Analytical precursor mining for personalized recommendation

Abstract

Systems and methods are disclosed for discovering precursors associated with a current user interaction event. One method comprises receiving a selection of a new item by a user and determining a plurality of similarities between the new item selected by the user and a plurality of historical items, the plurality of historical items being associated with prior user activity. Then a plurality of importance weights associated with the plurality of historical items are determined. Based on the plurality of similarities and the plurality of importance weights, at least one enhanced importance matrix is generated. Further, prior interactions of the user with the plurality of historical items are determined. Based on the enhanced importance matrix and the prior interactions of the user with the plurality of historical items, precursors for the new item selected by the user are identified and provided to a display.

1 - 20 . (canceled) 
     
     
         21 . A computer-implemented method of discovering precursors associated with a current user interaction event, the method comprising:
 receiving a selection of a new item by a user;   determining a plurality of similarities between the new item selected by the user and prior user interactions;   determining a plurality of importance weights associated with the prior user interactions by generating and training a predictive machine learning model;   generating at least one importance matrix based on the plurality of similarities and the plurality of importance weights;   identifying one or more precursors for the new item selected by the user based on the importance matrix and the prior user interactions; and   continuously updating the machine learning model with new importance weights based on user interaction events observed over time.   
     
     
         22 . The computer-implemented method of  claim 21 , further comprising providing, to a display, the identified precursors. 
     
     
         23 . The computer-implemented method of  claim 21 , wherein each of the plurality of similarities is determined by comparing content features of the new item with content features of each of the prior user interaction. 
     
     
         24 . The computer-implemented method of  claim 21 , wherein identifying one or more precursors for the new item selected by the user comprises:
 determining a personal history matrix of a plurality of values representing the prior user interactions; and   determining a product of the importance matrix and the personal history matrix.   
     
     
         25 . The computer-implemented method of  claim 21 , wherein each of the importance weights represents how each prior user interaction affects the current user interaction event collectively from all users, the current user interaction event being the user clicking the new item. 
     
     
         26 . The computer-implemented method of  claim 21 , wherein the prior user interactions include at least one or more of:
 positive prior interactions; and   negative prior interactions.   
     
     
         27 . The computer-implemented method of  claim 26 , wherein the negative prior interactions include at least one or more of:
 user skips; and   user dwell times that are less than a predetermined threshold.   
     
     
         28 . The computer-implemented method of  claim 21 , wherein determining the plurality of importance weights comprises:
 generating a current interaction matrix having entries representing current actions of users on a plurality of new items;   generating a history interaction matrix having entries representing prior actions of the users on a plurality of historical items;   generating a similarity matrix having entries representing similarities between the plurality of new items and the plurality of historical items;   generating a prediction matrix having entries representing probabilities of users selecting the plurality of new items; and   learning the importance weights of the plurality of historical items in a form of a weight matrix.   
     
     
         28 . The computer-implemented method of  claim 21 , wherein determining the plurality of importance weights comprises applying a gradient descent algorithm. 
     
     
         29 . A system for discovering precursors associated with a current user interaction event, the system comprising:
 one or more processors; and   a non-transitory computer readable medium storing instructions that, when executed by the one or more processors, cause the one or more processors to perform a method comprising:
 receiving a selection of a new item by a user; 
 determining a plurality of similarities between the new item selected by the user and prior user interactions; 
 determining a plurality of importance weights associated with the prior user interactions by generating and training a predictive machine learning model; 
 generating at least one importance matrix based on the plurality of similarities and the plurality of importance weights; 
 identifying one or more precursors for the new item selected by the user based on the importance matrix and the prior user interactions; and 
 continuously updating the machine learning model with new importance weights based on user interaction events observed over time. 
   
     
     
         30 . The system of  claim 29 , further comprising providing, to a display, the identified precursors. 
     
     
         31 . The system of  claim 29 , wherein each of the plurality of similarities is determined by comparing content features of the new item with content features of each of the plurality of prior user interactions. 
     
     
         32 . The system of  claim 29 , wherein identifying one or more precursors for the new item selected by the user comprises:
 determining a personal history matrix of a plurality of values representing the prior user interactions; and   determining a product of the importance matrix and the personal history matrix.   
     
     
         33 . The system of  claim 29 , wherein each of the importance weights represents how each prior user interaction affects the current user interaction event collectively from all users, the current user interaction event being the user clicking the new item. 
     
     
         34 . A non-transitory computer readable medium storing instructions that, when executed by one or more processors, cause the one or more processors to perform a method of discovering precursors associated with a current user interaction event, the method comprising:
 receiving a selection of a new item by a user;   determining a plurality of similarities between the new item selected by the user and prior user interactions;   determining a plurality of importance weights associated with the prior user interactions by generating and training a predictive machine learning model;   generating at least one importance matrix based on the plurality of similarities and the plurality of importance weights;   identifying one or more precursors for the new item selected by the user based on the importance matrix and the prior user interactions; and   continuously updating the machine learning model with new importance weights based on user interaction events observed over time.   
     
     
         35 . The non-transitory computer readable medium of  claim 34 , further comprising providing, to a display, the identified precursors. 
     
     
         36 . The non-transitory computer readable medium of  claim 34 , wherein each of the plurality of similarities is determined by comparing content features of the new item with content features of each of the prior user interaction. 
     
     
         37 . The non-transitory computer readable medium of  claim 34 , wherein determining the plurality of importance weights comprises:
 generating a current interaction matrix having entries representing current actions of users on a plurality of new items;   generating a history interaction matrix having entries representing prior actions of the users on a plurality of prior user interactions;   generating a similarity matrix having entries representing similarities between the plurality of new items and the plurality of prior user interactions;   generating a prediction matrix having entries representing probabilities of users selecting the plurality of new items; and   learning the importance weights of the plurality of historical items in a form of a weight matrix.   
     
     
         38 . The non-transitory computer readable medium of  claim 34 , wherein each of the importance weights represents how each prior user interaction affects the current user interaction event collectively from all users, the current user interaction event being the user clicking the new item. 
     
     
         39 . The non-transitory computer readable medium of  claim 34 , wherein the prior user interactions include at least one or more of:
 positive prior interactions; and   negative prior interactions.   
     
     
         40 . The non-transitory computer readable medium of  claim 34 , wherein identifying one or more precursors for the new item selected by the user comprises:
 determining a personal history matrix of a plurality of values representing the prior user interactions; and   determining a product of the importance matrix and the personal history matrix.