<< Back to Portfolio << Previous Next >>
-----------------------------------------------------------------------------------------------------------------------------------------------------------------------
Hindi Language Text-entry Solution for Handheld Devices
Aim: To make Hindi text-entry in handheld devices more usable within the existing technological constraints
Project Brief:
The increasing mobile penetration in India and particularly in the semi-urban and rural parts has resulted in a large user base for mobile interface in regional languages; Hindi being one of the most widely spoken amongst them. Text message input which is one of the most frequently used features of the interface is one area where it is particularly important to incorporate the usage of native language to enable easy communication. The major contributions of the project includes-
- Catering to the needs of "Digitally under-represented" segment of the users who can contribute to a very large consumer base of mobile phones/handheld devices, but are held back due to language constraints.
- Highlighting the major drawbacks of the existing keypad based Hindi text-entry solutions in mobile phones.
- Conducting a benchmark testing with one of the existing solutions, namely Multi keystroke method, to find the average rate of input. Using Predictive model to calculate the ideal or maximum possible rate of input to compare the rate of input.
- Finding the gap and suggesting alternate solution in the form of handwriting recognition. Proposing a template to balance the recognition and accuracy trade-off while maintaining reasonable speed of input and ease of use, for Hindi text-entry in handheld devices.
Project Accolade:
One International Journal Publication in IFIP 13.6, Springer Books, and one publication in the International conference proceedings of HWID2009 on the findings and proposed innovative solution.
Collaborators: Diya Gangopadhyay
Guide: Pradeep Yammiyavar
Nature of the Project: Semester Project
Duration: 4 months
Methodology adopted:
- Study and analysis of existing Hindi input solutions. There are three essentially distinct schemes of mapping the Hindi alphabet onto the mobile keypad. A Fitt's Law based predictive model was used to compare the text entry rate of these schemes. The input rate was found to be lower than desired and due to the nature of the alphabet it seemed difficult to improve it significantly.
- Usability testing of one existing solution, viz. multi-key stroke method and its comparison with free flow of handwriting. The rationale for choosing this solution particularly being its widespread use as text entry solution in mobile phones.
- Comparison of the observed rate of text entry with the ideal condition rate to observe the gap. Analyzing test results to find out the reasons for the observed gap.
- A concept was designed for a handwriting recognition based interface for a mobile phone with a stylus. Though, it was found to be remarkably more usable than keypad entry, the primary obstacle was the technological limitation with respect to handwriting recognition. However, we got to know from literature study that uniformity of letterforms makes recognition considerably more accurate.
Salient features of the proposed solution -
1) The solution provided templates for the user to write every letter in which ensured that the handwritten text was consistent in terms of size and proportion, thus making recognition easier for the system. Usability tests proved that this kind of a template did not make the process of text entry significantly harder or slower.
2) Another key feature of this solution was predictive input which was combined with handwriting based text entry to speed up the text entry. Based on the frequency of certain words in text entry, the system suggested the users the rest of the word based on the part of it which the user had already written.
Deliverables:
The main deliverable included an entensive report stating the observations from literature study, user surveys and testings of the paper prototype.
-----------------------------------------------------------------------------------------------------------------------------------------------------------------------
<< Back to Portfolio << Previous Next >>