Text prediction systems have become more and more popular for people with speech and motor disabilities, and for the short messaging system (SMS) generation. Personal digital assistants and phones incorporate advanced text prediction systems with the purpose of facilitating user input.

As the title implies, this paper provides a comprehensive survey of existing technology for text prediction systems. With a clear conclusion, the paper suggests the need for a uniform platform to evaluate such systems.

The authors focus on the following factors: text block size, dictionary features, prediction method, the language influence, and the usability and adaptability of the user interface. The human-interface issues, and their influence on the predictor’s performance and the usability of the system, are particularly interesting. Though prediction measurement has been investigated objectively using simulators for human behavior, a subjective analysis is also possible involving users in the tests. The lack of a standard platform for performance evaluation makes it difficult to compare existing systems.

The paper is a comprehensive and well-written survey of text prediction systems. It might serve as a starting point and initial set of guidelines for people who want to do research in this field.