@T: ALWAYS THE TWAIN

Barry Robson very recently wrote on a group thread: “@Twain: You are not only the great integrator, you are a true “Renaissance Man”.

This is an amazing compliment and I’m really humbled because Barry’s background is that he’s CEO of the Dirac Foundation, St Mary’s Hospital Imperial College London; a Council Member of IBM’s Deep Computing Institute and Strategic Advisor at IBM’s TJ Watson Research Center; previously Professorial Lecturer at Mount Sinai Medical Center and also lectured at Stanford University Medical School. His full bio is here: http://www.research.ibm.com/people/r/robson/ and he’s been involved with genuinely ground-breaking projects.

For non-scientists, the Dirac Foundation continues the work of Paul Dirac who was awarded the Nobel Prize for Physics in 1933 (when he was only 31!!!) alongside Erwin Schrödinger, the genius behind the infamous “Schrödinger’s Cat” paradox. Dirac is credited with originating equations that describe the behavior of fermions and the potential existence of…..anti-matter.

It’s even more amazing because this discussion group comprises, arguably, some of the smartest AI and Natural Language Programming (NLP) computer scientists in the world; people who graduated with PhDs and Professorships decades before I was even a twinkle in my parents’ eyes and who’ve been dedicated to developing intelligent systems for over 3 decades more than me. I’m an unknowing novice by comparison to these super, super-clever people who share and spark knowhow in very contextual and relevant ways; and whom I admire, respect and aspire to emulate along my knowhow adventures and evolution.

So what caused Barry’s compliment? Well……….I “twained” three seemingly mutually exclusive approaches to code coherency and interoperability:

(1.) Cultural and perceptual calibration a priori to processing;

(2.) NLP statement structuring for verification of lexicon anomalies; and

(3.) Quantum Mechanics notation.

Here’s how the thread took shape; part of my contribution’s at the top:

I also threw in a solution to resolve Hamiltonian issues (these arose from Einstein’s original Quantum theories and are specifically about time-time capture, btw). It simply made sense in my mind to reconfigure subjunctive tenses from Latinate languages and to re-imagine it as computational code that would fit into double integrations for context mapping as well as differentials for transitional time-position capture.

Originally, when Barry introduced QM code and Hermitian operators into our discussion on “Can we compute the answer to any question?” following a video by Stephen Wolfram of Wolfram Alpha that indicated we would be able to, I have to admit that it was relatively new to me (or to crack a physicist’s joke…….it was µ relativity, ha ha). Anyway, Barry wrote:

For practical purposes that basically means that in standard i-complex QM you rotate the square root of minus one to the hyperbolic number (square root of plus one), and write your relators (verbs prepositions etc. ) as Hermitian operators.

Now  for anyone without a maths / astrophysics degree all of this would read like gobbledy-gook with the exception of the words “For practical purposes that basically means” and “verbs prepositions etc.”. For someone like me I home in on “i-complex” and become interested because I’ve had a long held belief in trying to fuse matrix maths with DNA paths and linguistics to try and arrive at genuine semantics.

Naturally, when faced with something I know less than I should about, I like to ask questions so that I can construct frames of reference for myself and also sanity-check the theories and process themselves for comprehension and coherency. My grandmother did say I was a “curious child and asked so many questions!” — LOL.

My questions stemmed along lines of construct and substitution for ensuring code consistency and coherency and Barry was kind enough to explain and provide examples of what he means by QM, Hermitian operators and twister notation. Here’s his explanation:

Using QM notation such as <subject| verb | object> as analogous to <A operator |B> in QM, operators can be products of operators, so adverbs that qualify them belong between the “|” symbols. Adjectives are actually quite subtle. They could be regarded as Hermitian operators that are completely symmetric as in <grass| green | grass>, but that is unwieldy and I tend to think of them in orthodata to metadata as in <grass:=green| or arguably <green_things:=grass|. QM often ignores the “:=” bit. What they mean is <momentum:=2.3 mass-velocity units | position:= 3 Angstroms>

Time, subjunctive etc belong as verb qualifying operators, unless like in Japanese you want a time tense adjective. Time suggests considering the CPT operations of physics, which I am trying, rather than simpy add a new mindless dimension to the thesaurus. Subjunctive seems to me to have a probabilistic-conditional aspect about it, though one can appeal to the twistor forms (see below) as in <|| wish that | <I| am | correct> >.

The article could be held to have adjectival force, but ultimately belongs I think in between the “|” with the verbs as a matter of categorical relationship. It is more natural to write

<some cats| are | black> = <cats| may be| black> (existential qualification for the general case)

<cats| are |mammals> (universal qualification for the general case)

<Aristotle| is |a man> (universal qualification for the incidence case)

<The philosopher| is |a man> (universal qualification for the incidence case)

<A philosopher| is |a man> (existential qualification for the incidence case)

After about 10 seconds I grasped this notation convention, so I proposed these transformations:

A-ha, thanks, Barry. So it seems to me that even optimally QM and its Hermitian operators (at the moment) work along an equivalent functional way to how logic questions work in IQ tests of the type:

• All cats have tails. Some cats are black. Chester has a black tail.

Can we tell what color Chester is? Or whether Chester is a cat or in fact a dog?

There are 2 Hermitian operator applications of particular interest.

(1.) When you write “I tend to think of them in orthodata to metadata as in <grass:=green| or arguably <green_things:=grass|”, would it be possible to adjust it to these scenarios?

• <green_things:=naive|

In English there is a phrase “green about the gills” which indicates someone fresh and naive.

• <green_things:=new|

Universally, the color green is associated with growth and new shoots.

• <green_things:=prosperous|

In Chinese, the homophone for the word for color green is the word for prosperity and happiness.

• <green_things:=fresh|

In Italian, the word verde for green also has connotations with verdura (vegetables which are fresh).

(2.) You also wrote, “Subjunctive seems to me to have a probabilistic-conditional aspect about it, though one can appeal to the twistor forms (see below) as in <|| wish that | <I| am | correct> >.”

The subjunctive tense is deployed in French, Italian and Spanish when expressing:

* events which are uncertain/doubtful to happen;

* emotions (hopes, fears, etc.);

* opinions rather than facts (subjectivity involved); and

* beliefs.

Time-wise, these are then sub-categorized into future subjunctive, present subjunctive, imperfect subjunctive, past perfect subjunctive and present perfect subjunctive. 

So…..could the “twistor” notation look something like this:

<|would || wish that | <I| am | correct> >

<|might || wish that | <I| am | correct> >

<|did || wish that | <I| am | correct> >

<|had || wish that | <I| am | correct> >

<have had || wish that | <I| am | correct> >

OR could it look something of the form:

<|time || wish that | <I| am | correct> >

I thought about this a little more, in amongst putting together the Startup Shuttle initiative (more in another post), and had one of my “Twain Synergy Epiphany” moments (TSE which will henceforth become some kind of energy or force unit like Joules or Newtons and be pronounced as “T-see”, LOL). It started to distill, crystallize and map in my mind how QM could work with NL and perception capture a priori, so that’s what I shared and that’s what caused Barry Robson to write:

“@Twain: You are not only the great integrator, you are a true “Renaissance Man”.

Readers should be aware that one of my personal heroes is Leonardo da Vinci and never am I more grateful that it’s the likes of him rather than Barbie that my parents taught me to appreciate. For sure, without their grounding and early orientation on the world’s most extraordinary talents, I would now not have reference frames or the intelligence to interact with these super, super-clever people whom I admire and respect.

So now having been suitably impressed and inspired by Barry Robson I’m going to apply to IBM’s SmartCamp competition:

* http://www-05.ibm.com/ie/smarterplanet/smartcamp/index.html

Hopefully, my friend David Price of debategraph.org is reading this post and will also apply since debategraph definitely would contribute to a Smarter Planet as IBM envisions, :*).

Additionally, readers should watch out for IBM testing their Watson machine on ‘Jeopardy’ soon. It was originally announced back in April 2009 and has been scheduled to happen soon (http://www.nytimes.com/2010/06/20/magazine/20Computer-t.html):

There are three specific reasons I’ve enrolled into Italian classes:

(1.) Project ART — I need to translate some material from Italian-English and vice versa.

(2.) 360-2020® — I’d like to incorporate some Italian idioms into the system.

(3.) Personal reasons — my boyfriends tend to be Italian or somehow connected to Italy, so it helps if we can communicate in Italian as well as in English; it would be asking too much for them to be able to speak Chinese, French (and enough German and Spanish to find restaurants and order food, but so long as I can we won’t starve on vacation)! Moreover, Italy is one of my favorite countries in the world and I love visiting it and it’s a lot more fun to be able to speak with the locals!

Anyway, I’ve been allocating some time to constructing my Italian grammar tables and an example for AMARE (to love) can be viewed here if readers click on the image; Firefox is better since Safari seems to exclude the table borders:

I discovered two beautiful phrase examples to use:

(1.) Ogni persona che abbiamo amato è una pianta che fruscii nel vento nel giardino della nostra anima — Every person we loved is a plant that rustles in the wind in the garden of our soul.

(2.) Si amarono fuori dagli schemi ma dentro la loro logica, si cercarono più di loro stessi — They loved outside the box but inside their own logic, they sought more than themselves.

Now, readers can search the entire Internet and all the Italian grammar books out there and they’re unlikely to find a single source, one-page overview of every tense related to a verb and how it’s constructed with examples of usage the way that I create my grammar tables.

This is because I learn languages (and do most things like most people) in my own unique, specific, rational and synergistic way so it makes more sense to draw up my own grammar tables — particularly since most online and book resources contain useful albeit disjointed information, and not what I need:

* logical, stranded timeline of tense applicability

* English equivalent of and equivalence with the tense

* Examples that allow clear differentiation between tenses, notably where the subjunctives are concerned.

My languages teacher in high school (who spoke French, Italian, Spanish and English) wrote in my report: “Twain has a natural flair for languages!” At the time I scored either 100% or high 90%s in English, French, German and Chinese (Cantonese and Mandarin) exams at school. There’s no magic or genius to this; an effective memory, some simple learning strategies and consistent application are helpful to the curious child. Later, she was fairly upset when, against her hopes, I decided to choose Physics, Chemistry and Computer Science as my electives and only French as a language. She believed I should study Modern Languages at university and then go and work for the UN or a diplomatic service.

Clearly, I’m not perfect (linguistically) or an AI robot since I didn’t get 100% all the time — ha ha. Additionally, it’s obvious from reading this blog and some of my online musings that I’m experimental with the grammatical structure, lexicon, vernacular and idioms of languages (whether foreign or code). Still, just because I do this doesn’t mean that I wasn’t properly educated and didn’t earn the appropriate educational qualifications.

I was published in a leading finance trade journal aged 22, was Editor of e-Intelligence and responsible for writing Strategic Investment reports, equity research reports, policy papers and business plans so when I need to write “professionally” I apply a different set of language rules to the ones I use on this blog.

There’s an adage for rebels / anarchists / groundbreakers that flows something like this:

YOU HAVE TO KNOW THE RULES TO BREAK THEM!

My philosophy and approach is more: We have to UNDERSTAND the rules to evolve something smarter.

Now, in recent years, there’s been an educational backlash against the established “rote learning” methodology of education that can still be found in most Oriental classrooms towards “creative learning”, as commented upon here:

· http://www.timesonline.co.uk/tol/life_and_style/education/article5270092.ece

All I can say is that if readers examine my Italian grammar table, it’s an example of CREATIVE ROTE LEARNING. The main reason I can be creative in my approach now as an adult is because as a child I learnt the rote foundations whether it was a language, a times table or the order of a recipe / equation / chemical reaction etc.

For some current educationalists to say that the Google generation doesn’t need to learn by rote at all and can simply go online and google answers and somehow make the structural connections between discrete points of information and “facts” is LAZY, MYOPIC and risks endangering the development and achievements of future intelligence.

There is NO WAY I will let my own child loose onto the Net without any rote learning structure, ability to contextualize and discern genuine facts from threaded untruths to back them up beforehand.

Now let’s make this observation in situ. If I hadn’t benefitted from the rote learning of grammatical structures in English, French, German and Chinese that now help with my accelerated acquisition of Italian, I would just go onto Google Translate, type in a phrase and accept the translation wholesale without any ability to discern its translation accuracy or knowhow to correct any mistakes myself. Why? Well, because those key reference points and structural connectivity that I’d normally develop during rote learning would be missing.

I like Google Translate and it’s good for some but not all uses. Hopefully, Google’s engineers will be able to refine the language filters and grammar construction codes towards more nuanced (semantic) meaning and understanding.

Here are some examples of Google Translation’s “lost in translations”.

(1.) Quando in profumeria vi venderebbero anche la luna.

* Google Translate: When you sell perfume in the moon.

* My human interpretation: (When you’re) in the perfume store they’d also sell you the moon.

(2.) Anche nel caso in cui abbiate venduto tutto e non avete piu’ nulla da riportare in Italia…

* Google Translate: Even if you sold everything and you’re out ‘nothing to report in Italy…

* My human interpretation: Even if you sold everything and you have no more to report in Italy…

(3.) Abbiamo venduto le nostre parti l’anno scorso.

* Google Translate: We sold our shares last year.

* My human interpretation: Abbiamo venduto le nostre azioni l’anno scorso.

(Parti is the literal translation for the plural share of a pie / house. The LATERAL translation of stock shares are azioni and Google Translate’s software interpreted literally, not laterally.)

(4.) Tu mi vendesti per pollastra!

* Google Translate: Thou vendesti for chicken!

* My human interpretation: Thou soldeth me for a chicken!

(Google Translate hasn’t quite mastered antiquated historical Italian idioms yet, :*))

(5.) Non me la sentirei di non farla più la politica.

* Google Translate: I do not feel like it no longer Policy.

* My human interpretation: I don’t feel like doing politics any more.

(Google Translate struggles with direct articles “il, lo, la, i, gli, le” associated with the verb that’s acting on the subject.)

(7.) I Greci sentirono ben presto la necessità di trovare allo Stato un fondamento intrinseco.

* Google Translate: The Greeks soon felt the need to find the state a basis intrinsic.

* My human interpretation: The Greeks soon felt the need to find an intrinsic base for the State.

(Again, this is the Italian literary past tense at play and interpretation of precedence relating to the adjective associated with the subject. The direct object are the Greeks, not the base.)

Readers will note that I refer to Google TRANSLATE whilst my own abilities as human INTERPRETATION. Interpretation embodies with it contextualization and the perceptual reading of sentiments / emotions / intent.

There are definitely challenging tenses for Google Translate and readers won’t be surprised that these tenses involve the expression of hope, desires, emotions, probability, doubt, uncertainty and undefined (non-specific) timelines. Principally:

* imperfetto (imperfect)

* congiuntivo (subjunctive across the board: present, imperfect, past perfect, present perfect)

* trapassato remoto (preterite perfect tense)

It’s well-known that translation software deploys some of the most sophisticated AI and NLP (natural language programming) out there. The fact that the software can’t semantically distinguish or derive tenses involving human emotions and ambiguity (whether in terms of sentiment probability or timeline) is a reflection that there is some way to go before AI agents will make human operators obsolete.

I also want to remark on the fact that Google is a US company and the majority of its employees’ mother language and mental orientation is English. Ergo, it’s not surprising that Google Translate’s reference structures for the imperfect and subjunctive tenses aren’t fully developed. This is because English — whether American English or English English — doesn’t make much use of it whereas in French, Italian, Spanish and Portuguese it’s “de rigeur” to know it and use it properly.

Also, it may be worth noting how Google Translate is performing with Chinese. My mother’s helping me interpret my ‘ Global Brain’ knol into Chinese (simplified) and Mandarin — her technical Chinese is stronger than mine — and she’s said more than once, “This Google translation makes no sense at all!”

LOL.

Yes, it’s possible that I may contribute to cracking the conundrum: “What can we program into machine code to enable them to understand human emotion, intent and multi-stranded (and not necessarily consequential) verb events?”

I believe that the answers will arise from people who can code at a high-level and are also multi-lingual, so their natural radars can spot where the bridges between machine rules and human context still need to be built and developed.

However, despite Google Translate’s offer for us to contribute to making their translations better, I’m not personally going to sit and spend time manually inputting all my human interpretations and corrections of Google Translate’s various faux pas and faux amis! I use the term “faux amis” deliberately — how can we sense someone is our true friend / language navigator / experience explainer or another human being? There’s some hype about AI agents attached to the Cloud making human customer services redundant. yet there are also voice-to-text translation services out there which have been uncovered and alleged to be little more than thousands of humans in a call center in India / Brazil / China doing the translations rather than the machines.

Let’s also reference back to our experiences with Elbot and the Turing Test:

LOL.

So here’s the reality check: a woman who believes in the Internet and its wonderful tools and is digi-savvy, still goes into a physical rather than virtual classroom to learn and prefers to interact with other human beans rather than online language bots.

[Note to my kid(s): 如果你正在閱讀這篇文章在2020年，媽媽說：“請回到您的公式 / 元素周期 / 表文法的工作了。謝謝。我愛你。理由的所有在這裡!"

Google Translate gets this wrong too, both sides of the translation, i miei bambini. That's why you have to go to Chinese school and be taught it properly. Also, please listen to your 姥姥 when she's explaining the nuances between logic and rationale! ]

:*).