university of kentucky :: dept. of computer science
|
|
Bayesian Advisor Project
university of kentucky :: dept. of computer science |
About the Bayesian Advisor Project... open project list ... In many academic institutions, advising duties come on top of increased pressure on faculty to improve their teaching and their department's retention of students, as well as boost research and funding thereof. All of these demands, including those inherent in the (sometimes Byzantine) requirements for students, take away from the human interaction component of advising. Thus, there are clear benefits to automating the mechanical parts of academic advising including: schedule consistency checking, progress toward degree completion, and optimization of the student's educational preferences. We hope, by automating the decision theoretic aspects of the advising process, to increase the time that advisors can spend actually listening to their advisees, imparting academic guidance and helping these students optimize their time at the university. In short, we endeavor to facilitate (augment) human advising, not to replace human advisors! The Bayesian Advisor tool aims to do just that. The advisor tool is, in practice, a "chain" of tools working together to support, ultimately, planning over a Bayes network. The advisor tool will check whether a student is on track to meet requirements in a way consistent with her preferences; it will suggest courses to further that process and will be able to compare different plans of action (sequences of courses), from disparate sources, and recommend those with the highest expectation of success for that student. |
We have completed an alpha version of POET (the online preference elicitation tool) and posted it here. Please remember that an alpha version comes very early in the development process; it's possible you'll discover some bugs and/or unimplemented features. Work on POET is on-going: we are still fixing problems and adding features. We have also posted a pre-alpha version of an XML instance document elicitator here. This tool is being designed to elicit, from domain experts, information about preferences and archetypes for use in POET. It's currently implemented as a CGI application. We appreciate feedback! Please let us know what you think! Are you interested in working on a project related to the Bayesian Advisor? We've posted a list of open projects on this site; why not take a look and see if anything grabs your attention. If something does, you can contact Dr. Goldsmith for more information ... Over the summer, members of the BAP group took a trip to Natural Bridge. Photos were taken. The results are here. It should be noted that not all members were able to attend due to dislike for summer, heat, bugs, outdoors ... The CS department will be conducting a joint AI seminar this semester. This is a merger of two previously disjoint AI seminars: one dealing with logic programming and constraint satisfaction, and the other on uncertain reasoning. The new seminar will be held on Thursdays at 8:30am in CRMS 209. Visit the seminar website for more information. Part of our group attended AAAI/IAAI 02 in Edmonton, Alberta. We presented our work on preference elicitation (POET) at the Preferences in AI and CP: Symbolic Approaches workshop. Checkout the Publications link if you want to see the paper. |
|