#hierarchical-semantic-resolution #embeddings #blockchain #ai #llm Created at 130223 # [Anonymous feedback](https://www.admonymous.co/louis030195) # [[Epistemic status]] #shower-thought Last modified date: 2023-02-13 Commit: 0 # Related - [[Computing/Intelligence/Machine Learning/Embedding is the dark matter of intelligence]] - [[Software 3.0]] # TODO > [!TODO] TODO # Hierarchichal semantic resolution The idea of hierarchichal semantic resolution is to structure embeddings data in a hierarchical way which allows for the use of more complex and nuanced semantic relationships. This is done by breaking down a single word or phrase into multiple layers, each representing a different level of abstraction. For example, a word such as "cat" can be broken down into "feline", "pet", "animal", and "living thing". This allows for more precise semantic resolution as each layer can represent a different meaning. By breaking down the words into multiple layers, it allows for more complex relationships to be established and analyzed. For example, if a sentence contains the words "cat" and "dog", the hierarchical semantic resolution can determine the relationship between the two words more accurately than a single-layer embedding. This can be beneficial in tasks such as sentiment analysis, where the context of the words is important for an accurate analysis. A hierarchical embedddings, like a [[Blockchain]], will have its value changed as soon as a parent node value change. For example a tree may have an embeddings "1", while one of its leaf is "13". Let's say the tree is being attacked by ants, it's embeddings will change to "2". All of the leaf nodes that are dependent on the parent node's value will also change to "14". This allows for the system to be more accurate in its semantic resolution, as it is able to track changes in the relationships of the words it is analyzing.