On this page
What ASReview LAB does
ASReview LAB is an open-source, browser-based research tool from Utrecht University’s AI Lab. In its screening workflow, a reviewer labels records and an active-learning model ranks remaining items for review. The project dashboard supports progress monitoring and the dataset can be exported. ASReview can help prioritize screening, but does not create the literature search, decide eligibility for you or prove that all relevant studies have been found. This profile is based on official documentation; we did not run a review or benchmark the model.
Where it fits, and where it falls short
Useful strengths
- The human reviewer supplies relevance labels in an interactive active-learning loop.
- A dashboard offers screening progress information and a configurable stopping threshold.
- The documented export includes imported data and labels in formats including RIS, CSV, TSV and Excel.
- The project is open-source and offers a separate self-hosted server option for multi-reviewer workflows.
Limits to consider
- It prioritizes records you have already collected; it does not establish that database searches were sufficiently broad.
- A stopping threshold is a decision aid. ASReview documentation says the ideal threshold remains under research.
- The screen order and model output are not a substitute for protocol-defined inclusion criteria or required independent screening.
- The local installation route requires technical setup; confirm whether the browser-based hosted experience is available and appropriate for your data.
A suggested ASReview LAB workflow
View the workflow diagram
- 1
Run the databases and other searches required by your review, and keep the database, query, date and original exports.
- 2
Prepare and import the records in a supported format; check the fields and identifiers before screening.
- 3
Label relevant and irrelevant records using written criteria, reviewing the prioritized queue while documenting uncertain cases.
- 4
Set and justify a stopping rule using your protocol and project context, then export and retain the records, labels and remaining screening decisions.
Use active learning to prioritize a screening queue
ASReview LAB is designed for systematic screening and labeling of text records. Its active-learning cycle begins with human labels: the reviewer marks records, and the model uses those decisions to rank which undecided text is more likely to be relevant. The researcher remains in the loop and supplies the decisions; a high position in the queue is not itself an eligibility judgment.
This is most useful after you have assembled a search set from the databases and sources your review requires. Keep the query, database, date, deduplication method and imported files outside the prioritization model as part of the review record. A focused queue can make screening easier to manage while leaving the search coverage question unanswered.
Define inclusion and exclusion rules before labeling. If a title or abstract does not contain enough information, record it as uncertain or follow your protocol’s full-text route instead of converting missing information into a confident exclusion. Keep the reason for each decision traceable to the criterion and source record.
Treat the stopping threshold as a documented judgment
The progress dashboard provides statistics and charts and allows a reviewer to set a threshold based on the number of consecutive irrelevant records since the last relevant one. The threshold is not set by default. The documentation says reaching it prompts the reviewer to decide whether to stop or continue, and notes that the ideal threshold is still being researched.
Do not present a reached threshold as proof that every relevant record has been found. Choose and justify a stopping approach with the review protocol, screening volume, topic and consequences of a missed study in mind. Check records you expected to find and investigate any relevant known items that appear late or are absent.
For a formal systematic review, follow the protocol and reporting requirements your project uses. Record the chosen stopping rule, reviewer judgment and unresolved items. An active-learning dashboard is a workflow aid, not a guarantee of recall or methodological completeness.
Supporting documentation: [2]
Export the labels and preserve the search separately
ASReview documentation says a dataset can be exported at any point, including the imported data, labels and additional variables. The listed export formats include RIS, CSV, TSV and Excel. Inspect the actual export and retain the original source files so the team can recover both the initial records and the decisions added during screening.
Keep search provenance with the project: database names, exact query strings, dates, filters, duplicate handling and any supplementary search route. The screening export is not, by itself, a complete search strategy or a record of why a database query was chosen.
If several reviewers are involved, verify the selected deployment and collaboration method before importing a live project. ASReview describes a self-hosted Server edition with authentication and multi-reviewer screening. We did not configure a server, inspect account permissions or verify a hosted team workflow.
Supporting documentation: [2]
Check the installation route before adopting it for a class
The documentation describes ASReview LAB as open-source and browser-based software, with installation guidance for running it locally. The project website also presents a live demo and a separate server option. These routes have different implications for setup, collaboration, institutional data policies and support; verify the current configuration before assigning students a workflow.
For a small classroom exercise, use a permitted sample set and write criteria that distinguish a clear include, a clear exclude and an unresolved abstract. Compare the screen order with a known relevant paper and keep the original set. This is a suggested exercise, not a test we performed and not a validation of model recall.
ASReview fits a screening-prioritization task. Rayyan may better fit a team that needs an interface centered on review stages and conflict reconciliation; Elicit offers a different research workflow that includes discovery and structured evidence extraction. Compare by the record and oversight process your project must retain.
Pricing, free access and plan checks
ASReview LAB is described by its documentation as open-source software, and the project also documents a self-hosted server edition. The public pages reviewed here do not establish a hosted personal subscription price. Check the installation and hosting route, current deployment requirements and institutional support before planning a team workflow. We did not install or run the software.
Confirm billing period, renewal terms, limits and regional availability on the provider’s current plan page before paying.
Questions about ASReview LAB
Does ASReview find all relevant papers for my review?
No completeness guarantee is established by the screening workflow. ASReview prioritizes records you import. Run and document the searches your review requires, check known relevant papers and follow your protocol.
Does the AI decide which studies to include?
The documented active-learning workflow ranks undecided records using labels supplied by the reviewer. You apply the inclusion criteria and make or verify the screening decisions.
Can I stop when the dashboard threshold is reached?
The threshold is a prompt to consider stopping, not proof that screening is complete. The documentation says the ideal threshold remains under research. Justify the decision under your protocol and project context.
What can I export from ASReview LAB?
The documentation says the dataset export contains imported data, labels and additional variables, with RIS, CSV, TSV and Excel among the formats. Retain your original records and search log separately.
Is ASReview free?
ASReview LAB is described as open-source software. Hosting, setup and support depend on the route you choose. The pages reviewed here do not establish a hosted personal price; check current installation and institutional options.
Did AI Essay test screening accuracy?
No. This is a source-based product profile. We did not install ASReview, run a screening project or measure recall, precision or time savings.
Sources and research notes
This guide synthesizes official product information with our editorial assessment of academic use. Suggested exercises are examples you can repeat; they are not results of tests we performed. We do not infer accuracy from a provider’s marketing claims.
- ASReview LAB documentation: purpose, active learning and human labels
- ASReview LAB documentation: dashboard, stopping threshold and exports
- ASReview: screening workflow, installation and project options
Found a changed feature or a factual error? Send the exact page and an official source to info@ai-essay.com. Read our editorial policy for how we handle corrections.