How does the prediction in The Grid 2 work?

The Grid 2 uses a prediction system specially created by Smartbox for The Grid 2. Here are a few details about how it works intelligently to provide you with the best suggestions as you type. 

Learning as you type 

The prediction dictionary learns as you type. Each time you complete a word (by typing a space or another character that separates words) the prediction dictionary is updated. 

Frequency of use 

If you use a word often, the prediction dictionary will promote that word. This means that the word will be suggested sooner (i.e. you will not need to type so many letters to display the word). 

Contexts 

The prediction system also learns the context of the words you write. This allows it to provide suggestions for the next word after you complete a word. It also looks at the last few words you wrote to help make better suggestions about the word you are typing. 

Spelling 

Words are spell-checked before being added to the dictionary, and predictions are spell-checked before being suggested. If you want to use a word that is not in the spelling dictionary (e.g. some unusual names of people and places) then you can add these to the dictionary. 

Large 

The dictionary can hold almost 25,000 words! If you approach this number of words in the dictionary, it will drop the words that you have not used for a long time to make space for new words. 

Well trained 

The Grid 2 is supplied with a well-trained prediction dictionary with around 7,000 common words stored in it. This means that you can start using the prediction straight away. As soon as you use it, it will start learn about the words you use. You can also start with a blank dictionary by removing all the words currently in the dictionary. 

Sounds-like prediction 

In addition to providing word completion, the system also provides “sounds-like” suggestions, using phonetic matching to identify words similar to the letters you have typed. 

Language independence 

The prediction system is language independent. The system is purely statistical and does not use information about grammar or other tricks to suggest words. This means that it can be used in any language.