Predictive event searching utilizing a machine learning model trained using dynamically-generated event tags

1. An apparatus comprising:

at least one processing device comprising a processor coupled to a memory;

the at least one processing device being configured to perform steps of:

obtaining event metadata for a plurality of events published by one or more event sources to an event platform, the event metadata comprising static event tags for respective ones of the plurality of events;

generating one or more dynamic event tags having an association with one or more event types based at least in part on analysis of real-time event traffic, the real-time event traffic comprising a subset of the plurality of events published by the one or more event sources to the event platform over a designated time period;

training a machine learning model utilizing the static event tags for respective ones of the plurality of events and the association of the one or more dynamic event tags with the one or more event types;

receiving a query comprising one or more event parameters; and

providing a response to the query by utilizing the trained machine learning model to match one or more of the plurality of events with the one or more event parameters in the query.

2. The apparatus of claim 1 wherein at least one static event tag for a given one of the plurality of events is assigned by a given one of the event sources in conjunction with publishing the given event to the event platform.

3. The apparatus of claim 1 wherein the event metadata for a given one of the plurality of events further comprises a source application, an event name and an event description.

4. The apparatus of claim 1 wherein the event metadata for a given one of the plurality of events further comprises a payload schema and a payload content type.

5. The apparatus of claim 1 wherein the designated time period comprises a sliding time window, and wherein the at least one processing device is configured to repeat the steps of generating the one or more dynamic event tags and training the machine learning model responsive to expiration of the sliding time window.

6. The apparatus of claim 1 wherein generating the one or more dynamic event tags captures linguistic context of the real-time event traffic by applying natural language processing with topic modeling to text in the event metadata for the subset of the plurality of events published by the one or more event sources to the event platform over the designated time period.

7. The apparatus of claim 1 wherein generating the one or more dynamic event tags comprises, for a given one of the one or more event types, creating a token dataset comprising occurrence frequencies for tokens representing at least a subset of a plurality of words in the event metadata for one or more events in the subset of the plurality of events associated with the given event type that are published by the one or more event sources to the event platform over the designated time period.

8. The apparatus of claim 7 wherein creating the token dataset comprises:

splitting text of the event metadata for the one or more events in the subset of the plurality of events associated with the given event type into one or more sentences;

splitting the one or more sentences into the plurality of words;

lemmatizing the plurality of words;

stemming the plurality of words; and

extracting ones of the stemmed words that are nouns and verbs to form the subset of the plurality of words.

9. The apparatus of claim 7 wherein generating the one or more dynamic event tags comprises, for the given event type, performing topic modeling on the token dataset.

10. The apparatus of claim 9 wherein performing topic modeling on the token dataset comprises:

creating a bag of words model for the tokens in the token dataset;

filtering the tokens in the token dataset; and

applying a latent Dirichlet allocation algorithm to the bag of words model with the filtered tokens to identify topics for respective ones of the events in the subset of the plurality of events published by the one or more event sources to the event platform over the designated time period.

11. The apparatus of claim 10 wherein filtering the tokens in the token dataset comprises:

removing tokens that occur in less than a first threshold number of event instances in the subset of the plurality of events published by the one or more event sources to the event platform over the designated time period;

removing tokens that occur in more than a second threshold number of event instances in the subset of the plurality of events published by the one or more event sources to the event platform over the designated time period; and

keeping a third threshold number of most frequent remaining tokens.

12. The apparatus of claim 1 wherein the machine learning model comprises a convolutional neural network model.

13. The apparatus of claim 1 wherein the one or more event sources comprise one or more applications running on at least one of physical and virtual computing resources of information technology infrastructure.

14. The apparatus of claim 1 wherein the static event tags and the dynamic event tags are stored in an event register maintained by the event platform, the event register exposing one or more application programming interfaces for receiving the query and providing the response to the query.

15. A computer program product comprising a non-transitory processor-readable storage medium having stored therein program code of one or more software programs, wherein the program code when executed by at least one processing device causes the at least one processing device to perform steps of:

obtaining event metadata for a plurality of events published by one or more event sources to an event platform, the event metadata comprising static event tags for respective ones of the plurality of events;

generating one or more dynamic event tags having an association with one or more event types based at least in part on analysis of real-time event traffic, the real-time event traffic comprising a subset of the plurality of events published by the one or more event sources to the event platform over a designated time period;

training a machine learning model utilizing the static event tags for respective ones of the plurality of events and the association of the one or more dynamic event tags with the one or more event types;

receiving a query comprising one or more event parameters; and

providing a response to the query by utilizing the trained machine learning model to match one or more of the plurality of events with the one or more event parameters in the query.

16. The computer program product of claim 15 wherein generating the one or more dynamic event tags captures linguistic context of the real-time event traffic by applying natural language processing with topic modeling to text in the event metadata for the subset of the plurality of events published by the one or more event sources to the event platform over the designated time period.

17. The computer program product of claim 15 wherein the static event tags and the dynamic event tags are stored in an event register maintained by the event platform, the event register exposing one or more application programming interfaces for receiving the query and providing the response to the query.

18. A method comprising steps of:

obtaining event metadata for a plurality of events published by one or more event sources to an event platform, the event metadata comprising static event tags for respective ones of the plurality of events;

generating one or more dynamic event tags having an association with one or more event types based at least in part on analysis of real-time event traffic, the real-time event traffic comprising a subset of the plurality of events published by the one or more event sources to the event platform over a designated time period;

training a machine learning model utilizing the static event tags for respective ones of the plurality of events and the association of the one or more dynamic event tags with the one or more event types;

receiving a query comprising one or more event parameters; and

providing a response to the query by utilizing the trained machine learning model to match one or more of the plurality of events with the one or more event parameters in the query;

wherein the method is performed by at least one processing device comprising a processor coupled to a memory.

19. The method of claim 18 wherein generating the one or more dynamic event tags captures linguistic context of the real-time event traffic by applying natural language processing with topic modeling to text in the event metadata for the subset of the plurality of events published by the one or more event sources to the event platform over the designated time period.

20. The method of claim 18 wherein the static event tags and the dynamic event tags are stored in an event register maintained by the event platform, the event register exposing one or more application programming interfaces for receiving the query and providing the response to the query.