C4.5: programs for machine learning
C4.5: programs for machine learning
IGTree: Using Trees for Compression and Classification in Lazy LearningAlgorithms
Artificial Intelligence Review - Special issue on lazy learning
The art of computer programming, volume 3: (2nd ed.) sorting and searching
The art of computer programming, volume 3: (2nd ed.) sorting and searching
LetterWise: prefix-based disambiguation for mobile text input
Proceedings of the 14th annual ACM symposium on User interface software and technology
Language modeling for soft keyboards
Proceedings of the 7th international conference on Intelligent user interfaces
A commonsense approach to predictive text entry
CHI '04 Extended Abstracts on Human Factors in Computing Systems
Text prediction systems: a survey
Universal Access in the Information Society
Word-based predictive text entry using adaptive language models
Natural Language Engineering
The effect of domain and text type on text prediction quality
EACL '12 Proceedings of the 13th Conference of the European Chapter of the Association for Computational Linguistics
Applied Computational Intelligence and Soft Computing - Special issue on Awareness Science and Engineering
| Hi-index | 0.00 |
Word completion is a basic technology for reducing the effort involved in text entry on mobile devices and in augmentative communication devices, where efficiency and ease of use are needed, but where a low memory footprint is also required. Standard solutions compress a lexicon into a suffix tree with a small memory footprint and high retrieval speed. Keystroke savings, a measurable correlate of text entry effort gain, typically improve when the algorithm would also take into account the previous word; however, this comes at the cost of a large footprint. We develop two word completion algorithms that encode the previous word in the input. The first algorithm utilizes a character buffer that includes a fixed number of recent keystrokes, including those belonging to previous words. The second algorithm includes the complete previous word as an extra input feature. In simulation studies, the first algorithm yields marked improvements in keystroke savings, but has a large memory footprint. The second algorithm can be tuned by frequency thresholding to have a small footprint, and be less than one order of magnitude slower than the baseline system, while its keystroke savings improve over the baseline.