vocabulary:
  name: Salesforce Einstein Vocabulary
  description: >-
    Operational vocabulary for Salesforce Einstein AI and ML APIs covering
    computer vision, natural language processing, predictive analytics,
    generative AI, and conversational AI concepts.
  version: "1.0"
  created: "2026-05-02"
  modified: "2026-05-02"
  tags:
    - Artificial Intelligence
    - Machine Learning
    - Salesforce Einstein

  domains:
    - name: Machine Learning
      description: "Core ML concepts across all Einstein APIs."
      terms:
        - term: Dataset
          definition: >-
            A labeled collection of examples used to train an Einstein model.
            Vision datasets contain images; Language datasets contain text documents.
          aliases: [Training Data, Training Set]
          related: [Model, Label, Training]

        - term: Model
          definition: >-
            A trained machine learning artifact that can make predictions on
            new data. Created by training on a dataset.
          aliases: [Classifier, AI Model, Trained Model]
          related: [Dataset, Prediction, Training]

        - term: Training
          definition: >-
            The process of building a model by exposing it to labeled examples
            in a dataset so it can learn patterns.
          related: [Dataset, Model, Accuracy]

        - term: Prediction
          definition: >-
            The output of a model when analyzing new input data. Includes
            probabilities/confidence scores for each possible label.
          aliases: [Inference, Classification Result]
          related: [Model, Probability, Label]

        - term: Label
          definition: >-
            A category or class assigned to training examples. For vision,
            a label might be "dog" or "cat". For language, "positive" or "negative".
          aliases: [Class, Category, Tag]
          related: [Dataset, Training, Prediction]

        - term: Confidence Score
          definition: >-
            A probability value (0–1) indicating how confident a model is in
            its prediction. Higher scores indicate greater certainty.
          aliases: [Probability, Score, Confidence]
          related: [Prediction, Model]

    - name: Computer Vision
      description: "Einstein Vision concepts."
      terms:
        - term: Image Classification
          definition: >-
            The task of assigning one or more labels to an image based on
            its visual content.
          aliases: [Image Recognition, Visual Classification]
          related: [Object Detection, Dataset, Model]

        - term: Object Detection
          definition: >-
            Identifying and localizing specific objects within an image,
            returning bounding boxes and labels for each detected object.
          aliases: [Object Recognition]
          related: [Image Classification, Bounding Box]

    - name: Natural Language Processing
      description: "Einstein Language concepts."
      terms:
        - term: Sentiment Analysis
          definition: >-
            Classifying the emotional tone of text as positive, negative, or neutral.
            Used to understand customer feedback, social media, and service cases.
          aliases: [Sentiment Classification, Opinion Mining]
          related: [Intent Detection, Text Classification]

        - term: Intent Detection
          definition: >-
            Identifying the purpose or goal behind a piece of text.
            Used to route customer queries to appropriate service flows.
          aliases: [Intent Classification, NLU]
          related: [Sentiment Analysis, Chatbot, Einstein Bots]

    - name: Generative AI
      description: "Einstein GPT and generative AI concepts."
      terms:
        - term: Prompt Template
          definition: >-
            A reusable template defining instructions and placeholders for
            generating AI content. Stored in Salesforce and parameterized
            with record field values at runtime.
          aliases: [Prompt, Template]
          related: [Content Generation, LLM]

        - term: Large Language Model (LLM)
          definition: >-
            A neural network trained on large text corpora capable of generating,
            completing, summarizing, and transforming text. Einstein GPT uses
            Salesforce's own models and integrated OpenAI models.
          aliases: [LLM, Foundation Model, Generative Model]
          related: [Prompt Template, Generation, Completion]

        - term: Grounding
          definition: >-
            Connecting an LLM to real Salesforce data (records, knowledge articles)
            to generate contextually accurate, personalized content.
          related: [Prompt Template, CRM Data, Retrieval Augmented Generation]

    - name: Conversational AI
      description: "Einstein Bots concepts."
      terms:
        - term: Einstein Bot
          definition: >-
            An AI-powered chatbot built on the Salesforce platform for handling
            customer service queries, with natural language understanding
            and integration with Salesforce data.
          aliases: [Bot, Virtual Agent, Chatbot]
          related: [Session, Intent, Dialog]

        - term: Session
          definition: >-
            A single conversation instance between a user and an Einstein Bot.
            Sessions maintain conversation context and history.
          aliases: [Conversation, Dialog Session]
          related: [Einstein Bot, Message]

        - term: Dialog
          definition: >-
            A configured conversation path in an Einstein Bot defining how the
            bot responds to specific intents and user inputs.
          related: [Intent, Einstein Bot, Session]