Thursday, March 31, 2011

Paper Reading #18: An Adaptive Calendar Assistant Using Pattern Mining for User Preference Modelling

Comment 1: http://chiblog.sjmorrow.com/2011/03/paper-reading-18-adaptive-calendar.html
Comment 2: http://csce436-nabors.blogspot.com/2011/03/reading-18-adaptive-calendar-assistant.html

An Adaptive Calendar Assistant Using Pattern Mining for User Preference Modelling
Alfred Krzywicki, Wayne Wobcke and Anna Wong
Intelligent User Interfaces

This paper is about a new user interface for calendars that allows users to not only see their events but also helps intelligently organize them. The calendar application has a few features that other calendars do not, one being, suggestion of attributes for appointments such as date and time as well as location and other factors. This can help the user to have a more detailed glance view when they mouse over an event or look at a short version when clicking on it. It also has a feature named SmartCal that tracks patterns in user appointments and better helps to give suggestions on how to organize events and when an upcoming event might happen. The algorithm is both intelligent and minimally invasive and allows the user to ignore suggestions as well as change them over time. The design also helps to point out conflicting and overlapping conflicts and allows the user to either device a solution of their own or gives suggestions of how it could be handled. The authors even give the basic code and talk a lot more about the design of SmartCal and how it generates solutions to various problems that the user encounters while using the application. They perofrmed an experiment of users trying the application over a four week period and then collected results of how often the user used the new features and how many times they listened to the suggestions made as well as the frequency they used them to resolve conflicts. Users seemed relatively happy and commonly used the suggestions but not to the extent that the authors thought they might. They did get to see their appointment suggestion feature in action which had results ranging from 55% accuracy to 100% accuracy based on the user. The tests seemed to be rather successful and they want to improve the robustness and accuracy of their algorithm in the future.



I think this application looks really cool and the idea of the algorith that can predict when my appointments is going to be is really cool. However I am starting to notice a pattern of the various article I am reading and I am getting a lot of these kinds of applications recently. It seems like a lot of my articles are "applications that are built to serve a very particular set of people" and for the most part are things I don't REALLY see myself using. This makes it hard for me to really say anything overly helpful about the application, despite this there is some things that I will say about SmartCal. First, I do like the suggestion algorithm and think it is a good idea. If I was able to generate solutions to my various calendar problems and overlaps and have them pointed out to me before they become a problem. I didn't like that it will generate suggestions on how to make my events better or the strange organization system that the authors wanted where they have three kinds of events divided into categories. Again in all I think this is going to be a very good aplpication for people in the office place who have lots of appointments and need to make sense of them. I will not be in this stage for a while and I dont see me using this application any time soon so it makes it hard for me to judge it. I am curious to see what the authors do with it next and if they are able to improve the prediction algorithm at all.

1 comment:

  1. I found it interesting that a researcher would spend so much time attempting to improve calendaring software. It seems like a linear view of appointments would be intuitive enough for most people.

    ReplyDelete