Six Degrees of Francis Bacon
I worked with English professors to design, develop, and launch Six Degrees of Francis Bacon (SDFB). SDFB is an interactive visualization that allows researchers to explore people’s relationships from the 16th and 17th centuries and contribute their knowledge. I had a major role in making this project a success.
Student Capstone Project
Aug. 2013 – Nov. 2014
(4 months, part-time)
Lead Developer, UX Designer
Jun. 2014 – Jul. 2015
(14 months, part-time)
Six Degrees of Francis Bacon (SDFB) is directed by Carnegie Mellon University.
Professor Christopher Warren (Director), Jessica Otis (Project Manager), Professor Raja Sooriamurthi (Advisor)
What We Made
In this example video, the researcher uses the Six Degrees of Francis Bacon (SDFB) visualization to search for a notable figure, review relationships, and contribute details. SDFB is also live. Try it out for yourself.
Three Things to Know About SDFB
1. SDFB’s interactive visualization enables researchers to discover all possible 1st and 2nd degree relationships for a historical figure.
Problem: Too many biographies to manually review
SDFB’s primary data source, the Oxford Dictionary of National Biography, has over 50,000 biographies of historical figures. For example, even if a researcher set the narrow goal of identifying all people that Francis Bacon associated with, she could easily miss a minor connection because it’s impossible for her to cross-check all biographies for mentions of him. To find 2nd degree connections, she would have to perform a cross-check for each of Francis Bacon’s connections. Impossible!
Solution: Programmatically parse biographies to generate the interactive social network visualization
Our statistician programmatically parsed the references that people had to each other in their biographies. He identified 200,000 relationships and estimated the strength of the relationship by tracking the the number of references to each other.
Impact: New discoveries of relationships
“You can see all of the people they knew, thus giving you new ways to consider communities, factions, influences and sources”
– Project Director Christopher Warren, associate professor of English at Carnegie Mellon University
2. SDFB facilitates scholarly collaboration
Problem: A lack of informal dialogue between researchers
There is a lack of informal dialogue between researchers studying the same notable figures and relationships, and that is a missed opportunity for researchers to share knowledge and receive critical feedback before publication.
Solution: User Contributions
Researchers can share what they know by suggesting new historical figures, relationships, or groups. They can also add more details by clarifying the type of relationship that two people had, connecting people to groups, or adding annotations. This facilitates a dialogue by enabling researchers to critique details provided by others.
I also designed a special curator role to help shape the community. There’s more about that in the Design Process Overview section.
Impact: Knowledge sharing, less bias, website stickiness
- As of March 2016, there are 1,582 new relationship type assignments, 523 new group membership assignments, 58 new annotations on people, and 12 new annotations on groups.
- Annotations help the small admin team cope with potential data errors and expand the limited perspective of the original dataset.
- The community improves website stickiness by encouraging the feeling of ownership.
3. SDFB uses a community-based editorial oversight
Problem: User contributions can lead to spam or low-quality content
There’s only 3 admin, and they need help, especially as SDFB scales up and there are more user contributions to monitor.
Solution: Approvals and edits by Admin and Curators
I designed a crowdsourcing system that ensures a high quality of contributions through approvals and edits.
In addition to the Standard User and Admin roles, I proposed a Curator role, loosely based on Wikipedia’s editorial oversight process. When a Standard User makes a large number of approved contributions, they will automatically be offered a position as a Curator. A user that is promoted to be a Curator can assist Admin with the approval and edits of user contributions. I worked closely with the professors to determine what each role should be authorized to do.
I also developed an audit trail, which is important in maintaining the integrity of the initial contributions.
Impact: High-quality content from a variety of perspectives and a strong feeling of ownership
- Scalability: Curators can assist admin in monitoring new content, especially as more user contributions come in as SDFB scales up.
- Taking Ownership: All users have the ability to contribute content, but the Curator role goes a step beyond that by making the user responsible for Six Degrees of Francis Bacon. The feeling of ownership can keep users active on the site.
- Recruiting New Users: Admin can use the promise of immediate Curator status to recruit notable historians and researchers to participate in Six Degrees of Francis Bacon.
- Limiting Bias: Each Curator views content through their own perspective.
Design Process Overview
My Contributions to the Design of SDFB
I was the first designer on this project. I worked with English professors to shape the vision for Six Degrees of Francis Bacon. Namely, I identified the need for the ability of researchers to contribute their knowledge. Then, I designed the community-based editorial system that allowed admin and curators to ensure quality contributions through approvals and edits.
Co-designing with Experts
I worked directly with English Professor Christopher Warren throughout the project and later with Post Doc Jessica Otis. The ability to reach out to the experts was great for understanding the needs of researchers. For example, I learned that exact dates were not often available when it came to relationships, so together we designed a date type tag with options “in”, “before”, “after”, and “circa”. They were also very helpful in connecting our team with their peers for additional feedback.
Working Prototypes and Testing with Students
Throughout the project, SDFB was used by students studying the Early Modern Social Network at courses taught at Carnegie Mellon University. The students’ feedback was especially helping in identifying usability concerns. The professors teaching the courses pushed us to design useful features, such as the ability to filter the social network.
Original Static Visualization
The original visualization was a GraphML format created by the iWakari team. This was the first time that a subset of the data was parsed and interpreted as a network with nodes (circles) representing people and edges (lines) showing the relationships. Even as a subset, you can see that there was a lot of data to work with. We ended up keeping this basic network model.
The Original Model with People and No Relationships
The Original Model with Both People and Relationships
For our senior capstone project as Information Systems students, my team worked with Professor Warren and iWakari to prototype a user interface for the first interactive visualization. I proposed the crowdsourcing component and presented it to them with a Balsamiq prototype (see below).
Findings based on Expert Feedback
- There needs to be different confidence estimates for each relationship type.
- The user should be able to add an annotation to explain the reasoning in addition to a bibliography.
- The bibliography should remain separate to encourage users to cite their work.
- Relationships are directional. For example, Person A is a father to Person B and not the other way around. There needs to be a way to represent the directions.
My role: Lead UX Designer
Team Members: Adetunji Olojede, Amiti Uttarwar, Miko Bautista, Leonard Sokol, and Raja Sooriamurthi (advisor)
First Working Prototype and Beta Test
Based on my designs, we created a working prototype using GraphML, HTML, and Ruby on Rails. It allowed researchers to navigate using the visualization and add annotations to existing relationships. This prototype was limited to only having a subset of data (about 20%) and did not have user accounts so users could not edit annotations. I was the Lead UX designer for the side panel in which researchers could read notes and contribute. I was also the lead developer for the crowdsourcing features and Admin pages that allowed them to browse through the data.
Findings based on beta tests with students in a college course
We ran a beta test with a class of 20 English students. They tested the interface by adding over 120 user annotations.
- The lack of a database meant it was very slow to query the data.
- Annotations became too long to read in the side panel.
My role: Lead UX Designer, Lead Backend Developer
Team Members: Adetunji Olojede, Amiti Uttarwar, Miko Bautista, Leonard Sokol, and Raja Sooriamurthi (advisor)
We learned from the previous prototype to keep scalability in mind, so for this iteration,we focused on laying a foundation that would accomplish that. I completely redesigned the database to handle big data, including the entire dataset from the Oxford Dictionary of National Biography (over 50,000 people and 200,000 relationships) and growth from user contributions. We also added user accounts which enable the user contributions model that ensures a high-standard of contributions with approvals and edits.
There are still many concerns with the interface that the team is continually addressing. But, SDFB now has a solid foundation.
Changes based on two beta tests with students in a college course and the public
- Due to the large amount of data being processed, the visualization timed out. We added a limit on the number of nodes being processed. The user can filter the query if they want.
- I made several usability improvements such as by improving the autocomplete search suggestions to search by all permutations of a person’s name (e.g. Ambrose Crowley, Sir Ambrose, Sir Ambrose Crowley, Sir Crowley, Ambrose)
- I added form validations to improve the quality of the user contributions.
- I developed an Export to CSV function so that the Admin can perform analysis on new data added from user contributions.
My Role: I was hired as the lead developer.
Team: Jessica Otis (project manager), Ally Sorge, Jeremy Lee, Zaria Howard, Ivy Chung, Sama Kanbour, Angela Qui, Chanamon Ratanalert
My contributions to the development of SDFB
As the lead developer for the current application, I created the entire Ruby on Rails and SQL database backend that made the data import, visualization, user contributions, and admin functions possible.
- Designed and implemented a scalable SQL database that includes the original dataset of over 50,000 people and 200,000 relationships
- Created secure user accounts with authorizations based on one of the 3 user roles: Standard User, Curator, or Admin
- Wrote efficient queries and passed large amounts of data from the database to the visualization.
I designed a SQL database that interfaces with the Ruby on Rails app.
- The database is denormalized to handle big data.
- All entities can be approved or rejected and made inactive by Admin or Curators.