Method and system for performing entity resolution

A computing device configured to perform entity resolution, the computing device comprising:

• a processor,
• a memory storing instructions;
• and
• a communication interface coupled to each of the processor and the memory,
• wherein the processor is programmed to cooperate with the instructions to perform operations comprising:
• receiving a set of data from at least one data source, wherein the set of data comprises a plurality of attributes associated with a set of entities;
• extracting a first set of attributes and a second set of attributes from the plurality of attributes, wherein each attribute from the first set of attributes is associated with at least one attribute from the second set of attributes and corresponds to an entity;
• generating using at least one trained machine learning-based model, a first set of embeddings for the first set of attributes and a second set of embeddings for the second set of attributes, wherein the generating of the first set of embeddings and the second set of embeddings includes populating missing categorical attributes with imputed values so as to generate a complete representation, and wherein the generating of the first set of embeddings and the second set of embeddings includes using the trained machine learning-based model to convert the received set of data into tokenized numeric data, wherein the at least one trained machine learning-based model includes a first trained model that is trained to identify textual attributes and generate textual embeddings, and wherein the at least one trained machine learning-based model further includes a second trained model that is trained to identify categorical attributes and generate graph node embeddings;
• combining the generated first set of embeddings and the generated second set of embeddings to generate a combined embedding by synthesizing categorical and textual embeddings into a single representation, wherein the combined embedding represents the entity;
• storing the combined embedding in a database;
• selecting a target entity from the set of entities;
• identifying a target set of embeddings associated with the selected target entity, from the stored combined embedding;
• comparing the target set of embeddings with the stored combined embedding;
• determining, via an agent framework, a set of candidate entities based on a result of the comparing, wherein the determining of the set of candidate entities includes deploying a plurality of large language model (LLM) powered agents that capture intrinsic nuances associated with the set of candidate entities to identify the set of candidate entities from the set of entities;
• and
• clustering the set of candidate entities.

A non-transitory computer readable storage medium storing instruction for performing entity resolution, the instructions comprising executable code which when executed by a processor, causes the processor to perform operations comprising:

• receive a set of data from at least one data source, wherein the set of data comprises a plurality of attributes associated with a set of entities;
• extract a first set of attributes and a second set of attributes from the plurality of attributes, wherein each attribute from the first set of attributes is associated with at least one attribute from the second set of attributes and corresponds to an entity;
• generate using at least one trained machine learning-based model, a first set of embeddings for the first set of attributes and a second set of embeddings for the second set of attributes, wherein the generating of the first set of embeddings and the second set of embeddings includes populating missing categorical attributes with imputed values so as to generate a complete representation, and wherein the generating of the first set of embeddings and the second set of embeddings includes using the trained machine learning-based model to convert the received set of data into tokenized numeric data, wherein the at least one trained machine learning-based model includes a first trained model that is trained to identify textual attributes and generate textual embeddings, and wherein the at least one trained machine learning-based model further includes a second trained model that is trained to identify categorical attributes and generate graph node embeddings;
• combine the generated first set of embeddings and the generated second set of embeddings to generate a combined embedding by synthesizing categorical and textual embeddings into a single representation, wherein the combined embedding represents the entity;
• store the combined embedding in a database;
• select a target entity from the set of entities;
• identify a target set of embeddings associated with the selected target entity, from the stored combined embedding;
• compare the target set of embeddings with the stored combined embedding;
• determine, via an agent framework, a set of candidate entities based on a result of the comparing, wherein the determining of the set of candidate entities includes deploying a plurality of large language model (LLM) powered agents that capture intrinsic nuances associated with the set of candidate entities to identify the set of candidate entities from the set of entities;
• and
• cluster the set of candidate entities.