at an amazing rate.  Watson learns from what it eats and grows as it links the low level conceptual words into higher levels of understanding, intelligence and even machine knowledge.

Ok, Watson, if you are so good at what you do that you can beat the best human chess player then put his in your   AlchemyLanguage  API and smoke it.  Tell me what you think (unlike what I may feel) about what Clive Crook says.

"A reasonable possibility is this was for health related reasons. Regardless, it is arrogant. Even if there was a good explanation for why people would be left, understandably, disappointed, why simply get up and go without saying it.

If Clinton had come out with something like “sorry, I’ve been running behind schedule all day for X- or Y- reason, I don’t want to have to cancel my next appointment at Z- event, I do hope you’ll understand etc. etc.” people would have, almost universally, appreciated that."

You, the reader, if any, may take what Clive says and put it in the AL API to see the results.  Nobody ever really looks at this blog in the first place.  I do not write for readers, only myself.  Readers are a figment of my imagination because the don't exist here.

AL API analysis:

Entities: Clinton.  My analysis: Right on!  I agree!  Writing something about "Clinton".

Cut to the chase and go to "Relations".

AL API says:  None Detected! WTF?  Clive said something.  I can read a bunch of words he said.  I quoted them in this blog.  I can read what AL API looks for when I go to the "Relations" explanation link.  I learned how to do sentence diagramming in 3rd grade.  I loved doing it.  AL API says no Relations detected?  IBM paid big bucks for this?  Somebody separated the relationship between IBM and its Big Bucks for this to enhance Watson's Deep Learning Capabilities"?

Words in a sentence have syntax and semantics

At the link of what Clive wrote and the AL API analyzed for Realtions and looking at the Naked Capitalism website responses to it I find that "abynormal" got what Clive said and had a belly laff.

In reply to "abynormal" Clive wrote:

"Doh! Beginners error there, wasn’t it!? You see, I forgot for a moment it was that Hillary Clinton we were talking about and hope tormented me. Thanks Aby, I’m back to reality now and have shaken off that regrettable lapse into naivety."

Applying what Clive wrote above to the AL API it parsed to this sentence in Relations: "You see, I forgot for a moment it was that Hillary Clinton we were talking about and hope tormented me." 

Relations further reduced this parsed sentence to its subject noun - verb- object noun simple structure:

"we hope tormented me"  We is the subject noun.  Hole is the verb. "Tormented me" is the object noun phrase.

At least the AL API found something that Clive said that had enough value to parse because it could be reduced to a fundamental thing we call a sentence at the most basic relationship level.  Subject noun - Verb - Object Noun.  What grade did I learn that in?

AlchemyLanguage simply does not get what Clive has to say.  By inference or deduction or whatever the AL API uses to do what it does the "Relations" conclusion is he had nothing to say that could be parsed in accordance with some algorithm of NLP?

Does AL API get what I am saying?   This requires a test.  I will use the prior paragraph beginning with:  "AL API says:"  and see what AL API says about it.

The result is long.  I will copy and condense it.  The AL API Relations displays the following:

"Clive said something
He said a bunch of words
IBM paid big bucks for this"

That in a nut shell is what I said but the point I intended to make was that Clive actually wrote something but the AL API "Relations" could not find any relationships among the words he used to say it.  In conclusion.....he said nothing with any value to parse?

Maybe what is needed is something more lengthy that Clive Crook wrote to discover if he has anything to say that the AL API can parse in accordance with what Relations seeks to express by means of its parsing functions.

For example what Clive says here: http://www.nakedcapitalism.com/2016/01/why-bitcoin-is-not-disruptive.html.

It is something that he seems to have written exclusively for Naked Capitalism.  He is a writer for major publications.  I can't find anywhere else on the WWW where this appears.  I believe that it is probably entirely in his own words without any external editing.  Let's just say it is without any filter.

Selecting the first 6 paragraphs of what Clive wrote and entering it in the text analysis of the AL API the result is that Relations did not find any relations in the words of the first 5 paragraphs.  Do the entry yourself and look a the results.  I took the same selection to a reading level analysis program and the reading level is very high.  Maybe that is the problem?

I read the same first four paragraphs and my analysis is that they are bombastic pomposity.  Bombastic pomposity is a high level conceptual abstract that Watson has yet to achieve in its artificial intelligence.

For example purposes these are the first four paragraphs of what Clive Crook wrote:

By Clive, an investment technology professional and Japanophile
Have you ever, as I have, watched something on television, read an article in a newspaper or online – or similar – where the topic has been that with which you’re especially knowledgeable about? Perhaps you’re, say, a teacher and the piece has been about education or child development. Or maybe you’re a mechanic and it is a feature on auto maintenance. You might have a hobby, let’s pick painting, and a non-expert is trying to explain what it is to paint.
Whenever I, being in finance, encounter the output of a journalist or – worse – an economist or a politician trying to tell the audience something about a financial matter (particularly if it is my specialism, consumer finance and money transmission) I rarely hear or read the first sentence without encountering an error of such fundamental magnitude that it renders pretty much everything that follows completely wrong. At first, I thought that it was because my knowledge is about an esoteric or complex subject. Finance really isn’t complex, no matter how many attempts are made to make it so, but I speculated that as it wasn’t exactly commonplace then maybe that excused to all-too-frequent inaccuracies I found when it was covered in the general media.

So I began to tell people I knew what I was encountering. They responded in exactly the same way. Educationalists bemoaned how woeful and wrongly reporting of schooling, teaching and the delivery of learning was compared to how it actually happened – as they knew, as a fact, from their own first-hand knowledge. Social workers told me that it appeared when the subject was reported, those doing the reporting seemed to have no concept whatsoever of what a social worker was. The guy who fixed my leaky pipe said that most of what was on television to do with plumbing was rubbish. And so on.

I suppose I should be expecting it. Journalism has descended into – along with a lot of “professions” a generalist skill set. Add in time pressures, often rigorous imperatives on word counts, readability (or using too long words, if it is television) along with the lack of subject matter knowledge and it is hardly surprising we’re drowning in misinformation.

There are a few – a very few – sources of good information. Specialised financial publications can be one such source. But even there, there’s no guarantees. The Financial Times, for instance, is invariably good to excellent. The Economist, however, is so in hoc to vested interests and conventional wisdom, it is not worth bothering with. Outside the niche media, though, if you read or see a feature on some aspect of finance or economics, you’re more likely to come away dumber than when you started. That is some achievement. I’d like to think that we make Naked Capitalism one such source of sufficiently good quality coverage to help begin to redress the balance of so much wrong thinking caused by wrong understandings."

The AL API finally finds something it can analyze in the first sentence of the 6th paragraph.

The subject matter that Clive is writing about is: "Why Bitcoin Is Not Disruptive".  He laments the journalist and other uninformed professionals misrepresentation of financial matters on which he, "being in finance" recognizes as a misrepresentation because he is a subject matter expert.

"Whenever I, being in finance, encounter the output of a journalist or – worse – an economist or a politician trying to tell the audience something about a financial matter (particularly if it is my specialism, consumer finance and money transmission) I rarely hear or read the first sentence without encountering an error of such fundamental magnitude that it renders pretty much everything that follows completely wrong."

Clive might be a finance expert.  He analyzes Bitcoin by means of  transactional relationship charts to show that it is not a disruptive technology.  He proves himself to be correct.

Clive himself falls into the misconception trap.  He fails to see, because his all seeing and knowing eye sees only finance transactions, that Bitcoin is a Blockchain application and Blockchain is going to be greatly disruptive to finance.  He does not see or understand that. 

Fortunately, his countryman does.  https://www.cryptocoinsnews.com/uk-chief-scientist-bitcoin-blockchain/Chief Scientist Sir Mark Walport sees beyond the application of Blockchain to bitcoin and its potential for substantial systemic changes in government.

Blockchain applied to finance is, I think, going to be very disruptive.  Clive Crook can't see the forest for the trees.  He is very good at looking at the trees in the financial forest that he lives in.  He has no view of a bigger picture.  I think that anything outside of his view of the legacy financial world claimed to be disruptive change is a threat to the status quo expressed by those that are not subject matter experts like he is.

Simple sentence:  Clive Crook is arrogant, bombastic and pompous. AL API extracts that in Relations exactly as I wrote it.  Looks like I said something that it could parse.  It will probably go into Watson's brain AI memory.  Someday Watson will be intelligent enough to give its conclusive judgment analysis of Clive Watson and determine if what I feel as well as think about Clive Crook compares to what Watson thinks about Clive Crook.

Someday in the far future if this blog and this entry lasts that long a grand son or daughter might discover what gradpa wrote and ask Watson what is analysis of all that is.  Maybe Watson will confirm that grandpa was a wild and crazy guy or what................Maybe Whatson will say he spent too  much time thinking. 





 








No comments: