Network computer system using sequence invariant model to predict user actions
Embodiments provide for a computer system and method to employ a sequence invariant model to determine user intentions, based on monitoring of real-time activities of the user.
1. A method for selectively engaging online users, the method being implemented by one or more processors and comprising:
monitoring each user of a group of users during a respective online session where the user performs a sequence of M activities, to selectively engage one or more users of the group;
wherein monitoring the group of users includes:
for each user of the group, determining an intention score of the user after each activity that the user is determined to perform over at least a portion of the respective online session, beginning after the user performs a corresponding activity that is sequenced as a Kth activity in the user's sequence of M activities until a determination event is detected for the user; and
wherein determining the intention score includes determining, using a predictive model that is adapted for a sequence of K activities, the intention score of the user based on information associated with a sequence of K preceding activities that that the user performed; and
wherein K and M are integers greater than 0, and K is less than or equal to M.
2. The method of claim 1 , wherein the determination event corresponds to at least one of (i) the intention score of the user exceeding a threshold value, or (ii) the user is determined to perform a corresponding activity that is sequenced as an Mth activity in the user's sequence of M activities.
3. The method of claim 1 , wherein the determination event corresponds to at least one of (i) the intention score of the user exceeding a threshold value, (ii) the user is determined to perform a corresponding activity that is sequenced as an Mth activity in the user's sequence of M activities, and/or (iii) the user is determined to perform a particular type of activity.
4. The method of claim 1 , wherein the determination event corresponds to at least one of (i) the intention score of the user exceeding a threshold value, (ii) the user is determined to perform a corresponding activity that is sequenced as an Mth activity in the user's sequence of M activities, and/or (iii) the user is determined to perform a particular sequence of activities.
5. The method of claim 1 , wherein the intention score of the user is based at least in part on contextual information associated with the sequence of K preceding activities.
6. The method of claim 1 , further comprising selectively engaging individual users of the group based at least in part on the intention score of the user.
7. The method of claim 2 , wherein selectively engaging individual users of the group includes engaging users for which the intention score indicates a likelihood that the user will perform a conversion event.
8. The method of claim 3 , wherein selectively engaging individual users of the group includes excluding users for engagement from which the intention score indicates a likelihood that the user will not perform a conversion event.
9. The method of claim 1 , further comprising:
monitoring individual users of the group to determine whether an activity that the user performs after the respective determination event is consistent with the intention score determined for the user; and
tuning a training of the predictive model based on the activity the user is determined to have performed.
10. A computer system comprising:
one or more processors;
a memory to store a set of instructions;
wherein the one or more processors execute the set of instructions to perform operations that include:
monitoring each user of a group of users during a respective online session where the user performs a sequence of M activities, to selectively engage users of the group;
wherein monitoring the group of users includes:
for each user of the group, determining an intention score of the user after each activity that the user is determined to perform over at least a portion of the respective online session, beginning after the user performs a corresponding activity that is sequenced as a Kth activity in the user's sequence of M activities until a determination event is detected for the user; and
wherein determining the intention score includes determining, using a predictive model that is adapted for a sequence of K activities, the intention score of the user based on information associated with a sequence of K preceding activities that that the user performed; and
wherein K and M are integers greater than 0, and K is less than M.
11. The computer system of claim 10 , wherein the determination event corresponds to at least one of (i) the intention score of the user exceeding a threshold value, or (ii) the user is determined to perform a corresponding activity that is sequenced as an Mth activity in the user's sequence of M activities.
12. The computer system of claim 10 , wherein the determination event corresponds to at least one of (i) the intention score of the user exceeding a threshold value, (ii) the user is determined to perform a corresponding activity that is sequenced as an Mth activity in the user's sequence of M activities, and/or (iii) the user is determined to perform a particular type of activity.
13. The computer system of claim 10 , wherein the determination event corresponds to at least one of (i) the intention score of the user exceeding a threshold value, (ii) the user is determined to perform a corresponding activity that is sequenced as an Mth activity in the user's sequence of M activities, and/or (iii) the user is determined to perform a particular sequence of activities.
14. The computer system of claim 10 , wherein the intention score of the user is based at least in part on contextual information associated with the sequence of K preceding activities.
15. The computer system of claim 10 , wherein the operations further comprise:
selectively engaging individual users of the group based at least in part on the intention score of the user.
16. The computer system of claim 11 , wherein selectively engaging individual users of the group includes engaging users for which the intention score indicates a likelihood that the user will perform a conversion event.
17. The computer system of claim 12 , wherein selectively engaging individual users of the group includes excluding users for engagement from which the intention score indicates a likelihood that the user will not perform a conversion event.
18. The computer system of claim 10 , wherein the operations further comprise:
monitoring individual users of the group to determine whether an activity that the user performs after the respective determination event is consistent with the intention score determined for the user; and
tuning a training of the model based on the activity the user is determined to have performed.
19. A non-transitory computer-readable medium that stores instructions, that when executed by one or more processors of a computer system, causes the computer system to perform operations that include:
monitoring each user of a group of users during a respective online session where the user performs a sequence of M activities, to selectively engage users of the group;
wherein monitoring the group of users includes:
for each user of the group, determining an intention score of the user after each activity that the user is determined to perform over at least a portion of the respective online session, beginning after the user performs a corresponding activity that is sequenced as a Kth activity in the user's sequence of M activities until a determination event is detected for the user; and
wherein determining the intention score includes determining, using a predictive model that is adapted for a sequence of K activities, the intention score of the user based on information associated with a sequence of K preceding activities that that the user performed; and
wherein K and M are integers greater than 0, and K is less than M.
20. The non-transitory computer-readable medium of claim 19 , wherein the determination event corresponds to at least one of (i) the intention score of the user exceeding a threshold value, (ii) the user is determined to perform a corresponding activity that is sequenced as an Mth activity in the user's sequence of M activities, (iii) the user is determined to perform a particular type of activity, and/or (iv) the user is determined to perform a particular sequence of activities.