Petteri Nurmi, Andreas Forsblom,
Patrik Floreen
Helsinki Institute for Information
Technology HIIT
Department of Computer Science, P.O. Box
68, FI-00014 University of Helsinki, Finland
firstname.lastname@cs.helsinki.fi
Peter Peltonen, Petri Saarikko
Helsinki Institute for Information
Technology HIIT
P.O. Box 9800, FI-02015 Helsinki
University of Technology TKK, Finland
firstname.lastname@hiit.fi
Summary
The usage of text predictions in creating shopping list was studied in this paper. They compiled an auto-generated list of 10K shopping lists to create their dictionary of suggestible words. They then implemented 8029 association rules assorted by their confidence. This alters the listing of the suggested words so that it fits the user rather than just listing them alphabetically. For instance, if the user types "cereal" and the user types the letter 'm', the program would suggest the word "milk" before listing the word "macaroni".
The user study was done using mobile devices equipped with text prediction and without(the control group). Within these groups the users were also asks to complete entering their shopping lists with either both their hands or with only one hand. The participants were given candy for being in the study.
The results of the study showed that users would input words at about 5 wpm faster than without text prediction. This result was even faster for people typing with two hands.
Discussion
The user study in this paper was hilarious! They gave the participants candy? Really? Don't most studies actually pay their participants? Also, it said that one participant was removed from the study because of their "substantially high typing error rate". I don't really understand why they decided to test the difference between one handed and two handed typing. It seemed like an arbitrary test to create further useless information.
No comments:
Post a Comment