Natural language understanding: Difference between revisions
(DIET classifier link added) |
(created a link for gpt-3) |
||
| Line 3: | Line 3: | ||
In recent years, especially in the chatbot engineering industry, we tend to use NLU to mean an intent/entities classifier, based on machine learning techniques (transformers, etc.). The main open source project / state of the art of this approach is probably the [https://rasa.com/blog/introducing-dual-intent-and-entity-transformer-diet-state-of-the-art-performance-on-a-lightweight-architecture/ RASA DIET classifier]. | In recent years, especially in the chatbot engineering industry, we tend to use NLU to mean an intent/entities classifier, based on machine learning techniques (transformers, etc.). The main open source project / state of the art of this approach is probably the [https://rasa.com/blog/introducing-dual-intent-and-entity-transformer-diet-state-of-the-art-performance-on-a-lightweight-architecture/ RASA DIET classifier]. | ||
Besides, in terms of linguistic, and psycho-linguistic/cognitive scientific disciplines, there is a great skepticism about naming "language understanding" a ML-based classifier of intents (and entities). A growing number of researcher linguists state that it's even impossible to understand language with machine language techniques (the more famous and currently debated is probably GPT-3). One of the scientist more active in this battle is [https://ontologik.medium.com/ Walid Saba]. | Besides, in terms of linguistic, and psycho-linguistic/cognitive scientific disciplines, there is a great skepticism about naming "language understanding" a ML-based classifier of intents (and entities). A growing number of researcher linguists state that it's even impossible to understand language with machine language techniques (the more famous and currently debated is probably [[GPT-3]]). One of the scientist more active in this battle is [https://ontologik.medium.com/ Walid Saba]. | ||
Revision as of 20:11, 3 January 2022
Natural Language Understanding (NLU) is a a misleading term, highly discussed in the Conversational AI / scientific community.
In recent years, especially in the chatbot engineering industry, we tend to use NLU to mean an intent/entities classifier, based on machine learning techniques (transformers, etc.). The main open source project / state of the art of this approach is probably the RASA DIET classifier.
Besides, in terms of linguistic, and psycho-linguistic/cognitive scientific disciplines, there is a great skepticism about naming "language understanding" a ML-based classifier of intents (and entities). A growing number of researcher linguists state that it's even impossible to understand language with machine language techniques (the more famous and currently debated is probably GPT-3). One of the scientist more active in this battle is Walid Saba.