Systems and methods for sequential model framework for next-best user state
Systems and methods of generating an interface including elements related to a next best state prediction are disclosed. A request for an interface including a user identifier is received. A next state prediction engine receives a sequence unit set including at least one sequence unit associated with the user identifier and a set of features associated with the at least one sequence unit and generates at least one next state prediction using a trained sequential prediction model. The trained sequential prediction model is configured to receive the sequence unit set and the set of features for the at least one sequence unit and output at least one predicted next state for the sequence unit set. An interface generation engine generates an interface including at least one element related to the at least one predicted next state and transmits the interface to a user device associated with the user identifier.
1. A system comprising:
a non-transitory memory;
a transceiver that receives a request for an interface including a user identifier, wherein the user identifier is stored in the non-transitory memory;
a next state prediction engine that:
receives a sequence unit set including at least one sequence unit associated with the user identifier, wherein the sequence unit is stored in the non-transitory memory;
receives a set of features associated with the at least one sequence unit in the sequence unit set; and
generates at least one predicted next state using a first trained sequential prediction model and a second trained sequential prediction model, wherein the first trained sequential prediction model receives the sequence unit set and the set of features for the at least one sequence unit and outputs at least one intermediate predicted next state for the sequence unit set, wherein the second trained sequential prediction model receives the at least one intermediate predicted next state and generates the at least one predicted next state for the sequence unit set, and wherein each of the at least one intermediate predicted next state and the at least one predicted next state each include a plurality of state elements associated therewith; and
an interface generation engine that:
generates an interface including at least one element related to the at least one predicted next state; and
transmits the interface to a user device associated with the user identifier.
2. The system of claim 1 , wherein the first trained sequential prediction model comprises one of a SASRec model or a TiSASRec model.
3. The system of claim 1 , wherein the sequence unit set comprises a set of sequence units representative of one or more prior intents associated with the user identifier.
4. The system of claim 1 , wherein the first trained sequential prediction model comprises a soft hot encoding process.
5. The system of claim 1 , wherein the first trained sequential prediction model comprises an aggregation layer including a concatenation process and an element wise sum multiply process.
6. The system of claim 1 , wherein the first trained sequential prediction model comprises a linear layer and an attention layer.
7. The system of claim 1 , wherein the set of features comprises at least one categorical feature and at least one continuous feature.
8. The system of claim 1 , wherein the transceiver receives actual next state data indicating whether a next state of the interface was the at least one predicted next state, and wherein a model training engine trains an updated sequential prediction model based, in part, on the actual next state data.
9. A computer-implemented method, comprising:
receiving, via a transceiver, a request for an interface including a user identifier, wherein the user identifier is stored in non-transitory memory;
receiving, by a state prediction engine, a sequence unit set including at least one sequence unit associated with the user identifier, wherein the sequence unit is stored in the non-transitory memory;
receiving, by the state prediction engine, a set of features associated with the at least one sequence unit in the sequence unit set;
generating, by the state prediction engine, at least one predicted next state using a first trained sequential prediction model and a second trained sequential prediction model, wherein the first trained sequential prediction model receives the sequence unit set and the set of features for the at least one sequence unit and outputs at least one intermediate predicted next state for the sequence unit set, wherein the second trained sequential prediction model receives the at least on intermediate predicted next state and generates the at least one predicted next state for the sequence unit set, and wherein each of the at least one intermediate predicted next state and the at least one predicted next state each include a plurality of state elements associated therewith;
generating, by an interface generation engine, an interface including at least one element related to the at least one predicted next state; and
transmitting, via the transceiver, the interface to a user device associated with the user identifier.
10. The method of claim 9 , wherein the first trained sequential prediction model comprises one of a SASRec model or a TiSASRec model.
11. The method of claim 9 , wherein the sequence unit set comprises a set of sequence units representative of one or more prior intents associated with the user identifier.
12. The method of claim 9 , wherein the first trained sequential prediction model comprises a soft hot encoding process.
13. The method of claim 9 , wherein the first trained sequential prediction model comprises an aggregation layer including a concatenation process and an element wise sum multiply process.
14. The method of claim 9 , wherein the first trained sequential prediction model comprises a linear layer and an attention layer.
15. The method of claim 9 , wherein the set of features comprises at least one categorical feature and at least one continuous feature.
16. The method of claim 9 , comprising:
receiving, via the transceiver, actual next state data indicating whether a next state of the interface was the at least one predicted next state; and
training, by a model training engine, an updated sequential prediction model based, in part, on the actual next state data.
17. A non-transitory computer readable medium having instructions stored thereon that, when executed by a processor, cause a device to perform operations comprising:
receiving, via a transceiver, a request for an interface including a user identifier, wherein the user identifier is stored in non-transitory memory;
receiving, by a state prediction engine, a sequence unit set including at least one sequence unit associated with the user identifier, wherein the sequence unit is stored in the non-transitory memory;
receiving, by the state prediction engine, a set of features associated with the at least one sequence unit in the sequence unit set;
generating, by the state prediction engine, at least one next state prediction using a first trained sequential prediction model and a second trained sequential prediction model, wherein the first trained sequential prediction model receives the sequence unit set and the set of features for the at least one sequence unit and outputs at least one intermediate predicted next state for the sequence unit set, wherein the second trained sequential prediction model receives the at least on intermediate predicted next state and generates at least one predicted next state for the sequence unit set, and wherein each of the at least one intermediate predicted next state and the at least one predicted next state each include a plurality of state elements associated therewith;
generating, by an interface generation engine, an interface including at least one element related to the at least one predicted next state; and
transmitting, via the transceiver, the interface to a user device associated with the user identifier.