调查问卷中开放题文本答案编码方法研究综述

A Review of Text Answer Coding Methods for Open-Ended Questions in Questionnaires

调查问卷作为联系调查者与被调查者之间的“纽带”,是采集数据、发现因素间内在联系和规律的有力工具。对其中的开放题收集的答案文本数据进行编码,有助于了解文本答案蕴含的固有特征,便于后期开展统计分析,是开放题研究的热点问题。为此,从人工编码、半自动编码及自动编码三个方面对开放题文本答案的编码方法研究进行了归纳、总结与展望,为进一步开展相关研究奠定良好的基础。结果表明:关于开放题文本答案的编码方法研究,经历了从最初的完全人工编码到“机器 + 人工”的半自动编码,逐步发展到目前“基于人工智能”的自动编码探索三个阶段;虽然编码效率得到了极大的提升,但是由于缺乏普适性的编码方法,因而不同领域开放题文本答案编码的准确性存在较大差异,提高编码的普适性、准确性仍是未来研究的重点。

As the “link” between the investigator and the respondent, questionnaire is a powerful tool to col-lect data and find the internal relations and rules among factors. Encoding the answer text data col-lected by the open-ended questions is helpful to understand the inherent characteristics of the text answers, which is convenient for later statistical analysis, and is a hot issue in the study of open questions. Therefore, this paper summarizes, concludes and prospects the research on the encoding methods of open-ended question text answers from three aspects: manual encoding, semi-automatic encoding and automatic encoding, which lays a good foundation for further related research. The results show that: As for the research on the encoding method of open-ended ques-tion text answer, it has experienced three stages from the initial completely manual encoding to the semi-automatic encoding of “machine + manual”, and gradually developed to the current automatic encoding exploration of “artificial intelligence”. Although the coding efficiency has been greatly im-proved, due to the lack of universal coding methods, the accuracy of open-ended question text an-swer encoding in different fields is quite different. Improving the universality and accuracy of cod-ing is still the focus of future research.