Chatbots are computer programs designed to chat with users via text or voice through the use of techniques of Web services, data analysis, and artificial intelligence (AI). Currently, the use of chatbot is becoming an important trend in the field of data science. An increasing number of chatbots are being built on social platforms, such as Facebook, LINE, and Slack. This has led to the development of numerous tools and online platforms for the construction of chatbots; however, most of these services do not provide comprehensive support for the visual representation and control of conversational flows, bi-directional Web service integration, or the systematic reuse of conversations. Developing a complex chatbot with external Web services requires the writing of extensive conversation scripts and additional coding. In this paper, we propose a visual, flow-based approach to the construction of chatbots on the Node-RED platform, referred to as FCF (Flow-based Chatbot Framework). This system is based on the newly-devised data format for Webhook, thereby allowing bidirectional service integration for software applications. Five chatbot dialogue patterns and three chatbot application scenarios are provided to be components for the construction of complex chatbot applications.