Systems, methods, and computer readable media for data augmentation

A system for improving database functionality using data augmentation, the system comprising:

• a network device;
• a memory comprising:
• a data augmentation model, wherein the data augmentation model comprises a machine learning model comprising a neural network, and a plurality of seed entries, wherein the neural network comprises an embedding layer and one or more hidden layers, the data augmentation model and the plurality of seed entries generated based on a first data set, the first data set being based on a first set of users on a first platform;
• a processor in communication with the network device and the memory, the processor configured to:
• receive, via the network device, a candidate data item in a second data set based on a second set of users from a second platform, the candidate data item comprising primary user data from the second set of users, wherein the first set of users and the second set of users share no common members, and wherein no association exists between the first data set and the second data set;
• generate a candidate seed corresponding to the candidate data item in the second data set based on the primary user data from the second set of users;
• determine a data feature, based on the data augmentation model, for the candidate seed;
• generate at least one matching seed by generating a matching score between the data feature of the candidate seed and each seed entry in the plurality of seed entries and applying a threshold score, wherein the matching score is generated by processing vector representations corresponding to the data feature and the seed entry using the data augmentation model, and wherein the vector representations corresponding to the data feature and the seed entry are generated by passing the candidate seed or the seed entry through the embedding layer and the one or more hidden layers of the neural network;
• augment the candidate data item in the second data set with data corresponding to the first data set by adding, to the candidate data item, additional information corresponding to the at least one matching seed in the plurality of seed entries generated based on the first data set;
• and
• store, in the memory, an association between the second data set and the first data set in a seed match database.

A method for generating a data augmentation model and a plurality of seed entries for improving database functionality, the method comprising:

• receiving a first data;
• generating a plurality of seed entries based on the first data set, the first data set comprising demographic data of a first set of users on a first platform, and each seed entry of the plurality of seed entries corresponding to at least one user feature of a user from the first set of users;
• generating an ancillary data model for predicting at least one ancillary data item for a user from the first set of users, the at least one ancillary data item comprising a gender and an ethnicity;
• sorting the first data set based on the plurality of seed entries into at least two categories, the at least two categories comprising a combination of the gender and the ethnicity;
• generating a data augmentation model based on the first data set, wherein the data augmentation model comprises a machine learning model comprising a neural network having an embedding layer and one or more hidden layers;
• assigning each seed entry in the plurality of seed entries based on the first data set to a data feature, the data feature being based on a candidate data item in a second data set based on a second set of users from a second platform, by generating a matching score between the data feature and each seed in the plurality of seed entries and applying a threshold score, wherein the matching score is generated by processing vector representations corresponding to the data feature and the seed entry using the data augmentation model, and wherein the vector representations corresponding to the data feature and the seed entry are generated by passing the candidate seed or the seed entry through the embedding layer and the one or more hidden layers of the neural network, wherein the first set of users and the second set of users share no common members, and wherein no association exists between the first data set and the second data set;
• augmenting the candidate data item in the second data set with data corresponding to the first data set by adding, to the candidate data item, additional information corresponding to at least one assigned seed entry in the plurality of seed entries generated based on the first data set;
• and
• storing an association between the second data set and the first data set in a seed match database.