Here’s the glossary to increase your understanding of how chatbots work under the hood.
Communicating with a chatbot starts to feel natural when you approach the customer service department of a company or other organisation. But do you really know how the technology behind the chatbot works? Enkronos has compiled a list of concepts you need to know to understand how bots work under the hood.
The first chatbot was created back in 1966 by MIT professor Joseph Weizenbaum. He created Eliza to mimic the behaviour of a psychologist. The program had a database of ready-made answers and counter-questions that were matched to different questions so that it would feel like communicating with a human. Today’s most advanced chatbots based on artificial intelligence are far better at interacting with humans.
API – Application programming interface is a set of tools that allows apps to communicate with other services. It allows developers to easily extend existing software with new features. For example, Facebook’s open API allows third-party chatbot makers to integrate them into Facebook Messenger.
Artificial Intelligence (AI) – A field of computer science concerned with providing computers and machines with what can be perceived as human intelligence. Artificial intelligence enables chatbots to interpret linguistic expressions, translate between languages and make their own decisions.
Chatbot – a chatbot, or chatbot, is a program designed to mimic human communication behaviour. Chatbots can be integrated into a website or into various messaging services, such as Facebook Messenger, and can then respond to different questions. The chatbot either handles the entire conversation or, if a question arises that it fails to answer, the matter is passed on to a human.
Chatbot builder – A ready-made technology platform that allows users to build and launch a chatbot without in-depth programming skills. Several providers of chatbot platforms are currently on the market.
Entities – Keywords and phrases that have similar meanings and are important to your specific business. For example, these could be the names of products or services you offer. For example, by creating lists of entities, the chatbot can show the product to the customer at the right moment.
Fallback – When the chatbot is unable to identify and respond to the customer’s message, the fallback is the default response delivered, for example: “I don’t quite understand what you mean, can you ask me again?”. In this situation, many providers offer a so-called Hand Over to a human agent who helps the customer instead.
Chatbots: what they are for and how they can be used by companies
Intent – Intent refers to the goal the customer has when writing a question to the chatbot. In order for the chatbot to be able to successfully determine what the customer is looking for, it must be skillfully programmed and also trained with as large a data set as possible.
Greeting – The short message that welcomes visitors to the chatbot and encourages them to ask a question.
Contextual understanding – A challenge for chatbots is to interpret the question based on the context in which it is asked. Chatbot developers solve this by allowing the bot to analyse conversation history, timing, tone of conversation, sentence structure and more. Contextual understanding can be offered in some Chatbot Builders to create conversation topics.
Conversational channel – The channel used for communication between human and bot. It can be directly on a website or in services such as Whatsapp, Facebook Messenger, Slack, Kik or over text messages.
Machine learning – A field of AI that allows chatbots to identify patterns in large amounts of data and learn from previous conversations with humans. This means that the chatbot gets better at delivering relevant answers the longer it is used.
NLP (natural language processing) – An area of artificial intelligence that involves giving computers the ability to understand human language, recognise linguistic patterns and convert text into speech. It is NLP that determines the intent of a conversation.
NLU (natural language understanding) – Although computers are good at recognising and interpreting human language, interpreting our intentions (see Intents above) presents a much greater challenge. Thanks to NLU, the chatbot can often understand the intent even if we misspell, omit certain words or express ourselves ironically or humorously.
Sentiment analysis – Through advanced linguistic analysis, the chatbot can interpret the customer’s mood and determine whether he or she is satisfied or dissatisfied with a particular product or service. This can to some extent replace the traditional type of customer satisfaction surveys.
Chatbot: What is it, what is it for and how do they work?
Transfer – When the chatbot is unable to resolve an issue on its own, it can be passed on to a human contact centre agent through a so-called transfer.
Typing delay – Adding a delay before the chatbot responds to a message from the visitor can make the conversation flow more naturally.
Webhook – An automated transfer of the information the customer types in a message to the chatbot, such as an email address or phone number. The information is sent to one of the company’s other systems, such as a crm system, and can be used to improve customer service.
Quick reply – By displaying a number of pre-programmed, possible quick replies that the customer can click on, the issue can often be resolved more quickly than if the customer types in all the answers themselves.
Would you like to create your own chatbot development? Contact Enkronos team today.
