Crowdsourcing Scholarly Discourse Annotations
summary
A web interface asks authors to mark key sentences in uploaded PDFs, converts those annotations into a scholarly knowledge graph, and limits selections to a suggested maximum of two sentences with a warning if exceeded. Machine help includes extractive summarizer highlights and zero-shot classifier class suggestions drawn from a 25-class Discourse Elements Ontology. A study with 23 researchers yields feasibility evidence: usability score 76 (SUS) and workload 35 (NASA‑TLX), and most participants accept annotating at camera‑ready submission.
AI-written summary · about
related talks
20mKnowledge Scout: From Talks and Slides to Smart Summaries
12mGarden of Papers: Finding, Reading, and Organizing Research Papers in a Visual, Integrated, and F...
13mPaperBridge: Crafting Research Narratives through Human-AI Co-Exploration
12mSynthia: Visually Interpreting and Synthesizing Feedback for Writing Revision
18mFacilitating Document Reading by Linking Text and Tables
13mAbstractExplorer: Leveraging Structure-Mapping Theory to Enhance Comparative Close Reading at Scale
16mInteract with Your Technical Documentation via Generative AI
08mMedKnowts: Unified Documentation and Information Retrievalfor Electronic Health Records