Using the Scholarcy Chrome Extension
Our Chrome extension processes the contents of the current page in your web browser, if it is publicly accessible and not behind a login/paywall. You can also upload files directly from your computer. In both cases, your data will be processed on our server, deleted, and the extracted information is returned to you. When you click on the Scholarcy extension and choose ‘Options’, you can modify the way Scholarcy processes your content.
We use Google Analytics to gather anonymised statistics on the features that are being used (or not being used), to help us prioritise improvements to the service. None of this data is shared and we do not use this data for marketing or advertising purposes. Please read our Privacy and Cookies Policies, and also our FAQ for more information on how your data is processed.
By default, all options apart from Extract figures are enabled.
Basic options
Generate summary
This creates the Scholarcy summary section, comprising a summary of the main document text with a fixed word limit set by the word count option, or as a fraction of the original document set by the % option. If set, the % will override any value for the fixed word count. However, when the Structured summary and Extract snippets options are checked (default – see below), setting a target summary length will have less of an effect, depending on the structure and length of the original article.
Summary engine
This can be set to stable, which gives a good result for a broad range of documents, or experimental, which may work better for longer, more narrative texts.
Rewrite in 3rd person
This rewrites the Scholarcy summary section into a neutral third person (for example ‘The authors’ results suggest …’) to make it easier to quote from and reference sections of the paper.
Skip dialogue/quotes in summary
When selected, any dialogue or long sections of quoted text will not be considered as input to the summary engine, so only descriptive text will be summarised.
Structured summary
When checked, the AI will attempt to structure the summary into Introduction, Objectives, Methods, Results, Conclusion.
Extract sections
This identifies the main sections in the document (Introduction, Methods, Conclusion etc) and either creates summary sections for each of these when Extract snippets is checked (default), or outputs the full section text when Extract snippets is unchecked.
Extract statistics
This identifies population, intervention and outcome measures and reports these in a Participants and statistics section.
Extract tables
This identifies tabular data and their captions, showing the captions in a Tables section along with a download button to save the tables to an Excel file on your computer.
Extract figures
This extracts figures and their captions into a Figures section. You can then click on each figure to show a larger version in a new browser tab. Each Figure will be linked to callouts in the text. Figure extraction is very CPU intensive and, for large documents with high-resolution images, may not always be successful, so it is disabled by default.
Highlight facts
This highlights in yellow important facts stated by the paper.
Smart highlighting
Individual facts may span a few words and may lack context, so Smart highlighting extends the fact to a complete clause.
Highlight contributions
This highlights in pink the main goals that the authors set out to achieve and the main contributions of the paper.
Link citations
This creates links between inline citations such as [3, 4] or (Smith 2007a) to the relevant bibliographic entry.
Key points
Determines how many key points are displayed in the Scholarcy highlights section – 5 (default) is a good starting point for most articles.
Advanced options
You probably won’t want to or need to change these options very often, but you may like to experiment with them to speed up processing or improve the results.
External metadata
This locates metadata from external sources such as CrossRef, arXiv and others to correct the results of the automated extraction of title, authors and abstract where necessary.
Require captions
If checked, only extract images and tables that have associated captions.
PDF engine
Reflowing text in the correct order from PDF files is surprisingly difficult. You can change this to Experimental which may give improved results on ‘difficult’ PDFs.
Background reading list
- Fast: maps key terms to a Wikipedia search query but does not attempt to verify or disambiguate the results
- Precise (default): maps key terms to the correct Wikipedia entry, which may add a few seconds to processing time
- Broad: A combination of the Fast and Precise, and includes Wikipedia entries that include the key term in the title
- Broader: As for broad, but also includes Wikipedia entries that include the key term in the entry.
Key terms
- Narrow (default): Identifies specific key terms in the document
- Broad: As for narrow but also includes acronyms
- Broader: As for broad but also includes frequent noun phrases
Sampling
- Representative (default): Harvests key terms from a representative sample of the document
- Full text: Harvests key terms from the full document
Summary of cited works
- Brief (default): Displays the key findings of the cited work
- Detailed: Displays a longer summary of the cited work
