A beginner-friendly glossary of chatbot terms
Lionbridge AI is now TELUS International. But don’t worry! We are currently updating our great content to our new home. This article will be looking good again in no time!
A chatbot is an artificial intelligence software that simulates human conversation through mobile apps or social media messaging platforms like Facebook messenger or Slack. You’ve probably used chatbots before, perhaps for online shopping or to book a flight or hotel. Chatbots allow businesses to automate customer service, without having to employ a customer support representative to help each customer who reaches out with a question. For some audiences, chatbots might even increase customer satisfaction.
Chatbot Terms for Beginners
- Automatic speech recognition: Automatic speech recognition is the process of taking the user’s speech as input, and using computer hardware and software techniques to identify what words were actually said.
- Autoresponder: An autoresponder is an automatic reply that is triggered when a user sends their first message, or a specific keyword, to the chatbot. For example, The autoresponder feature is available on Facebook messenger. If you message a brand on Facebook messenger and get an instant reply, they are probably using autoresponder.
- Broadcast: A broadcast is a message that is sent to all users in your list of chatbot users. For example, you might broadcast a message to all users who liked your company’s Facebook page.
- Chat logs: Chat logs are past records of all spoken and typed interactions between a user and chatbot.
- Classifier: Classifiers are a way to categorize user inputs into different categories. Humans naturally classify objects into sets. Pianos are instruments, t-shirts are clothing and happy is an emotion. Similarly, chatbots break down sentences and categorize the segments, to understand the intent behind each user input.
- Compulsory input: A compulsory input is information that the user must enter before moving on in the chatbot conversation. The chatbot won’t ask a different question or otherwise move forward in the conversation, until the user provides the missing information.
For example, flight number might be a compulsory for airline support chatbots. The chatbot would need to know which flight you're booked for to be able to assist you. Another example would be order tracking number, since an online e-commerce chatbot needs to know which shipment you'd like to cancel or order again.
- Entity extraction: Entity extraction is an umbrella term that refers to the process of adding structure to text data so that your chatbot can read it. The chatbot uses entity extraction to identify words from user utterances, and respond accordingly. If the chatbot needs more information to complete a task, it will prompt the user for an additional entity.
- Intent: Intents are the purpose of a user’s input into a chatbot. For example, if a restaurant has a chatbot on their website, then a customer might use it to inquire about business hours. This intent can be expressed in different ways, such as What are your hours of operation? or What time do you open and close?
- Intent classification: Intent classification is the process of categorizing utterances into predefined intent groups. This is important because chatbots need to accurately match utterances to specific intents, to be able to respond, continue the conversation, and provide the right answers.
- Intent recognition: Intent recognition (also called intent detection or intent extraction) is the process of extracting relevant information from a user utterance, so that a chatbot can understand the intent behind it. Intent recognition is a critical natural language understanding task for intelligence user interfaces to determine what kind of support a user is looking for, and how the UI can offer help.
- Natural language processing: Natural language processing is a field of artificial intelligence that specializes in a machine’s ability to recognize what is said to it, understand its meaning, determine the appropriate action and respond with language that the user can understand. Chatbot development relies heavily on natural language processing, since chatbots mimic human conversations.
- Natural language understanding: Natural language understanding is a subfield of natural language processing that aims to understand the intended meaning of chatbot utterances. Human speech is peppered with nuances, subtleties, idioms and mispronunciations, but natural language understanding aims to sift through these complexities of human speech, to extract the user’s intent.
- Platform: The platform, also called the channel, refers to the application that hosts your chatbots. These days, you can host chatbots in most social media sites, such as Facebook messenger and Twitter direct messages.
- Semantic annotation: Semantic annotation is the task of annotating various concepts within text, such as people, objects, or company names, to train a chatbot. The chatbot will refer to the semantic annotation to categorize new user input, and respond accordingly.
- Utterance: Lastly, this is one of the most important chatbot terms that you should know. Chatbot utterances are anything that the users says or types as input into the chatbot. For example, if a user types What is the current time in Tokyo, Japan?, the entire sentence is the utterance.
Beyond the glossary, there is much to chatbots and their potential. Check out our one-minute white paper to learn how chatbots are leveling up digital customer experience. From there, explore our AI & Bot digital experience solutions and get in touch to drive your business forward with chatbots.