Dissect

Reading a paper properly, without reading all of it.

Library

Loading the library

Add a paper

Drop a PDF here, or click to choose one

The same file uploaded twice reuses the existing parse.

Or paste an arXiv id, a DOI or a link

Works with 2407.01449, arXiv:1706.03762, a doi.org link, or any link ending in .pdf.

ExtractionBuild the vector indexRuns the bundled local encoder. No key, no cost. Off leaves keyword search only, which is faster and still finds exact values.Read text inside figuresOCR over every figure and any scanned page. Slow, and the only way to search a chart's axis labels.

Deep extraction is unavailable: