A Visualization‑based Approach to Present and Assess Technical Documentation Quality

Abstract

Technical documentation has moved from printed booklets to electronic versions that need to be updated continuously to match product development and user demands. There is an imminent need to ensure the quality of technical documentation, i.e., information that follows a product.In order to ensure the quality of technical documentation, it is important to be able to measure it in a constructive way. In this paper, we approach quality from a software quality perspective and rely on automated measurements and analyses. It is generally not possible to assess quality attributes such as “ease of understanding” using automated measurements. To assess such quality attributes, we suggest the use of visualizations as a communication medium between the machine results and technical writers, and define a visualization‑based quality assessment approach for technical documentation.In order to test our approach, we use it to assess the quality of 3 real‑world documentations from a Swedish mobile phone manufacture, a Japanese camera manufacturer, and a Swedish warship producer. The study shows that our approach can be used to identify potential quality defects. For example, we tested an unclassified subset of the warship’s technical documentation and found that 49% of it was redundant text clones. We performed the study in collaboration with a Swedish company that is in charge of creating and maintaining the 3 documentations, and they acknowledge that our approach has great potential and that our results proved helpful to them.