This article covers how scite works by explaining how we are building out our database of citations, and how we classify them into supporting, mentioning, and disputing.

How does scite work?

In order to build out our database of Smart Citations, we access full-text PDFs and XMLs directly through indexing agreements with various publishers and other sources (such as preprint servers).

You can find how scite works in detail here: https://direct.mit.edu/qss/article/2/3/882/102990/scite-A-smart-citation-index-that-displays-the

Let's say that there is a source paper which makes a citation to a target paper.

scite will:
access the full-text articles of both papers
detect that the source paper makes a citation to the target paper
extract the text from the source where the citation happened
internally track, for the target paper, all of the different source papers which cited it and the information around each citation

Each citation it extracts is referred to as a Smart Citation because it contains information about:
which papers are citing each other
the textual context of the citation and the location within the citing paper
a classification describing whether the citation is supporting, mentioning, or contrasting the cited work's claims

Why aren't some papers showing up?

We are indexing new content every day, and any gaps will be filled over time as we form more partnerships and gain access to more papers. If you are part of a journal or research organization publishing papers, we'd love to get in touch and partner to index your content here: drop us a line at hi@scite.ai

How are citations classified?

Check out our article here: How are citations classified?