Abstract
- The creation of scientific research proposals and publications leads to a high effort for scientists, which has increased significantly in recent years. At the same time, technologies from the fields of information retrieval, machine learning, and semantic technologies offer potentials to relieve scientists and to increase the quality of scientific texts. Taking these technologies into account, this paper focuses on the determination of requirements for a generalistic system that enables the automated assessment of the text quality of scientific texts. First, a criterion-based requirements analysis was performed. As a result, ten specific criteria were identified and enriched by quality-dependent criteria. These formed the basis for the derivation of a total of 26 non-functional and qualitative requirements. These were analyzed, and dependencies were checked and transferred into a total of four requirement clusters. The requirement clusters comprise the requirements for data quality, text analysis, text quality assessment, and other qualitative requirements. Thus, a basis for the creation of a system architecture was created, which annotates, trains and evaluates texts, and contributes to the assessment of the text quality of scientific texts.