Within the Synesis One ecosystem a Kanon Collection is comprised of 10,000 keyword Kanons, designed to celebrate digital human knowledge and its evolution at the time of minting. Each selected Kanon is based on the estimated frequency of the individual keyword’s usage. This presents a unique opportunity to consider what’s rare in the realm of keyword NFTs? Which words have more meaning to me? Which words do I want to collect? What words are used frequently?
What is a Kanon?
A Kanon is a semi-fungible token (NFT) representing a keyword or a fractional unit of the Synesis Metaverse. It is a collectable. And like a stamp collection it encapsulates a moment in history, a hero, or an object worth commemorating or to be cherished by many. Each Kanon will entitle its holder a right to claim a network reward in Synesis (SNS).
What is Natural Language Processing?
One of the most innovative sub-fields of AI is natural language processing (NLP), which is the study and application of specific tools that enable machines to process, analyze, and interpret human language. What makes NLP so impressive in its abilities? For starters, human language is incredibly complex given the variations across populations and the ambiguity and imprecise nature of natural languages. This makes it difficult for even humans to analyze and understand the tens of thousands of words flying at us each day.
What Ambiguous Words Exist?
Ambiguity in AI systems refers to words, sentences and phrases that can have two or more possible interpretations:
- Lexical ambiguity: A single word can be used as a noun, verb, or adjective.
- Semantic ambiguity: The context impacts the interpretation of a sentence.
- Syntactic ambiguity: A sentence can be interpreted in more than one way due to its ambiguous nature.
AI often needs to be trained more intensively on these specific words and phrases in order to be accurate.
We’re not spilling the beans on which words will be available in the first collection yet, but here are 10 possible Kanon collectables that are really difficult for humans and AI:
The word break is one of the most ambiguous words in human language, meaning successful machines must be trained to recognize the differences. It can indicate an abrupt occurrence, unexpected good luck, a personal or social separation, and much more.
Another one of the ambiguous words of the English language, run can present some difficulties for AI systems. Not only can the word have multiple meanings, but it also sounds the same, meaning AI systems must learn to differentiate between them in context.
We’re breaking the rules a bit with this one and using three words — their, there and they’re. Homonyms, or two or more words that are pronounced the same but that have different definitions, are especially difficult for question answering and speech-to-text applications since these applications are not written in text form.
The word love is one of the most commonly used English words. AI systems must be trained to recognize whether it is being used to indicate a strong positive emotion of affection, an object of devotion, a person, or one of the other meanings that could dramatically change the nature of a sentence.
Irony and sarcasm can also pose difficulties for AI, which brings us to our next word — whatever. It can be interpreted many ways by us humans, especially since it can have either a positive or negative annotation. AI must be trained with certain cues in order to understand the meaning behind the word.
AI systems struggle with pronouns, and this is what often leads to bias creeping into the systems unnoticed. Most major AI systems being used today have a bias against her.
A neologism is defined as a relatively new or isolated term, word, or phrase that is entering common use. One of the most popular is retweet, which many of us encounter in our everyday lives. AI systems must be trained to understand these new words that have yet to be accepted into mainstream language.
The word won’t always has an ambiguity associated with it. It can often confuse AI systems, which struggle with deciding whether to treat it as one word or two words, as well as determining its meaning.
The word make, which is one of the most common in the English language, is another ambiguous term that can require more training. AI must recognize whether it’s a noun or verb, and what exactly it is referring to.
Closing out this list is the word clear, another highly-used English term. The word fits into many categories including adjective, noun, adverb, or verb, and there are many different meanings behind it.
These are just 10 of the many English words that can trip up AI systems. If we humans sometimes struggle with ambiguous words despite our ability to understand context, it only makes sense that AI systems would do the same.
At Synesis One our ecosystem utilizes both a play-to-earn model allowing anyone to have fun while training AI on how to interpret and understand these ambiguous words — and an NFT exchange, which allows you to trade, or collect and hold your favorite words to earn rewards.