Presage 0.8.2 is now available for download.

Presage 0.8.2 comes with improved gprompter and pypresagemate demo applications. gprompter new features include autopunctuation implementation, support for keyboard accelerators to access menu items, bug fixes, and updates to the text editing component. pypresagemate now supports standard command line parameters.

Presage 0.8.2 delivers a number of fixes and improvements to the core predictive engine, such as a fix for a defect in the learning code and the prediction with multimaps. The internal core predictor hierarchy has been refactored and the XML parsing code subsystem updated. Compilations issues on Solaris and various compilation warning have also been fixed.

Presage 0.8.2 also includes other bug fixes and enhancements. Please see the ChangeLog for more details.

Presage 0.8.1 is now available for download.

Presage 0.8.1 sports significant performance improvements in its smoothed n-gram predictor. Runtime execution was sped up by approximately a factor of 5 by tuning some expensive SQL queries to the embedded SQLite database.

Presage 0.8.1 comes with refactored configuration and profile handling subsystems. Configuration is read from system-level, installation-level, user-level XML profiles and from an optional user-specified profile. Changes to configuration variables made at runtime through the config() API can now be persisted to file by calling the new save_profile() API method. Presage components are also immediately notified of configuration changes and adjust accordingly.

Presage 0.8.1 also includes other bug fixes and enhancements. Compilation problems on recent stricter compilers, compilation warnings, and minor memory leaks were fixed in this release. Please see the ChangeLog for more details.

Presage 0.8 is now available for download.

Presage 0.8 includes two new predictive applications, gprompter and pypresagemate.

Gprompter is a cross-platform predictive text editor. Gprompter displays predictions in a contextual pop-up box as each letter is typed. Predictions can be easily selected and inserted in the document.

Pypresagemate is a universal predictive text companion. Pypresagemate works alongside any AT-SPI aware application. The Assistive Technology Service Provider Interface (AT-SPI) is a toolkit-neutral way of providing accessibility facilities in applications. Pypresagemate works in the background by tracking what keystrokes are typed and displaying predictions in its window. When a prediction is selected, text is sent to the active application.

Presage 0.8 provides a new callback-aware programming interface to make it easier to develop interactive presage applications. Presage applications no longer need to track user interaction by explicitly updating the context. Instead, a callback object decouples the user application from the chosen text buffer (which might be a simple string, a graphical text widget, a file stream, etc.) and frees the caller from having to explicitly notify of any updates to the context. with presage and this allows presage to retrieve contextual information from the application whenever needed.

Presage 0.8 also includes other bug fixes and enhancements. Please see the ChangeLog for more details.

Today marks the launch of the new and improved look of the presage project website.

Both the website theme appearance and the website core were updated to the latest drupal and zen theme releases, and then modified to taste with the most sophisticated web design tools such as emacs and vim.

This website redesign anticipates new features soon coming to presage. Stay tuned!

Presage 0.7.3 is now available for download.

Presage 0.7.3 includes the new predictive ARPA plugin. The ARPA plugin enables the use of statistical language modelling data in the ARPA N-gram format. In the ARPA format each N-gram is stored with its discounted log probability and its Katz backoff weight. Probabilities are estimated by applying Katz backoff smoothing to the maximum likelihood estimates based on n-gram counts data.

Presage 0.7.3 also provides (in a separate tarball) an extensive language model generated from the British National Corpus (BNC) containing 20.001 unigrams, 517.537 bigrams, and 1.648.226 trigrams. This language model was constructed by computing the smoothed Katz backoff trigram model using the CMU-Cambridge Statistical Language Modeling toolkit.

Presage 0.7.3 also includes other bug fixes and enhancements. Please see the ChangeLog for more details.

Presage 0.7.2 is now available for download.

Presage 0.7.2 adds support for predictive plugins filters and a predict method returning an ordered multimap of pairs. The filters feature enables lookahead prediction.

Presage 0.7.2 incorporates a number of bug fixes, such as handling of utf-8 encoded text, initialization bug in presage demo program, build problems fixes in gpresagemate. Prompter comes with user interface improvements, such as text size controls.

Presage 0.7.2 also includes other bug fixes and enhancements. Please see the ChangeLog for more details.