Date created: | Last Updated:

Category: Project

Description: Abstract Objectives: To adapt, validate and refine topic search filters meant to retrieve Swiss social work bachelor theses focusing on children’s homes written in French and indexed in RERO DOC Study design: Search filter external validation, refinement and internal validation Methods: Full-text assessment of all 385 bachelor theses published by HETS Valais students and indexed in RERO DOC, extraction of titles and assessment of whether they focus or not on children’s homes. Adaptation of the two previously built filters for RERO DOC, measurement of their performances on this new dataset. Development of three new Results: Performances of the original topic search filters dropped significantly when applied to this external validation set. The highly sensitive children’s homes search filter performance went from 79.45%/95.12% sensitivity/specificity to 66.32%/78.89% and the highly specific search filter performance went from 69.86%/98.37% sensitivity/specificity to 54.74%/83.39% when tested on the RERO DOC dataset. Reasons included: a lower prevalence of specific keywords within children’s homes titles (now out of date or highly canton-specific) and limited functionality of searches making use of character accentuations. A highly sensitive (“exhaustive”), a highly specific (“focused”) and a balanced (“best trade-off”) RERO DOC specific search filters were developed iteratively using a search simulator. The “exhaustive” search filter achieved 87.37% sensitivity and 71.88% specificity. The “focused” search filter achieved 62.11% sensitivity and 93.40% specificity. And the “best trade-off” filter achieved 74.74% sensitivity and 84.38% specificity. Novel keywords included in these filters include: reference, fugue, mineur, rupture, plainte and St-Raphael/Saint-Raphael. Limitations: The RERO DOC search filters were used without using data from abstracts or the full extent of advanced search features available and could be improved further. The new search filters should be used with caution on different databases, if used to search works from non-Swiss settings (for instance works from France), if used to search for works published prior to 1980/1990 or if used to search for works published in the following decades. Funding: No funding was received for this work. Registration: Not done, previous report versions nevertheless available at https://osf.io/96b5h/ . Data and materials: See https://osf.io/96b5h/files. All other data is otherwise included within this manuscript. Keywords: Children’s homes, search filter, sensitivity and specificity, indexing, literature searching, RERO DOC