Some non-research projects I have been involved in:
- Duke computer science office map
- Interactive event map
- Web site design
- Group publication, management, and comment system
- Automatic generation of drug names
The computer science department has a database backend which maintains up-to-date information about students and professors, such as their office location, email addresses, phone numbers, etc. Hanging off of our main webpage is one front end to this, but since I maintain the office lottery, I felt that a more graphical display would allow one to more easily find all the members of an office.
This is essentially just a different view of the same model, so whenever the database updates, this page will as well. It is written in object-oriented php (imagine that!) and extracts all the information from the database in a single SQL query. Other features include: linking to rooms with schedules, intelligent cropping and resizing of oddly shaped images, and the ability to add and remove people from offices without changing the database.
I helped design a simple google maps mashup of the locations of people coming to a Professor's party. This allows one to see where from around the country all other party-goers are coming from. Although the design requirements of this small web app were simple, this could be extended to include different colors for different responses, allowing people to leave messages attached to their names, etc. This would be a great alternative to a paper RSVP for a wedding.
I helped design my advisor's website in a way that would be easily extensible and interface well with all of our group's projects. In particular, the code to manage user downloads of our research group's software was designed to be secure and stable as possible. Since these pages have been up, several department class web pages have adopted the same layout and design.
Although still in the draft stages, when completed this project will allow users to identify, tag, comment on, and search for publications. This would be particularly useful for a group of users with overlapping interests (such as an academic research group) to share knowledge about papers they have read. The final version will include: access control lists, automatic creation of bibliography files from selected publications, and hopefully automatic text extraction for searching the body of publications.
Due to the scale and intent of this project, it will likely be designed within the Django or the Ruby on Rails framework.
This doesn't quite qualify as a project, but given the offbeat brand names of drugs these days, I thought this would be interesing. Using list of the brand names of known drugs, a Markov model of a specified order is trained and new drug names are generated according to the parameters of the learned model.