Text prediction integration

A computing device comprising a processor and memory, the memory storing executable instructions that, when executed by the processor, cause the computing device to perform operations comprising:

• instantiating a user interface comprising:
• a keyboard comprising a character entry display;
• an input text field for displaying inputted text;
• and
• a predicted items pane for displaying predicted messages contacts, conversations or documents the user may wish to interact with, based on text input by a user;
• wherein the user interface is operable to input characters via a gesture across the user interface, wherein the gesture is a continuous touch gesture and the predicted items pane displays one or more items based on the continuous gesture, even when there is no text committed to the text input field;
• and
• upon receiving an input gesture relating to a sequence of two or more terms, instantiating a text prediction to perform operations comprising:
• generate a plurality of word predictions for a first term of the sequence of two or more terms the word prediction comprising a first word comprising a subset of the selected characters and a second word generated from a text prediction engine;
• generate at least one word prediction for a second term of the sequence of two or more terms, wherein each word prediction for the second term is based on the plurality of word predictions for the first term and the plurality of selected characters;
• assign each word prediction an associated probability value determined by the text input and a contextual analysis;
• generate a ranked distribution of predicted sequences of two or more terms using probability values of each of the plurality of word predictions for the first term and the at least one word prediction for the second term to determine a cumulative probability of each predicted sequence;
• filter, search, and order, by the computing device, a plurality of predicted items associated with content of an application by at least:
• using the ranked distribution of predicted sequences of two or more terms to determine a subset of most likely items;
• generating a relevancy score for each item in the subset based on the ranked distribution and a content vector generated from words associated with each item;
• and
• ordering items in the subset based on the relevancy score;
• and
• outputting the subset of most likely items on the predicted items pane.

A method for searching a plurality of items, the method comprising:

• receiving, at a keyboard associated with a user interface on a computing device comprising a processor and memory, user text input comprising a plurality of selected characters indicative of a sequence of two or more terms, wherein the user interface is operable to input characters via a touch location or continuous gesture;
• generating, by the computing device, a plurality of word predictions for a first term of the sequence of two or more terms the word prediction comprising a first word comprising a subset of the selected characters and a second word generated from a text prediction engine;
• generating at least one word prediction for a second term of the sequence of two or more terms wherein each word prediction for the second term is based on the plurality of word predictions for the first term and the plurality of selected characters;
• assigning each word prediction an associated probability value based on the text input and a contextual analysis;
• generating, by the computing device, a ranked distribution of predicted sequences of two or more terms using probability values of each of the plurality of word predictions for the first term and the at least one word prediction for the second term to determine a cumulative probability of each predicted sequence;
• filtering, ordering, and searching, by the computing device, a plurality of items associated with content of an application by at least:
• using the ranked distribution of predicted sequence of two or more terms to determine a subset of most likely items;
• generating a relevancy score for each item in the subset based on the ranked distribution and a content vector generated from words associated with each item;
• and
• ordering items in the subset based on the relevancy score;
• and
• outputting, on a user interface of the computing device, the subset of most likely items to a user.