HCII PostDoctoral Fellows Presentation

04 Sep 2020 - Arun Balajiee

Speaker: Cynthia Bennett

Paper of the talk: Cynthia L. Bennett, Daniela K. Rosner, and Alex S. Taylor. 2020. The Care Work of Access. In Proceedings of the 2020 CHI Conference on Human Factors in Computing Systems (CHI ‘20). Association for Computing Machinery, New York, NY, USA, 1–15. DOI:https://doi.org/10.1145/3313831.3376568

Dr. Cynthia Bennett talked about the interesting applications of AI with accessibility justices. Specifically called the “Care of Work Access”, in her talk she explained the depth of work in the area of providing accessible designs for the differently abled. People are interdependednt on each other and their relationships in simple situations makes a lot of difference. The idea is to consider accessibility as a problem with continuous improvement throught mundane attunements rather than as something that can be achieved or not. Similarly, the other idea discussed is to orient analysts to work to build access for differently abled with the consideration that they also know some things and have experience with access through non-innocent authorizations. In addition to this, it is also important to consider moment-by-moment sense of how well people interact and feel together. The contribution is to incorporate these ideas into Assitive Technologies with AI. Building such a technology should set teh path to ableim of the differently abled. She then talked about a few transcripts that explain the idea of interactions between people on day-to-day mundane activities, which could pose serious challenges as problems to be solved.

We think this idea ias a important step towards inclusiveness in HCI and building technologies that are not just assistive but enable the differently abled to be self-reliant.

Speaker: Amy Pavel

Paper for the talk: http://vis.berkeley.edu/papers/vidcrit/

Dr. Amy Pavel talked about producing narratory for videos and media content to become more accessible and reduce the access barriers posed by them. Much of the talk was about adding textual commentary to videos to make them ore accessible. The interface built by Dr. Pavel provides supports to include long pauses, optimize the desciption, iterate through the algorithmic operations. The descriptions can also be categorized into inline and extended, based on the timeline of the video. When tried with novice describers, 60 % of the vide greatly overlapped with the underlying audio.

After watching her talk and reading this paper, it is a really novel field to explore. Adding accessiblity to humorous content as well as to videos is a challenge and overcoming it with the help of Text-To-Speech techniques is an interesting way to deal with the challenge. Dr. Pavel chimed with the idea of applying natural language techniques in the use of her implementation for podcasts, shorter versions of audio clips, with the idea of sacrificing a little on the content or the correctness of the text. Also incorporating a few techniques to support easy human intervention and error correction might be another approach to solving this challenge better.

Speaker: Lu Lawrence

Paper of the talk: Lawrence, L. (2019). Exploring the design process of an interdisciplinary team building a collaborative orchestration tool. LSGS Conference 2019 Contexts, Complexity, and Communities: Reflecting on and Reshaping Research on Learning. Chicago, IL: Learning Sciences Graduate Students Conference.

Dr. Lu Lawrence’s talk explored the idea of building collaborative support tools for teachers – named the “Teacher Orchestration Tool”. Her goal was to introduce the idea of “Design Based Research”. In her talk she presented the implementation of CSTEPS – collaborative support tools for engineering problemt solving. The tool is used to support students of engineering classrooms, with collaborative interaction and through teacher orchestrations. She analyzed how these affordances affect the interactions in the classrooms. From their results, her team and Dr. Lawrence foudn that the about 78 times of the clicks on the prompts from the orchestration tools were from the course instructors. There were cases of improvements in interactions for students struggling to work collaboratively. However, there were several instances where students were distracted away from using the tool.

Some of the limitations of using the tool happen to be inline with the philosophy of teaching of the instructor. It also about people who are motivated to collaborate in a classroom environment. In some cases, the expectations of collaborations and the scalibility of the technology has room for exploratory work. It is also important to consider diversity and what the instructors plan to do in the class. More can be found at this link: CoLearnLab. Dr. Lawrence also mentioned her excitment to be exploring the idea of extending this work in the implications of post COVID era of online teaching.

Speaker: Michael Mogessie

Dr. Michael Mogessie’s talk was about the fusion between best Software Engineering practices that could allow for smooth flow of research, without disruption. In this final talk of the session, Dr. Mogessie talked about the larger implications of maintaining a code base for building research tools and the best software engineering practices to be followed on this regard. It is important to train the developers to be good at the project they are working on or hiring professionals. It is important to consider separate professtionals for frontend and backend work as well as the design of the tools. Professional software developers build code that are scalable, well-documented, understand what is going in the software. All this requires spending some money from the beginning and the PI needs to consider these issues before starting their work on the research. It is also important to train students, if they are being hired to work on the project and may not know a lot on the topic. In a lot of the cases, the project maintainers depent on one developer and there is a lot of code that doesn’t make sense, for someone being onboarded later on.

Hence it is highly important to keep the software and tools required for your research to be well documented, built professionally and with usability in mind. It should also be built so that anyone coming afterwards can read and understand the code and be able to start working on the project quickly