How to build a chatbot. A semiotic and linguistic approach

Chatbots are Artificial Intelligences able to simulate a written or vocal conversation in natural languages through software, applications or websites. Their aim is to provide users with the information they need through a faster and simpler user experience. Many businesses and organisations use chatbots as tools to improve user experience, provide 24/7 customer service, simplify online shopping and payments, manage reservations and so on. Semiotic research has recently analysed chatbots. Its aim has mainly been to advance the understanding of how to make the conversation of chatbots “more human” as possible in order to avoid disruption and frustration in users. Semiotic analysis is accurately used to respond to this challenge, focusing on how humans make meanings and interpret them, against a machine centred approach. However, it makes theoretical generalisation from analysis of existing chatbots in specific domains, e.g. healthcare. The domain in which a chatbot operates is important, so is the cultural context, that is notably absent in previous semiotic analysis of chatbots. This paper presents a semiotic and linguistic approach to chatbot design that focuses on three dimensions: a) the user/chatbot goal; b) the target user, her context and interpretations; c) the chatbot personas. This approach can prove useful to identify how users make and interpret meanings during conversations with chatbots in a given social and cultural context.