Identifying entities absent from training data using neural networks
AssigneeSap SE
InventorsSean Saito, Auguste Byiringiro
Life of this patent
Estimating the term…
Abstract
Methods, systems, and computer-readable storage media for a machine learning (ML) model and framework for training of the ML model to enable the ML model to correctly match entities even in instances where new entities are added after the ML model has been trained. More particularly, implementations of the present disclosure are directed to a ML model provided as a neural network that is trained to provide a scalar confidence score that indicates whether two entities in a pair of entities are considered a match, even if an entity in the set of entities was not accounted for in training of the ML model.
Claims20
A computer-implemented method for determining whether entities in a pair of entities match one another using a machine learning (ML) model, the method being executed by one or more processors and comprising:
- training a ML model using training data comprising a first set of entities and a second set of entities;
- defining an entity pair comprising a first entity and a second entity, the first entity being associated with first data representing the first entity and the second entity being associated with second data representing the second entity, the second entity being absent from the training data used to train the ML model using one-shot learning;
- providing a first entity string comprising at least a portion of the first data and a second entity string comprising at least a portion of the second data;
- processing the first entity string and the second entity string through a ML model to generate a scalar confidence score by:
- providing, by an embedding layer of the ML model, a first embedding based on the first entity string and a second embedding based on the second entity string,
- generating, by a decomposable attention layer of the ML model, a vector based on the first embedding and the second embedding, and
- generating, by an activation layer of the ML model, the scalar confidence score;
- and
- determining that the first entity matches the second entity based on the scalar confidence score.
A non-transitory computer-readable storage medium coupled to one or more processors and having instructions stored thereon which, when executed by the one or more processors, cause the one or more processors to perform operations for determining whether entities in a pair of entities match one another using a machine learning (ML) model, the operations comprising:
- training a ML model using training data comprising a first set of entities and a second set of entities;
- defining an entity pair comprising a first entity and a second entity, the first entity being associated with first data representing the first entity and the second entity being associated with second data representing the second entity, the second entity being absent from the training data used to train the ML model using one-shot learning;
- providing a first entity string comprising at least a portion of the first data and a second entity string comprising at least a portion of the second data;
- processing the first entity string and the second entity string through a ML model to generate a scalar confidence score by:
- providing, by an embedding layer of the ML model, a first embedding based on the first entity string and a second embedding based on the second entity string,
- generating, by a decomposable attention layer of the ML model, a vector based on the first embedding and the second embedding, and
- generating, by an activation layer of the ML model, the scalar confidence score;
- and
- determining that the first entity matches the second entity based on the scalar confidence score.
A system, comprising:
- a computing device;
- and
- a computer-readable storage device coupled to the computing device and having instructions stored thereon which, when executed by the computing device, cause the computing device to perform operations for natural language explanations for determining whether entities in a pair of entities match one another using a machine learning (ML) model, the operations comprising:
- training a ML model using training data comprising a first set of entities and a second set of entities;
- defining an entity pair comprising a first entity and a second entity, the first entity being associated with first data representing the first entity and the second entity being associated with second data representing the second entity, the second entity being absent from the training data used to train the ML model using one-shot learning;
- providing a first entity string comprising at least a portion of the first data and a second entity string comprising at least a portion of the second data;
- processing the first entity string and the second entity string through a ML model to generate a scalar confidence score by:
- providing, by an embedding layer of the ML model, a first embedding based on the first entity string and a second embedding based on the second entity string,
- generating, by a decomposable attention layer of the ML model, a vector based on the first embedding and the second embedding, and
- generating, by an activation layer of the ML model, the scalar confidence score;
- and
- determining that the first entity matches the second entity based on the scalar confidence score.
Dependencies are inferred from the claim text.