Abstract: The subject technology provides for temporal reasoning. A system can receive contextual information from a plurality of data sources on an electronic device. The system can identify a predetermined pattern that is indicative of a particular activity in the contextual information within a time interval. The system can determine a confidence score for the particular activity based at least in part on one or more confidence values of a corresponding activity signal associated with the time interval. The system can update a graph-based data structure by adding a representation of the particular activity as a node to the graph-based data structure when a confidence score of the particular activity exceeds a confidence threshold. The system also can provide, for display on the electronic device, a user activity interface that provides access to an indexed collection of events organized by activity type by querying the graph-based data structure.

Abstract: Aspects of the subject technology provide a relevance prediction based on a database of entities and statistics of past interactions between the entities. A relevance of a target entity in the database given a context of entities in the database may be predicted by deriving a probability set of at least one conditional probability from the database for each of the context entities, and then applying a machine learning model to the probability set to produce the relevance of the target entity.

Abstract: Methods and apparatus, including computer program products, are provided for activity tracking. In some example embodiments, there may be provided a method that includes deriving, from radio frequency based data, pseudo sensor data representative of at least an activity of a user, the pseudo sensor data derived based on at least a machine learning model configured to transform the radio frequency based data into the pseudo sensor data; and providing the pseudo sensor data. Related systems, methods, and articles of manufacture are also described.