Bento: Redesigning Academic Research
Contextual Research and Product Vision
Problem
Through our interviews with academic researchers, my team found that the research process is tremendously fragmented. Different types of academic resources — articles, datasets, bookmarks, etc. — all reside in different tools and repositories. Because of this fragmentation, it’s difficult to start a new project from scratch, or to share resources with a colleague.
Solution
Our concept, Bento, allows researchers to keep all of their resources for each project in one place. Researchers can share their repositories with others using “fork” and “merge” functions, just like in a version control system (e.g. Git). A discovery tool visualizes the connections between repositories in different disciplines. A platform like Bento would help researchers learn from their colleagues in other fields while devoting more time to their actual work.
Scroll to the bottom for our concept poster, which includes some storyboards that describe Bento in action.
Process
This seven-week project involved contextual interviews, synthesis, modeling, and visioning. It culminated in a product concept, but no precise feature definitions or concrete screen designs.
How might we leverage our technology and expertise to open up new markets?
This client — a data analytics software firm — came to our team with this question. Their main product enables non-technical users to access and analyze big data. The platform is particularly adept at dealing with large volumes of unstructured data spread across many documents (for example, a large collection of police reports).
This technology has applications in fraud detection, risk analysis, regulatory compliance, and behavior analytics. Our client has customers in diverse fields, from law enforcement/intelligence to financial services, but the CEO wants to find new markets.
We identified academic research as an opportunity area. Researchers need to integrate knowledge from many types of resources, including plain-text academic papers. We utilized our network at Carnegie Mellon to learn about common practices and problems in the academic research process.
Primary research
We conducted hour-long contextual interviews with four participants:
1. A professor of machine learning who studies social networks and natural language processing
2. A highly collaborative professor of game design
3. A research librarian with experience in cybersecurity
4. A design consultant-turned-academic program director
We asked participants to walk us through their most-recent research project. We had some focused questions about how they start projects, what data types they need to work with, and how they work together with their colleagues and assistants. We encouraged the participants to show us the actual documents and systems they’ve been working with.
This turned out to be very important — one participant showed us his collection of 3,000+ web bookmarks! He estimated that he spends 10 to 20 percent of his time simply trying to organize his resources.
After each interview, we met as a team to create interview notes and Contextual Design-style models. Once all of our interviews were done, we created an affinity diagram from the notes. We also consolidated the models to get a picture of the overall academic research experience. We were able to identify some workflows, needs, and breakdowns that all of our participants shared.
Major insights
- The research process is fragmented. Most research-related platforms only serve one type of resource. Articles, datasets, bookmarks, code, etc. all “live” in separate places.
- It’s difficult to start a new project from scratch, especially when working in an interdisciplinary problem space. Different fields often have different terminology, so it’s hard to tell if an exploratory literature review has been exhaustive and representative.
- Researchers need better tools for communication, planning, and sharing. Qualitative data is especially tough to share within a team, or archive for future use.
Ideation and Visioning
With these lessons in mind, we set out to generate some ideas with a Holtzblatt-style “wall walk.” We put up all of our research work, then spent some time making connections between the models and capturing every idea that came to mind. Next, we selected the best ideas as starting points for a visioning session. Through sketching and improvisation, we explored possible preferred states.
We picked the strongest visions and translated them into concrete storyboards. We tried to hit different angles of each vision, keeping some storyboards rather conservative, but pushing others to include future technologies and new behaviors. We took these “narrative prototypes” to our interview participants to gauge their response, searching for the most-advanced-yet-acceptable opportunities.
Proposed concept: “Bento”
Our most successful concept envisioned a collaborative, version control-inspired “research guide” network. Researchers would be able to pull all of their resources into one platform, then share them with others. Our concept is named for the Japanese “bento box,” which makes it easy to carry diverse food items on the go. We created a poster to communicate this solution’s value and connect it to our client’s expertise and strategic position.
Carnegie Mellon University MHCI
Course: User Centered Research and Evaluation, Fall 2016
Instructors: Amy Ogan, Chris Connors, Jim Morris
Teammates: Gena Hong, Sara Stalla, Yining Zhao, Danny Choo
