addressalign-toparrow-leftarrow-rightbackbellblockcalendarcameraccwcheckchevron-downchevron-leftchevron-rightchevron-small-downchevron-small-leftchevron-small-rightchevron-small-upchevron-upcircle-with-checkcircle-with-crosscircle-with-pluscrossdots-three-verticaleditemptyheartexporteye-with-lineeyefacebookfolderfullheartglobegmailgooglegroupshelp-with-circleimageimagesinstagramlinklocation-pinm-swarmSearchmailmessagesminusmoremuplabelShape 3 + Rectangle 1ShapeoutlookpersonJoin Group on CardStartprice-ribbonShapeShapeShapeShapeImported LayersImported LayersImported Layersshieldstartickettrashtriangle-downtriangle-uptwitteruserwarningyahoo

Rebecca Passonneau on Natural Language Understanding

Wisdom-of-the-Crowd Content Assessment: Human-Mediated Text Representation

An influential cognitive theory of human text comprehension developed in the 70s and 80s by Kintsh and van Dijk still guides research on student learning of reading and writing skills. It posits a conceptual model that differs from the kinds of text representation prevalent in AI and NLP research at that time, or used now, in that it depends on what is in the text, what is inferrable from the text, and what the reader knows. This talk will present an approach to content assessment of text that was originally developed to evaluate machine-generated summaries. It is distinctive in its reliance on model summaries written by a crowd (proficient individuals working independently). The crowd supplies an emergent property of observed content: relative importance. The talk will present similarities of this content assessment approach to coding rubrics used in research on reading and writing skills; two automated approaches to the content assessment; results of a recent experiment comparing the automated approaches to a content coding rubric applied to summaries written by community college students. The conclusion will touch on questions of the feasibility and utility of deriving such a content model without the intermediary wise crowd.

Rebecca J. Passonneau is the Director of the Center for Computational Learning Systems (CCLS) at Columbia University, and a Senior Research Scientist.  Her research addresses how the same combination of words has different meanings in different contexts, for both written and spoken language. This question is a particularly challenging object of study, because context has many dimensions. Her recent work investigates content assessment for summarization and reading and writing skills, meaning and action in human-machine dialogue, data mining that links textual and non-textual sources, and word sense annotation and representation. She received her Ph.D. from the Department of Linguistics at the University of Chicago in 1985, and worked at several industry and academic research labs before joining CCLS in 2006. She has over 100 publications in journals and refereed conference proceedings, and has been Principal Investigator or co-Principal Investigator on 17 sponsored projects with funding from 11 government and corporate sources.

Join or login to comment.

  • Pierre de L.

    The Video for Rebecca Passonneau's talk is now available.

    https://vimeo.com/166960455

    1 · May 17

  • James Q

    Any sides or videos of this? Would love to have been there

    2 · May 13

  • Bennett T.

    The earlier part, where the technique automatically ranked the Ebola articles by completeness, suggests an application. I'd like that in a web service, where I can lob a bunch of article links in, and get a suggested reading order. Not as noble as aiding teaching, but speaking selfishly, I'd like to use it.

    2 · May 13

  • Sami B.

    Rebecca Passonneau gave a great presentation of a text comprehension model based "wise crowd".
    It was also an interesting glimpse of what NLP looks like in academia.

    3 · May 13

  • Bennett T.

    Exciting topic, enthusiastic presentation.

    1 · May 13

Our Sponsors

  • NYI.net

    Colocation and hosting in downtown Manhattan

People in this
Meetup are also in:

Sign up

Meetup members, Log in

By clicking "Sign up" or "Sign up using Facebook", you confirm that you accept our Terms of Service & Privacy Policy