Publication monitoring

Repeatable searches for new publications under fixed criteria.

Result

You produce a dated update that distinguishes new search results, screened records, included publications, exclusions, and changes since the previous run. Another researcher can repeat the search from the recorded protocol.

Run record

Record each database, exact query, filters, date and time, result count, export file, deduplication rule, screening decision, reason for exclusion, and stable identifier. Preserve the raw export separately from the reviewed selection.

Actions

1. Define the research question, databases, date range, languages, publication types, and inclusion and exclusion criteria.
2. Copy the exact search strings and filters into the run record before execution.
3. Run each search and preserve its raw export or result snapshot unchanged with the run date.
4. Normalise and deduplicate in a working copy while retaining every original database identifier.
5. Screen records against the stated criteria and record a reason for each exclusion at the agreed stage.
6. Compare included identifiers with the previous accepted run and classify additions, removals, and changed metadata.
7. Produce a change table with search date, source database, stable link or identifier, decision, and reviewer.

Check

A second person can reproduce the query and reconcile raw count, deduplicated count, screened count, included count, and exclusions. Every reported addition maps to a raw record and decision. A title, abstract, or link is not represented as evidence from a full text unless the full text was reviewed.

If it fails

Preserve the failed or null run. If results are too broad, change one query element or filter at a time and record the revision. If too few, report the null result before relaxing a criterion. If a database or export failed, do not merge partial results into the accepted update without labelling the gap.

Next: Literature review.