Colandr

What Colandr is

Colandr is a free, web-based platform for evidence reviews and other document reviews and syntheses. It uses machine learning to rank citations by relevance, helping reviewers screen titles and abstracts, while natural language processing supports classification into user-defined categories and full-text data extraction. Reviewers retain the final decisions about inclusion, exclusion and classification, and teams can collaborate and cross-check one another’s work. Colandr accepts citation imports in TXT, RIS and BIB formats, with a 40 MB limit per import, and exports screening decisions and extracted data as CSV. It does not connect to citation databases, so reviewers export citations from another system; the platform can help generate Boolean search strings. Users must locate full-text papers themselves and import them one at a time, as there is no mass-upload feature. Relevance rankings update after every 10 citations are screened, while metadata extraction suggestions begin after 50 hand-coded extractions in a review. Colandr describes itself as open-source and open-access, with front-end and back-end code repositories. Its community site says it runs on an AWS server hosted by Conservation International and launched in July 2017. The team directs users to a help email and bug-reporting form, noting that volunteers run the project.