Basic NLP concepts
Concept of NLP coaching
Sample
Samples are customer utterance samples used for NLP training. Each sample expresses one or several certain intentions.
The sample sentence consists of 3 components: Intent, Entity, and Emotion.
Intent
Is the desire that the customer conveys in the sentence
Entity
These are information fields that appear in a sentence, modifying the intention.
Sentiment
Emotions are conveyed in customer statements. The levels of emotions in this system are categorized into positive, negative, and neutral.
For example:
Sample sentence | Intent | Entity | Emotion |
“I am quite satisfied with VN-airline's customer service” | Feedback from customer service Intent: feedback_service | Entity: “VN-airline” Entity type: brand_name | “Satisfied” (Positive) Emotion: positive |
“I want to book a flight to Hanoi” | Book your flight Intent: book_flight | "Hanoi" Entity type: city | Normal feelings Emotion: Neutral |
Keywords
Used for training the bot to recognize words of the same type, abbreviations, borrowed words, synonyms... that have the same meaning as a certain word.
No | Keyword | Entity type | Word/phrase |
1 | rice | food | Hue rice, chicken rice, lam rice (same rice) |
2 | Ho Chi Minh | city | HCM, Saigon |
3 | staff | people | Employee, staff |
Stopwords
These are words that appear a lot in natural language but only carry a little meaning. For example: ah, eh, oh, um, hey, etc...
Teencodes
These are abbreviations in English. For example idk (I don't know), sup (what's up), etc.
Depending on the business, it is necessary to build separate dictionaries for teencodes and stopwords for NLP training.
Last updated