We are working together to design, build, and operate shared infrastructure
to support researchers who want to experiment with personalized news
systems.  The US National Science Foundation has funded us to carry out the
planning phase of this activity, and part of that involves reaching out to
the research community to find out what you want and need in such
infrastructure.  Based on what we find, we will be putting together a
research infrastructure proposal and larger team to build and operate the
infrastructure.



We describe what we’re considering below, but our request is simple.  If
you have a research project that such infrastructure would support or
enable, we’d like you to tell us about it.  We plan to summarize the
responses for NSF and also to follow up with individuals to seek more
details where needed to present them in the proposal.  Please submit your
research project ideas through this survey (one survey entry per research
project, we welcome multiple submissions):



    https://z.umn.edu/RecNews



A few words of introduction:



The need:  We believe that the research community (and particularly the
academic research community) needs the resources to conduct more
experimental research in all aspects of personalized news delivery from
recommendation algorithms to content processing and summarization, from
interface designs to social aspects of news sharing. Most researchers lack
access to systems with long-term user bases and the ability to carry out
online experiments.



The synergy:  While we have initiated this effort in response to needs in
the recommender systems community, we intend for the resulting
infrastructure to support a wide variety of research and inquiry including
natural language processing, information retrieval, information
visualization, and human-computer interaction.



The vision:  Our goal is to design a news recommender (both back-end and
front-end) designed from the ground up to support multiple parallel
experiments.  Researchers could bring forward experiments to be carried out
on a selection of users across its user base, and would get back detailed
data describing both short-term and long-term results.  Experiments could
range from changes in the app interface to new summarization techniques to
different rankings to select which news articles to display.  And most
important, these experiments could evaluate not just the short-term impact
in click-through or reading rate, but longer-term patterns of consumption
or even (through pop-up surveys) content awareness.



What's missing:  That all sounds good, but any real system needs to start
with priorities and the definition of a "minimum viable product" that would
support the community. The survey noted above is intended to help us answer
some key questions for creation of our infrastructure proposal:



   -

   What would you use such an infrastructure for?  (we'll be collecting
   case stories)
   -

   What features or capabilities are essential to you? (we'll also be
   polling to see how important different options are)
   -

   Do you want to help?

Given that we are seeking NSF funding, we will have to emphasize US-based
research and at least initially a US-based news service, but we plan to
make the case that the field is best served in the long-term by robust
shared international infrastructure.  We welcome researchers from anywhere,
and anyone who may be interested in helping.



Again, if this is infrastructure you’d like to use, please fill out this
survey:



https://z.umn.edu/RecNews





Robin Burke, University of Colorado

Joseph A. Konstan, University of Minnesota

Ed Malthouse, Northwestern University



--
Joseph A. Konstan
Associate Dean for Research, College of Science and Engineering
Distinguished McKnight Professor and Distinguished University Teaching
Professor
Department of Computer Science and Engineering
University of Minnesota
[log in to unmask]

    ----------------------------------------------------------------------------------------
    To unsubscribe from CHI-ANNOUNCEMENTS send an email to:
     mailto:[log in to unmask]

    To manage your SIGCHI Mailing lists or read our polices see:
     https://sigchi.org/operations/listserv/
    ----------------------------------------------------------------------------------------