Situation forecast mechanisms for internet of things integration platform

Abstract

A method of consolidating Internet of Things (IoT) devices connected via an IoT network is disclosed. An IoT integration platform implemented by a computer system can collect data from one or more of IoT devices, IoT solution specific server systems, third-party server systems, general-purpose user computing devices, or any combination thereof. The IoT integration platform can label the data based on entity-specific context. The entity-specific context can correspond to a user account, a device, a location, or any combination thereof. The IoT integration platform can generate an entity-specific profile based on the labeled data. The IoT integration platform can generate, based on the entity-specific profile, a situation forecast associated with a target entity and with a timeframe yet to occur.

1 - 23 . (canceled) 
     
     
         24 . A computer-implemented method comprising:
 receiving, by an Internet of things (IoT) integration platform, multiple data streams from multiple data sources;   training multiple machine learning models based on a history of the multiple data streams, wherein the multiple machine learning models comprise at least (1) a subpopulation model configured to capture common behavioral patterns and (2) a device-specific model configured to characterize device usage patterns of the multiple data sources;   forecasting a contextual situation associated with a target entity based on a weighted combination of likelihood functions of the multiple machine learning models; and   generating, by the IoT integration platform, a recommended action for the target entity based on the forecasted contextual situation.   
     
     
         25 . The method of  claim 24 , further comprising:
 deriving one or more context indicators associated with the target entity based on the multiple data streams, wherein the one or more context indicators are configured to change adaptively over time.   
     
     
         26 . The method of  claim 25 , wherein a context indicator is computed as a function of multiple types of activity data from the multiple data streams. 
     
     
         27 . The method of  claim 25 , wherein at least one of the one or more context indicators is a function of another context indicator. 
     
     
         28 . The method of  claim 27 , further comprising:
 updating the one or more context indicators iteratively to capture interdependent impact of the one or more context indicators on each other.   
     
     
         29 . The method of  claim 24 , wherein forecasting the contextual situation comprises:
 generating a set of possible contextual situations based on the multiple data streams, wherein a possible contextual situation is associated with one or more attributes of the target entity; and   selecting the contextual situation from the set of possible contextual situations according to the weighted combination of likelihood functions.   
     
     
         30 . The method of  claim 24 , further comprising:
 classifying the multiple data streams into one or more events using the multiple machine learning models.   
     
     
         31 . The method of  claim 24 , further comprising:
 determining one or more trackable entities associated with the target entity using an entity graph; and   forecasting the contextual situation associated with the target entity using information of the one or more trackable entities.   
     
     
         32 . The method of  claim 24 , wherein at least one of the multiple machine learning models is specific to a user, a device, a location or place, a group of users, a group of devices, a group of locations or places, or any combination thereof. 
     
     
         33 . The method of  claim 24 , wherein the multiple data streams include at least one of a user reported activity, a third party application observed activity, or an IoT integration platform observed or inferred activity. 
     
     
         34 . An Internet of things (IoT) integration system, comprising a processor that is configured to:
 receive multiple data streams from multiple data sources;   train multiple machine learning models based on a history of the multiple data streams, wherein the multiple machine learning models comprise at least (1) a subpopulation model configured to capture common behavioral patterns and (2) a device-specific model configured to characterize device usage patterns of the multiple data sources;   forecast a contextual situation associated with a target entity based on a weighted combination of likelihood functions of the multiple machine learning models; and   generate a recommended action for the target entity based on the forecasted contextual situation.   
     
     
         35 . The system of  claim 34 , wherein the processor is configured to:
 derive one or more context indicators associated with the target entity based on the multiple data streams, wherein the one or more context indicators are configured to change adaptively over time.   
     
     
         36 . The system of  claim 35 , wherein the processor is configured to compute a context indicator as a function of multiple types of activity data from the multiple data streams. 
     
     
         37 . The system of  claim 35 , wherein at least one of the one or more context indicators is a function of another context indicator. 
     
     
         38 . The system of  claim 37 , wherein the processor is further configured to update the one or more context indicators iteratively to capture interdependent impact of the one or more context indicators on each other. 
     
     
         39 . The system of  claim 34 , wherein the processor is configured to forecast the contextual situation by:
 generating a set of possible contextual situations based on the multiple data streams, wherein a possible contextual situation is associated with one or more attributes of the target entity; and   selecting the contextual situation from the set of possible contextual situations according to the weighted combination of likelihood functions.   
     
     
         40 . The system of  claim 24 , wherein the processor is configured to classify the multiple data streams into one or more events using the multiple machine learning models. 
     
     
         41 . The system of  claim 34 , wherein the processor is configured to:
 determine one or more trackable entities associated with the target entity using an entity graph; and   forecast the contextual situation associated with the target entity using information of the one or more trackable entities.   
     
     
         42 . The system of  claim 34 , wherein at least one of the multiple machine learning models is specific to a user, a device, a location or place, a group of users, a group of devices, a group of locations or places, or any combination thereof. 
     
     
         43 . The system of  claim 34 , wherein the multiple data streams include at least one of a user reported activity, a third party application observed activity, or an IoT integration platform observed or inferred activity.