Tag Archives: NL

Why Gatfol Technology is Extremely Timeous…Siri for Everybody…

General-purpose conversational assistants by design use voice
recognition technology to isolate “key terms”
to mine sources for useful results…

Here’s the problem:

Imagine any sentence of about 8-10 words…

i.e. “I really appreciate my mother in the morning”

What would happen if we replace each word with – let’s say –
ten equivalent words that fit both grammatically and semantically?

i.e. “I definitely/positively/demonstratively…” “like/admire/love my mother…”

Taking the original phrase and randomly inserting the
replacement words in all possible groupings that still make sense, we get
100 million phrases that are ALL grammatically intact and semantically equivalent
– and
we are still only saying that we feel positive about our mother some time early in the day!

…Even the smallest body of text of even minimum complexity has trillions upon trillions of equivalent
semantic permutations.
In terms of conversational assistants – and without a Gatfol functionality-
we just do not have the
backend concept-combination multiplication power
to even begin to cover the permutation problem…

Five years ago…

 …Apple’s SIRI was not a reality…
…Augmented reality was in theoretical infancy…
…Semantic replacement technology was not on the radar…
…Apple was not promoting wearable computing devices with natural language interfaces…

Today…

…the world is starting to realise that we have to merge the immense richness
and depth of human
everyday language with the limited actionable
instructions of software programs and databases
if we want
to rely on digital machines to guide our lives…

Five years from now…

…SIRI equivalents and augmented reality will be everywhere…

 …many commercial suppliers of semantic language phrase replacement technology will exist…


…Gatfol is already the first…

Could Permutations Kill Ray Kurzweil’s Google AI Dream?

I remember reading somewhere online about a recent interview
(amongst many) with Ray Kurzweil – Director of Engineering at Google and the
world’s staunchest AI visionary – where he mentioned something about his plans in the line of…

to get the Google computers to understand natural language, not just do search and
answer questions
based on links and words, but actually understand the semantic content

I cannot remember where I saw the interview and will Google it…

…but I have a problem…

I know my keywords are fuzzy and I use Gatfol to show me some sample permutations on my input
…and realise that with 13 keywords and only 10 same-meaning single word replacements for each
keyword, as well as allowing for multiword to semantic equivalent multiword replacements, I am
looking at a minimum of 10 000 000 000 000 000 000 000 000 different ways to phrase my input
…all with the same basic meaning…

I also realise that I want a very specific search return that semantically equates to my input and
does not just hit and return pages corresponding to the trillions of different keyword combinations…

I quickly calculate that just by browsing my own thoughts and looking at my immediate
physical surroundings, I can theoretically come up with trillions upon trillions of meaningful
search input queries…each with trillions upon trillions of semantically equivalent phrasings…

The thought hits me that if Google wants to morph into Ray Kurzweil’s vision of a search
mind with which I can communicate and interrogate as if human – perhaps in five to ten years –
it will have to perform massively scalable multiword to multiword semantic equivalence transformations…

…Gatfol currently does this…