Literature pipeline

From a research area to a review you can defend.

Search, screen, full text, extract, categorise, questions, report. The search step runs here: describe an area in your own words and it writes eight candidate searches, counts each one across the databases, and has a sample judged record by record.

Start a search See the pipeline
example One area searched four ways: wider wording finds more papers and a smaller share of them belong.
Wording of the searchPapers foundOf those, relevant
the exact terms only93996%
with the usual synonyms40,80188%
plus the neighbouring field135,49561%
the widest wording tried446,94034%

The pipeline

in the browser Search A scope becomes candidate searches, counts per database, a judged sample and an exported list. in the browser Screen Every record decided in or out against a criterion you wrote, with the reason kept, and the criterion measured against a sample you label yourself.
script Full text PDFs fetched for what passed, and the text pulled out of them.
script Extract Method, data and finding lifted from each paper into the knowledge base.
script Categorise Subareas induced from what was extracted, then every paper assigned to one.
in the browser Questions Research questions drawn from what the area has not answered.
script Report The written review, with the evidence behind every claim kept next to it.

Where it looks

Recent runs