Rime Frost is a Pinyin input method schema for the Rime input framework that provides an optimized word frequency dictionary for Chinese character and word input.
The project addresses inaccurate character and word frequency statistics in existing Rime schemas by rebuilding the dictionary from scratch. Starting from a base configuration, it removes unhealthy vocabulary, eliminates rare words with frequency of one that cannot be properly segmented, and removes non-words like "的吧" and "的了". The tool then reprocesses a high-quality corpus of over 745 million characters through segmentation and statistical analysis to normalize character and word frequencies, resulting in more accurate predictions during typing. Both full pinyin and double pinyin input methods are supported.
Developers choosing this schema should know it works as a drop-in replacement for other Rime schemas with minimal configuration changes. It suits anyone using Rime on Windows or other platforms who wants improved input accuracy without relying on cloud-based prediction. The project includes optional auxiliary code support activated by backtick, allowing character component lookup and other features like symbol input, date and time insertion, lunar calendar conversion, and calculator functionality. A standalone Windows input method called Moqi IME bundles this schema as its default option for users preferring a dedicated application.
Development activity shows consistent maintenance focused on the dictionary and word frequency data. The project includes open-source evaluation code that allows reproducible testing of input accuracy across different Rime schemas and comparison with commercial input methods. Documentation covers installation paths, configuration options, and detailed explanations of auxiliary code splitting rules through linked resources.
