The search phase of software engineering systematic literature review: barriers and solutions
The Systematic Literature Review (SLR) is an approach for conducting literature reviews that provides less bias and more reliability than the more typical ad hoc approach. One of the first phases of the SLR process is to conduct a comprehensive search for current literature. Conducting this search on different sources of literature (i.e., digital libraries) is a manual and exhausting task that results in an SLR process that is more costly than necessary. Therefore, the goals of this dissertation are to: (1) find empirical evidence about the SLR search problem and the current status of tool support; (2) understand the barriers and the tooling requirements to support the SLR search phase; and (3) develop and evaluate a solution to help reduce the barriers. To address these goals, this dissertation consist of three articles. Article 1 describes the results from three empirical studies that produce a list of tooling requirements across the entire SLR process. The results of these studies identify numerous gaps between needs of SLR authors during the search phase and the current tool support. Article 2 consists of an SLR and a survey, to identify the specific tool requirements to support the SLR search phase. The SLR produced a list of SLR search barriers that were reported in the literature. The survey of SLR authors confirmed the results of the SLR and expanded the list to include issues and requirements from the community beyond what is reported in the literature. Article 3 describes the development and evaluation of the Searching Systematic Reviews Tool (SSRT) tool to address the problems identified in Articles 1 and 2. SSRT provides one interface to search multiple digital libraries at once, store the results, and remove duplicates. The article also describes the evaluation of SSRT and the future extensions of SSRT.