Have a look at the videos that we have produced to help you gain a better understanding of Pat

Meaning w/o nouns or verbs

Using sentences where either noun or verb form predicates unlock meaning is demonstrated. Some consider it the Holy Grail of NLU, but regardless it is essential for conversational A.I.

Published on Feb 17, 2019

Pat bAbI Converse

In this video we show Pat going through one of Facebook’s bAbI tasks.

Published on Sep 13, 2018

Pat input processing

In this video, we show how Pat’s NLU breaks down sentences and questions to engage in meaningful conversations

Published on Sep 13, 2018

Pat conversation demo

Take a look at Pat taking note off and responding to statements & questions. This is a demo of our NLUs capabilities in engaging in a meaningful conversation.

Published on Sep 13, 2018

NLU with Disambiguation

Here we show how context resolves the meaning of conversation and illustrate the impossibility to converse with today’s mainstream approaches. Scalability comes from an effective model.

Published on Aug 30, 2018

Pat's Facebook AI Test

Today we look into how our NLU solved the Facebook bAbI tasks, and found errors in the machine learning input data, causing undetected output errors.

Published on Aug 8, 2018

Patom Theory Explains Language Acquisition: Part 2

Today, the theory behind the meaning matcher’s phrase design are explored before the results and internals of the linking approach are demonstrated.

Published on Jul 28, 2018

Patom Theory Explains Language Acquisition: Part 1

The brain-based theory is introduced with its application to language learning. This video demonstrates elements from the companion medium article.

Published on Jul 22, 2018

Meaning Matching Without Parts of Speech

Today we demonstrate how an NLU differentiates between the word forms of ‘run’ based on meaning/definitions. This is currently known as an open scientific problem under the name Word-sense Disambiguation (WSD) and is central to real NLU.

Published on Jul 17, 2018

Why grammatical elements are bad for parsing

The problem of parts of speech is explained as part of the wrap up for why parsing is a bad concept for NLU systems. Not only have experts said it is impossible for full human languages, but it creates ambiguity at many levels resulting in the unsolvable combinatorial explosion.

Published on Jun 29, 2018

Bonus demonstration: why semantics, not syntax

This video provides more demonstrations with the meaning matcher, showing how semantic manipulation of language is superior to syntax-only solutions.

Published on Jun 27, 2018

Why Semantics, Not Syntax, creates NLU

This is the first video in the series on NLU: why syntax alone is inadequate. We look at the underlying hypothesis and approach to syntactic structures and compare it to meaning-based systems.

Published on Jun 27, 2018

Pat's NLU Series - Overview

This is an overview of the series on NLU with the science of NLP and the linguistics of language understanding.

Published on Jun 24, 2018