Table of Contents
This tutorial assumes you have already installed Python and ASReview. If this is not the case, please check out the installation page.
Launch ASReview LAB by running the following command in the command line (CMD.exe for Windows or Terminal for MacOS/Linux):
To start screening, a project file needs to be made. The following steps show how to create a project.
Start a new project by clicking the red button in the bottom right corner.
Next, provide a project name, your name and a short description on your systematic review project.
After a successful project initialization, a project dashboard will be shown. Click on the Start Setup button to add a dataset and prior knowledge.
Select the dataset you want to review, which should contain at least the titles and/or abstracts of all documents or publications you want to screen.
There are four ways to select a dataset:
- upload your own dataset. Read more about the format on our datasets page.
- import a dataset with an URL
- select a dataset from an extension (for example our COVID-19 extension)
- choose one of the built-in test datasets. Read more about testing on our user testing page.
After a successfull upload of the data, move to the next step.
Our active learning models require prior knowledge to work. This knowledge is used to train the first algorithms. In this step you need to provide at least one relevant and one irrelevant document. To facilitate this, it is possible to search within your dataset or ask the software to present a couple of random documents.
Let’s start with finding a relevant document. Probabily the most efficient way to do this is by searching for a document or publication you know is relevant. Click the search button and search your dataset by authors, keywords and title, or a combination thereof. Enter your search terms (Use ‘enter’ to start searching).
Click the publication you had in mind and click Relevant (Clicking Irrevant results in an irrelevant document).
The Prior Knowledge step will know show 1 relevant document. Repeat this step until you’ve selected your 1-5 relevant documents. Note that there are no restrictions on the number of publications you need to provide, but preferably provide 1-5 relevant documents.
You also need to provide some irrelevant documents. Given that the majority of publications in the dataset is probably irrelevant (spare problem), the publications presented here will most probable be irrelevant for your study. Click on random to present a couple of random documents.
Indicate for each publication whether it is relevant or irrelevant to your systematic review.
After labeling a couple of randomly selected documents, ASReview LAB will asked you whether you want to stop. Click on Stop and go to the next step.
.asreview project file by clicking the red button in the bottom
right corner of the home page. Choose the project file from your computer and
click Import. After a successful project initialization, a project dashboard
will be shown.
As soon as the active learning model is ready, a button appear with Start Review. Click the button to start screening. ASReview LAB presents you a document that the software is most confident about in terms of relevancy for your systematic review.
You are asked to make a decision: relevant or irrelevant for your systematic search?
While you review the documents that the software presents you with, the software continuously improves its understanding of your decisions, constantly updating the underlying model.
The Statistics panel shows information on the current project and labeling progress. The panel can be opened and closed during labeling as you like. To hide your progress, click the arrow in the upper right corner.
The software keeps presenting you with the most relevant publication until there are no more publications left to review, or until you decide to stop reviewing.
As you keep reviewing abstracts, your set of relevant documents will increase while the number of unlabeled abstracts left in the dataset will decline.
Now, in a ‘traditional’ systematic review, you would continue reviewing until you have seen all abstracts in the dataset. However, ASReview orders the publications in such a way that you see the most relevant publications first. In other words, there is a point where you can be confident that you have seen (almost) all publications you need, and you can stop reviewing.
When to stop is left to the user. A stopping criterium could be stopping after the last X presented abstracts were labeled irrelevant, or if your time is up. You can use the chart in the statistics panel to follow your progress.
Open the project dashboard by clicking the menu icon in the upper left corner. Click the download button next to Start Review. A dialog will show the download options.
Choose from the menu whether you would like to download your results as a CSV or an Excel file and click Download. A file is downloaded with the results of your review.