Entity Extraction
Definition
Entity extraction, also known as named entity recognition (NER) or slot filling, identifies structured data points within free-form text. While intent recognition answers what the user wants, entity extraction answers the specifics. For example, in 'I'd like to book an appointment for Friday at 3pm', the intent is 'book_appointment' and the entities are 'date: Friday' and 'time: 3pm'. Entities can be custom (order IDs, product names) or system-level (dates, numbers, email addresses). Modern LLM-based systems extract entities without requiring explicit entity definitions through instruction following.
Why It Matters
Entities are the variables that make automated workflows possible. Without entity extraction, a chatbot can only respond with generic answers. With it, the bot can personalize responses, pre-fill forms, query databases with specific parameters, and complete transactions — all based on information the user provided naturally in conversation.
How It Works
Rule-based entity extraction uses regex patterns and lookup tables. ML-based approaches train sequence-labeling models to tag entity spans in text. LLMs perform entity extraction by following natural language instructions: extract the date and time from this message. Extracted entities are stored in the conversation state and used to parameterize API calls or database queries.
Real-World Example
A user says 'Can I get the tracking info for order 98765 placed last Monday?' Entity extraction identifies the order ID and the date reference. The chatbot uses these to query the order management system and return the specific shipment status.
Common Mistakes
- ✕Not defining custom entities for domain-specific terms like product names or internal IDs, relying only on generic entity types.
- ✕Failing to validate extracted entities — accepting nonsense order numbers that do not exist in the database.
- ✕Assuming a single entity instance per message — users often mention multiple products or dates in one message.
Related Terms
Intent Recognition
Intent recognition is the process by which a chatbot identifies the goal or purpose behind a user's message. It classifies free-form user input into predefined categories (intents) — such as 'check order status', 'request refund', or 'get pricing' — enabling the bot to route the conversation appropriately.
Slot Filling
Slot filling is the dialogue management process of collecting all the required pieces of information (slots) needed to complete a task. The chatbot systematically asks for any missing slots — like date, time, or account number — until it has everything needed to fulfill the user's request.
Natural Language Understanding (NLU)
Natural Language Understanding (NLU) is the AI capability that interprets the meaning behind human text or speech — identifying what the user wants (intent) and extracting key details (entities). NLU is the 'comprehension' layer of a chatbot, translating raw input into structured information the system can act on.
Dialogue Management
Dialogue management is the component of a conversational AI system that tracks conversation state and decides what the bot should do next — ask a follow-up question, retrieve information, take an action, or hand off to a human. It is the 'brain' that orchestrates a coherent, goal-directed conversation across multiple turns.
User Utterance
A user utterance is any message, phrase, or spoken input a user sends to a chatbot. It is the raw input that the NLU layer processes to determine intent and extract entities. Understanding the variety of utterances users produce for the same intent is essential for training accurate, robust chatbot models.
Ready to build your AI chatbot?
Put these concepts into practice with 99helpers — no code required.
Start free trial →