State Machine Based Human-Bot Conversation Model and Services

Task-oriented virtual assistants (or simply chatbots) are in very high demand these days. They employ third-party APIs to serve end-users via natural language interactions. Chatbots are famed for their easy-to-use interface and gentle learning curve (it only requires one of humans’ most innate ability, the use of natural language). Studies on human conversation patterns show, however, that day-to-day dialogues are of multi-turn and multi-intent nature, which pushes the need for chatbots that are more resilient and flexible to this style of conversations. In this paper, we propose the idea of leveraging Conversational State Machine to make it a core part of chatbots’ conversation engine by formulating conversations as a sequence of states. Here, each state covers an intent and contains a nested state machine to help manage tasks associated to the conversation intent. Such enhanced conversation engine, together with a novel technique to spot implicit information from dialogues (by exploiting Dialog Acts), allows chatbots to manage tangled conversation situations where most existing chatbot technologies fail.