Personalized Document Template Recommendations Using Machine Learning

When creating a new document, a user typically starts from a blank document in the default template and must manually apply the desired formatting for the type of document. The types of documents and the frequency with which each type of document is created varies across users. However, word processors do not currently include a mechanism to select a document template based on the formatting of the frequent document types created by a user. This disclosure describes techniques to present personalized recommendations for suitable templates based on the documents and templates a user constructs frequently, thus saving the user the time and effort of manual formatting. With user permission, key parameters for each document previously created by the user are used to train a machine learning model in which K-means clustering is employed to output a ranked set of recommended document templates. The recommended document templates can be presented when the user creates a new document as a dialog or a sidebar