Quality Assessment Methods for Conversational Interfaces (bot)

Virtual assistants are being part of our everyday life.A computer program that simulates human conversations, by analysing and identifying their requests and giving to them the correct answer, is becoming important and a must to every business.Through Natural language processing, the bots are designed to convert text or speech from user queries into structured data.The data is used to choose a relevant answer for each request of the end user.They bring to the business cost savings by offering to the customers convenience and many services.It helps to resolve many types of user queries and issues while reducing the need for the human interaction and making the conversation friendly and useful for the users in order to accomplish their requests.Some of the main key attributes that makes chatbots very important are the reduction of the customer waiting time, the 24/7 availability to engage customers with immediate answers and easy scalability. User experience is an important aspect to take in consideration while creating a chatbot with a specific or general purpose.It should give to the user a good performance, functionality and efficiency during the conversation flow.In order to have a high user experience, before launching a specific bot is important to make a lot of tests and to make sure that all his functionalities works on time and in an efficient way. During this thesis work is presented a systematic literature review for the quality issued and attributes related to the contemporary issues of the chatbot development and is presented a methodology to evaluate the quality of the chatbots by using some attributes and their metrics during a conversation flow with an end user. Manychat, Alexa and Dialogflow skills are used to present the flow that this methodology follows for giving a result for the performance and the quality of a chatbot during the interaction with the user.