Comment 1
Comment 2
Reference Information:
Title: Designing Adaptive Feedback for Improving Data Entry Accuracy
Authors: Kuang Chen, Joseph M. Hellerstein, Tapan S. Parikh
Presentation Venue:UIST’10, October 3–6, 2010, New York, New York, USA
Summary:
Keeping track of data today is very important and the quality of the data is very important. For example in a clinic the data needs to be accurate taken from patients. USHER provides a theoretical, data-driven foundation for improving data quality during entry. Based on prior data, USHER learns a probabilistic model of the dependencies between form questions and values.
The best form designs include pre-determined constraints to reject or warn the user when they enter unlikely or illegal values. Some designs even insert cross-validation questions to double-check the accuracy of some questions. This double-check is something like a birthday entry field along with age.
The model USHER uses a given a subset of answers for a form and tries to accurately predict values for the unanswered questions. USHER also creates the ordering of questions on the form. This can be viewed like the game twenty-questions. You can narrow the answer down greatly depending on what order you ask the questions. USHER uses a static and a dynamic algorithm to select ordering.
 |
| USHER model |
To make filling out forms quicker widgets with either radio buttons, drop down menus, or auto-complete text fields were used based on the number of possible answers in the answer domain.
To test the technology and methods a study was done with professional data entry clerks working with
real patient data from six clinics in rural Uganda. From the study the percent of errors increases as the domain size increased, which was highest for the drop down box and auto-complete form.
 |
| This is the normal process clerks go through to fill out forms, USHER is trying to shorten the this process |
A suggestion for improvement includes the double-entry method which a different study showed reduced errors but increased the time to fill out a form. The USHER is supposed to be the compromise between these two things.
Discussion:
I actually think the study done in this paper will be useful. Because data quality is very important in fields such as the engineering and medical fields it makes sense to find better and more efficient for clerks to record it. It's very interesting how the USHER model tries to predict answers based on previous answers. This is an interesting HCI subject because it deals with data quality which is important. It also effects people that may not have any relevant computer experience when it comes to this kind of interface, the clerks. It's up to the HCI community to implement this system and improve it.
I agree that this is a great idea. I actually kind of wonder how much of an effect double data entry combined with this system would cause on errors.
ReplyDeleteUnderstanding how questions are related involves learning by collaborative filtering. I'd be interesting in learning more about how this system was designed and what algorithms were used.
ReplyDelete