Automation of systematic literature reviews: Development of a gold standard of articles in PubMed 2020–2021, a research protocol

Abstract

Evidence-based practice is an expectation that is becoming common in all fields. The systematic literature review (SLR) method aims to synthesize the scientific evidence available on a given scientific question reliably and transparently. As the amount of information available increases, the demand for reliable research results increases accordingly, the importance of SLRs is also growing. To reduce the cost of the SLR process, several automation methods have been presented by computer scientists. Given the proliferation of AI techniques and SLRs on their use in a wide range of applications, identifying articles where AI is specifically used to assist the SLR process is becoming challenging. It is important to find these articles so that we can follow the development of the field, and to be able to summarise the results of the researches. A solution to this problem would be to develop a filter to help find literature on the subject. A gold standard collection of all articles is essential to develop such a filter. This project aims to develop a gold standard collection of articles on the automation of SLRs from PubMed published between 2020 and 2021. To support this goal, a methodology is developed that combines manual work, text analysis, and the snowball method in an iterative search process, which is presented in the following study protocol.

Ähnliche Arbeiten