[dcchairs2016] UbiComp/ISWC 2016 Doctoral School notification - #114

[dcchairs2016 at ubicomp.org]

Dear Mohammad Rafayet Ali,

Please find enclosed the reviews for your submission for the Ubicomp/ISWC 2016 Doctoral School.

114: "Automated Conversational Skills Training with Real-time Feedback"

Despite not being able to accept your submission at this year's Ubicomp/ISWC Doctoral School, committee members provided guidance and feedback on your submitted paper. We highly encourage you to follow the valuable advices that the committee member entered in their reviews towards improving on your doctoral work.

Thank you for submitting to the UbiComp 2014 Doctoral School.


Max Mühlhäuser
Nadir Weibel
Rene Mayrhofer

UbiComp 2016 Doctoral School Chairs


------------------------ Submission 114, Review 1 ------------------------

Title: Automated Conversational Skills Training with Real-time Feedback


Confidence

   3  (Very confident - I am knowledgeable in the area)

Contribution to UbiComp

   This submission proposes to develop a system that automatically gives
   real-time feedback on non-verbal cues (eye contact, volume, body
   movement, and smile) during conversations. The contribution is more in
   term of testing the different ML approaches and enable the natural and
   real-time feedback, less about Ubicomp.

Overall Rating

   2  (Probably reject: I would argue for rejecting this paper.)

The Review

   This is interesting and potentially impactful research. The initial
   studies that have been achieved are interesting and yielded promising
   data for the continuation of the work.

   However there are two weak point in this submission. First of all the
   future work, in particular in terms of specific research plan and
   objectives to reach and activities to accomplish are not clear. Beyond
   the vague concept of improving the ML approach and make it work in a
   naturalistic setting we don't know much about how the author want to
   achieve this. 

   Additionally, it is not clear what is the core contribution to Ubicomp
   here. Definitely there is an HCI component, a ML component and even a
   component that is interesting in terms of the multimodal approach, but
   the submission does not really highlight what is the Ubicomp
   contribution.

   In summary the proposed research is too broadly defined and not specific
   enough to be able to capitalize for a presence at the Ubicomp DC. The
   advice is to define better what the doctoral student want to do to
   accomplish his/her goals and then see what is the community that he/she
   needs more guidance in terms of completing the thesis.


------------------------ Submission 114, Review 2 ------------------------

Title: Automated Conversational Skills Training with Real-time Feedback


Confidence

   2  (Somewhat confident - I have passing knowledge)

Contribution to UbiComp

   DOubtful, see below....

Overall Rating

   2  (Probably reject: I would argue for rejecting this paper.)

The Review

   The author reports on work towards DETECTING (via ML methods) and
   CONVEYING conversational skills of users that interact through or with
   computers, with the aim to improve these skills during the interaction by
   means of appropriate feedback.
   One problem with the paper is that it is difficult to follow since it
   jumps back and forth a couple of times between past, present, and future
   work. The second problem concerns UbiComp pertinence: conversational
   skills improvement on conventional computer settings is hardly related to
   Ubicomp. A third set of problems concerns the content quality: in the
   “speed dating” (human-in-the-loop) experiment, members of the
   organization (“RAs”) had to determine whether the reported approach
   (developed at the organization) outperformed conventional instructions
   about conversational skills – and guess what: they had the impression
   that it does :-); in other words: this study appears to be very biased.
   As to the results reported from the first attempts towards automatic (HMM
   based) detection of conversational skills cues (smile, eye contact etc.),
   all but the “volume” detection (which can probably be directly
   measured with better results than those shown in table 1) yield false
   positives of 30% and more; this means that the system is currently hardly
   usable so that its application in a UbiComp related setting is not
   feasible yet.