System and method for tagging untagged events
A tagging system gathers all events (tagged and untagged) generated by remote sensors at a location or facility over time. Based on the gathered events the tagging system uses machine learning to train a model to learn the sensor layout of a facility or location and the timing between the triggering of sensors. Once trained, the model can predict the movement and location of individuals and objects throughout the facility based on a starting tagged event. Given a series of tagged and untagged events, the system can use the movement predictions of the model to tag the untagged events in the series with the identification of an individual or object that triggered the generation of the untagged event.
1. A method for tagging untagged events, comprising:
receiving a first set of tagged events as a first set of tagged data sets and a second set untagged events as a first set of untagged data sets from a set of sensors in a sensor system;
transmitting the first set of tagged data sets and the first set of untagged data sets from the sensor system to a tagging system;
generating, by a training preprocessing component, one or more training data sets from the first set of tagged data sets and the first set of untagged data sets;
training, by a causality modeling component, an event prediction model with the one or more training data sets;
receiving a second set of tagged events as a second set of tagged data sets and a second set of untagged events as a second set of untagged data sets at a tagging application preprocessing component;
generating, by the tagging application preprocessing component, one or more data set groupings from the second set of tagged data sets and the second set of untagged data sets;
applying, by the causality modeling component, the trained event prediction model to each second tagged data set in the data set grouping;
generating, by the trained event prediction model, one or more predicted events for each second tagged data set in the data set grouping;
receiving the data set grouping and the one or more predicted events for each second tagged data set at a tagging component;
comparing, by the tagging component, each predicted event to each second tagged data set received to identify matching predicted events and second tagged data sets; and
tagging one or more second untagged data set by assigning an identification tag associated with the predicted event to the second tagged data set when the second untagged data set and the predicted event are identified as matching.
2. The method of claim 1 , wherein each tagged data set is a data structure including fields for a sensor identification, an event time, and an identification tag.
3. The method of claim 2 , wherein generating one or more training data sets comprises:
removing the identification tag from each tagged data set in the first set of tagged data sets;
ordering the first set of untagged data sets and the first set of tagged data sets by the event time associated with each tagged data set and an event time associated with each untagged data set; and
grouping the ordered first set of untagged data sets and first set of tagged data sets into training data sets.
4. The method of claim 1 , wherein the event prediction model is one of an unsupervised deep learning model, a statistical model, or a supervised deep learning model.
5. The method of claim 1 , wherein each of the tagged data sets in the second set of tagged data sets include a same identification tag associated with each tagged data set.
6. The method of claim 5 , wherein generating one or more data set groupings comprises:
ordering each tagged data set from the second set of tagged data sets and each untagged data set from the second set of untagged data sets based on an event time associated with each tagged data set and each untagged data set; and
grouping a first tagged data set from the ordered data sets with all subsequent untagged data sets from the ordered data sets with associated event times between the event time associated with the first tagged data set and the event time associated with a second tagged data set from the ordered data sets to generate a first data set grouping.
7. The method of claim 6 , wherein generating one or more data set groupings further comprises repeating grouping each subsequent tagged data set from the ordered data sets with all subsequent untagged data sets from the ordered data sets with associated event times between the event time associated with the subsequent tagged data set and the event time associated with a next tagged data set from the ordered data sets to generate a next data set grouping for each subsequent tagged data set.
8. A system for tagging untagged events, comprising:
a memory comprising computer readable instructions;
a processor configured to read the computer readable instructions that when executed causes the system to:
receive a first set of tagged events as a first set of tagged data sets and a second set untagged events as a first set of untagged data sets from a set of sensors in a sensor system;
transmit the first set of tagged data sets and the first set of untagged data sets from the sensor system to a tagging system;
generate, by a training preprocessing component, one or more training data sets from the first set of tagged data sets and the first set of untagged data sets;
train, by a causality modeling component, an event prediction model with the one or more training data sets;
receive a second set of tagged events as a second set of tagged data sets and a second set of untagged events as a second set of untagged data sets at a tagging application preprocessing component;
generate, by the tagging application preprocessing component, one or more data set groupings from the second set of tagged data sets and the second set of untagged data sets;
apply, by the causality modeling component, the trained event prediction model to each second tagged data set in the data set grouping;
generate, by the trained event prediction model, one or more predicted events for each second tagged data set in the data set grouping;
receive the data set grouping and the one or more predicted events for each second tagged data set at a tagging component;
compare, by the tagging component, each predicted event to each second tagged data set received to identify matching predicted events and second tagged data sets; and
tag one or more second untagged data set by assigning an identification tag associated with the predicted event to the second tagged data set when the second untagged data set and the predicted event are identified as matching.
9. The system of claim 8 , wherein each tagged data set is a data structure including fields for a sensor identification, an event time, and an identification tag.
10. The system of claim 9 , wherein generating one or more training data sets includes computer readable instructions that further cause the system to:
remove the identification tag from each tagged data set in the first set of tagged data sets;
order the first set of untagged data sets and the first set of tagged data sets by the event time associated with each tagged data set and an event time associated with each untagged data set; and
group the ordered first set of untagged data sets and first set of tagged data sets into training data sets.
11. The system of claim 8 , wherein the event prediction model is one of an unsupervised deep learning model, a statistical model, or a supervised deep learning model.
12. The system of claim 8 , wherein each of the tagged data sets in the second set of tagged data sets include a same identification tag associated with each tagged data set.
13. The system of claim 12 , wherein generating one or more data set groupings includes computer readable instructions that further cause the system to:
order each tagged data set from the second set of tagged data sets and each untagged data set from the second set of untagged data sets based on an event time associated with each tagged data set and each untagged data set; and
group a first tagged data set from the ordered data sets with all subsequent untagged data sets from the ordered data sets with associated event times between the event time associated with the first tagged data set and the event time associated with a second tagged data set from the ordered data sets to generate a first data set grouping.
14. The system of claim 13 , wherein generating one or more data set groupings further includes computer readable instructions that further cause the system to repeat grouping each subsequent tagged data set from the ordered data sets with all subsequent untagged data sets from the ordered data sets with associated event times between the event time associated with the subsequent tagged data set and the event time associated with a next tagged data set from the ordered data sets to generate a next data set grouping for each subsequent tagged data set.
15. A non-transitory computer readable medium comprising computer readable code to tag untagged events on a system that when executed by a processor, causes the system to:
receive a first set of tagged events as a first set of tagged data sets and a second set untagged events as a first set of untagged data sets from a set of sensors in a sensor system;
transmit the first set of tagged data sets and the first set of untagged data sets from the sensor system to a tagging system;
generate, by a training preprocessing component, one or more training data sets from the first set of tagged data sets and the first set of untagged data sets;
train, by a causality modeling component, an event prediction model with the one or more training data sets;
receive a second set of tagged events as a second set of tagged data sets and a second set of untagged events as a second set of untagged data sets at a tagging application preprocessing component;
generate, by the tagging application preprocessing component, one or more data set groupings from the second set of tagged data sets and the second set of untagged data sets;
apply, by the causality modeling component, the trained event prediction model to each second tagged data set in the data set grouping;
generate, by the trained event prediction model, one or more predicted events for each second tagged data set in the data set grouping;
receive the data set grouping and the one or more predicted events for each second tagged data set at a tagging component;
compare, by the tagging component, each predicted event to each second tagged data set received to identify matching predicted events and second tagged data sets; and
tag one or more second untagged data set by assigning an identification tag associated with the predicted event to the second tagged data set when the second untagged data set and the predicted event are identified as matching.
16. The non-transitory computer readable medium of claim 15 , wherein each tagged data set is a data structure including fields for a sensor identification, an event time, and an identification tag.
17. The non-transitory computer readable medium of claim 16 , wherein generating one or more training data sets includes causing the system to:
remove the identification tag from each tagged data set in the first set of tagged data sets;
order the first set of untagged data sets and the first set of tagged data sets by the event time associated with each tagged data set and an event time associated with each untagged data set; and
group the ordered first set of untagged data sets and first set of tagged data sets into training data sets.
18. The non-transitory computer readable medium of claim 15 , wherein the event prediction model is one of an unsupervised deep learning model, a statistical model, or a supervised deep learning model.
19. The non-transitory computer readable medium of claim 15 , wherein each of the tagged data sets in the second set of tagged data sets include a same identification tag associated with each tagged data set.
20. The non-transitory computer readable medium of claim 19 , wherein generating one or more data set groupings includes causing the system to:
order each tagged data set from the second set of tagged data sets and each untagged data set from the second set of untagged data sets based on an event time associated with each tagged data set and each untagged data set; and
group a first tagged data set from the ordered data sets with all subsequent untagged data sets from the ordered data sets with associated event times between the event time associated with the first tagged data set and the event time associated with a second tagged data set from the ordered data sets to generate a first data set grouping.