Auto-correction techniques implemented in software keyboards attempt to mitigate user errors in text entry by analyzing user input and automatically generating corrections. This approach can result in inaccurate corrections (auto-corruptions), e.g., when users enter words that are out of vocabulary, uncommon words, abbreviations, etc. When such phrases are sent, e.g., as messages, recipient users may not understand the sender’s intent or make sense of the received text.
This disclosure describes techniques to decode received text that has errors produced by auto-correction. The techniques include reconstructing possible touch point sequences that led to the phrase, determining alternative phrases that correspond to the determined sequences, and presenting the alternative phrases on the recipient device. The techniques also enable the sender to quickly send post-corrections.
Creative Commons License
This work is licensed under a Creative Commons Attribution 4.0 License.
Deselaers, Thomas and Carbune, Victor, "Decoding text with errors due to auto-correction", Technical Disclosure Commons, (September 22, 2017)