Information and interaction management in a central database system
A central database system allows entities to easily manage human resources functions. An entity requests that the central database system executes an employer function on its behalf, and the central database system determines a probability of the entity defaulting before the entity can finalize the employer function with the central database system. The central database system can train and apply a machine-learned model to dynamically determine the default probability for the entity. Based on the default probability, the central database system determines a risk tolerance associated with the employer function and determines whether to process or challenge the employer function based on the risk tolerance and the default probability.
1. A method to streamline database interactions comprising:
accessing, by a central database system, a training data set comprising historical interactions by historical entities with the central database system and characteristics of the historical entities;
training, by the central database system, a machine-learned model based on the accessed training data set, the machine-learned model configured to determine a rate of requiring a manual upload of files to the central database system from entities associated with interactions with the central database system;
receiving, by a central database system and via a first interface generated by the central database system, requests from target entities for target interactions with the central database system;
accessing, by the central database system, information describing characteristics of the target entities;
determining, by the central database system, a rate of requiring the manual upload of files to the central database system from the target entities by applying the machine-learned model to the accessed information and to characteristics of the target entities, the machine-learned model configured to output the rate of requiring the manual upload of files to the central database system; and
in response to determining that the rate of requiring the manual upload of files to the central database system is too high, retraining, by the central database system, the machine-learned model by adjusting weights and parameters of the machine-learned model such that an updated rate of requiring the manual upload of files to the central database system associated with the retrained machine-learned model is lower than the rate of requiring the manual upload of files to the central database system.
2. The method of claim 1 , wherein characteristics of the historical entities includes one or more of: past failures associated with each historical entity, an age of an account associated with each historical entity, and a number of years in operation of each historical entity.
3. The method of claim 1 , wherein the machine-learned model is further trained based on additional information associated with the historical entities received from one or more third-party systems.
4. The method of claim 1 , wherein the files are manually uploaded via a second interface generated by the central database system.
5. The method of claim 4 , wherein the second interface generated by the central database system comprises one or more of: an upload interface enabling the target entity to upload files and a third-party API interface enabling the target entity to log in to a third-party system that provides the files.
6. A non-transitory computer-readable storage medium containing computer program code that, when executed by a processor, causes the processor to perform steps comprising:
accessing, by a central database system, a training data set comprising historical interactions by historical entities with the central database system and characteristics of the historical entities;
training, by the central database system, a machine-learned model based on the accessed training data set, the machine-learned model configured to determine a rate of requiring a manual upload of files to the central database system from entities associated with interactions with the central database system;
receiving, by a central database system and via a first interface generated by the central database system, requests from target entities for target interactions with the central database system;
accessing, by the central database system, information describing characteristics of the target entities;
determining, by the central database system, a rate of requiring the manual upload of files to the central database system from the target entities by applying the machine-learned model to the accessed information and to characteristics of the target entities, the machine-learned model configured to output the rate of requiring the manual upload of files to the central database system; and
in response to determining that the rate of requiring the manual upload of files to the central database system is too high, retraining, by the central database system, the machine-learned model by adjusting weights and parameters of the machine-learned model such that an updated rate of requiring the manual upload of files to the central database system associated with the retrained machine-learned model is lower than the rate of requiring the manual upload of files to the central database system.
7. The non-transitory computer-readable storage medium of claim 6 , wherein characteristics of the historical entities includes one or more of: past failures associated with each historical entity, an age of an account associated with each historical entity, and a number of years in operation of each historical entity.
8. The non-transitory computer-readable storage medium of claim 6 , wherein the machine-learned model is further trained based on additional information associated with the historical entities received from one or more third-party systems.
9. The non-transitory computer-readable storage medium of claim 6 , wherein the files are manually uploaded via a second interface generated by the central database system.
10. A system comprising:
a hardware processor; and
a non-transitory computer-readable medium containing instructions that, when executed by the hardware processor, cause the hardware processor to:
accessing, by a central database system, a training data set comprising historical interactions by historical entities with the central database system and characteristics of the historical entities;
training, by the central database system, a machine-learned model based on the accessed training data set, the machine-learned model configured to determine a rate of requiring a manual upload of files to the central database system from entities associated with interactions with the central database system;
receiving, by a central database system and via a first interface generated by the central database system, requests from target entities for target interactions with the central database system;
accessing, by the central database system, information describing characteristics of the target entities;
determining, by the central database system, a rate of requiring the manual upload of files to the central database system from the target entities by applying the machine-learned model to the accessed information and to characteristics of the target entities, the machine-learned model configured to output the rate of requiring the manual upload of files to the central database system; and
in response to determining that the rate of requiring the manual upload of files to the central database system is too high, retraining, by the central database system, the machine-learned model by adjusting weights and parameters of the machine-learned model such that an updated rate of requiring the manual upload of files to the central database system associated with the retrained machine-learned model is lower than the rate of requiring the manual upload of files to the central database system.
11. The system of claim 10 , wherein characteristics of the historical entities includes one or more of: past failures associated with each historical entity, an age of an account associated with each historical entity, and a number of years in operation of each historical entity.
12. The system of claim 10 , wherein the machine-learned model is further trained based on additional information associated with the historical entities received from one or more third-party systems.
13. The system of claim 10 , wherein the files are manually uploaded via a second interface generated by the central database system.