A Review of Research on Coding Methods for Open-ended Text Responses in Survey Questionnaires

Abstract

In fields such as social sciences and market research, open-ended questions can collect richer data information, but how to effectively count and analyse these text answers becomes a key issue. The study mainly explores the three coding methods of open-ended questions in questionnaires, including the definition, process, and application of manual coding, semi-automatic coding, and automatic coding. According to existing literature and data, manual coding has high flexibility and accuracy, but it is inefficient when processing large-scale data; semi-automatic coding combines manual coding and machine learning technology, which can improve efficiency while maintaining a certain degree of accuracy; automatic coding relies on natural language processing technology and deep learning models, which greatly improve coding efficiency, but there is a problem of insufficient accuracy when facing complex semantics. Future research can focus on improving the accuracy of automatic coding through deep learning, developing intelligent semi-automatic systems that reduce manual intervention, and incorporating real-time feedback mechanisms for continuous misappropriation.