Project Proposal: Fast Chinese Text Entry on Mobile Phone Keypads

Text messaging on mobile phones is increasingly common.  However, mapping the alphabet to twelve phone keys introduces challenging ambiguities for text entry (e.g., does 2-8-8 indicate "cat", "bat", or "act"?).  This challenge is exacerbated in Chinese by large phonetic alphabets and homophonic Chinese characters.  In response, we have developed an algorithm to generate non-alphabetically ordered keypad layouts that reduce ambiguity for Chinese text entry. Our solution uses local search to create zhu-yin (Taiwanese phonetic alphabet) keypad layouts that are high-quality according to a metric that roughly models the time cost of disambiguation to the user.

Design Problem

Determine how modified keypad layouts compare to existing commercial layouts, considering in particular training effects.  Is realistic zhu-yin text entry on mobile phones facilitated by disambiguating keypad layouts? Do non-alphabetically ordered layouts significantly slow users' learning on the keypads?  How can the existing metric be modified to more realistically reflect the time cost of entry to users (e.g., by incorporating factors that effect training speed)?

This project will stress careful usability testing, including development of a clear hypothesis and prototype and study design, execution, and analysis.

Contact Person:

Steven Wolfman, Ph.D.
Instructor, UBC CS
wolf@cs.ubc.ca

 

Here are some other notes I thought might be useful.  I've attached the thesis and BREB forms.

Resources :

A thesis describing the keypad generation algorithm and the metric at its core are available, as are several high-quality layouts generated by the algorithm.  If needed, we can also provide code implementing the algorithm in Java.  Finally, we have BREB approval for an initial user study design, the rough study design itself, and a detached USB keypad that may be helpful in designing a prototype.  (Of course, it is not necessary to follow any of the initial directions described in these last materials!)

Limited funds are available for purchasing materials for the study and for reimbursing participants.  More details available upon request.

P.S. The other attractive point for students would be that a carefully designed study, together with the problem statement and algorithm design, could make a competitive submission to a good conference.