Can you remember the last time you had to search for folders within folders for a shared document? One of the greatest challenges of a collaborative project with many shared files and versions of various document types over a long period time is managing the growing list of resources in an efficient way. Bento is a concept for a product that would allow for version control and contributor management for academic researchers in the same way that Github make collaboration amongst developers a delightful experience.
7 week-long project for CMU course, User Centered Research and Evaluation | Fall 2016
Danny Choo, Sara Stalla, Palmer D'Orazio, Yining Zhao
Contextual Inquiry, Interviews, Affinity Diagramming, Sequence Flow Models, Cultural Models, Storyboarding, Speed-dating, Visioning
How might we leverage our client's technology and expertise to new domains? For this project, our client--a data analytics software firm--wanted to explore new applications of their current product. The platform is designed to deal with large volumes of unstructured data spread across many disparate data sources that could range from a news article to a CSV file. While the client's product is currently used in fraud detection and other legal contexts, they wanted to find new markets.
Our team identified academic research as an opportunity area for product expansion particularly because the client's current technology is poised to address issues around managing resources amongst multiple collaborators. With this user group in mind, we reached out to our network at Carnegie Mellon to start an exploratory research process.
We began our research by identifying our user group (academic researchers with a high level of collaboration) and conducting interviews with participants including: a professor of machine learning who manages PhD student collaborators, a librarian specializing in science and engineering inquiries who supports student researchers, and a professor of game design who leads several student research groups. These contextual interviews were a great opportunity to practice interviewing best practices such as asking about a specific example rather than generalized experiences.
We structured our interpretation sessions so they were at the most, 48 hours after each interview. Each interpretation session had a designated modeler who captured flows and other helpful visualizations on the whiteboard, a facilitator who led the interpretation session, and a notetaker who translated what we heard from interviewers into interpretation notes.
We used a structured affinity diagramming process to synthesize our interpretation notes from all of our interviewees. This bottom-up process allowed us to identify central themes as they emerged from the notes, rather than from pre-determined categories. This synthesis process took place over a few days and allowed the whole team to get immersed in the data together as we placed and moved around notes.
We created models of the process in order to map the current process and the breakdowns.
We used the method of "walking the wall" to identify the key issues, design ideas, and questions that came out from our affinity diagram. Taking these ideas into consideration, we used an improv method to rapidly generate future visions.
Storyboarding & Speed Dating
After our visioning session, we created eight storyboards illustrating concepts that ranged from conservative to provocative use of future technologies. We tested these storyboards through speed dating sessions and continued to iterate on our concepts based on feedback from participants.
Final Concept & Client Presentation
Through our research and contextual inquiry, we envision a resource curation service that supports cherry-picking and synthesizing selected resources across different resource types. This service would allow users to browse collections of interdisciplinary resource “Bundles” and select pieces that are most relevant to their research, thus curating their own focused Bundle. An accompanying visualization would display semantic connections on a map, and the proximity of a repository to each concept would represent where it fell on a spectrum of ideas. Bundle placement could be determined by tags or a text search of the bundle contents, and positioned on a corresponding area of the visualization.