Semantic template matching
AssigneeInternational Business Machines Corporation
InventorsZhong Fang Yuan, Tong Liu, Li Juan Gao, Peng HuangFu, Si Heng Sun, Yi Chen Zhong
Life of this patent
Estimating the term…
Abstract
A system and method for field extraction including determining a key position of a key in an electronic file, isolating candidate key values based on a distance from the key position, selecting a key value from the candidate key values based on an output of a trained neural network, and extracting the key and the key value from the electronic file, regardless of a key-value structure.
Claims20
A method comprising:
- determining, by a processor of a computing system, a key position of a key in an electronic file;
- isolating, by the processor, candidate key values based on a distance from the key position measured in at least three directions, wherein the isolating includes:
- searching for candidate bounding boxes that contain the candidate key values based on the distance from the key position, and
- filtering the candidate bounding boxes based on a relative position of a candidate bounding box with respect to the key position and other candidate bounding boxes;
- selecting, by the processor, a key value from the candidate key values based on an output of a trained neural network;
- and
- extracting, by the processor, the key and the key value from the electronic file.
A computing system, comprising:
- a processor;
- a memory device coupled to the processor;
- and
- a computer readable storage device coupled to the processor, wherein the storage device contains program code executable by the processor via the memory device to implement a method comprising:
- determining, by the processor, a key position of a key in an electronic file;
- isolating, by the processor, candidate key values based on a distance from the key position measured in at least three directions, wherein the isolating includes:
- searching for candidate bounding boxes that contain the candidate key values based on the distance from the key position, and
- filtering the candidate bounding boxes based on a relative position of a candidate bounding box with respect to the key position and other candidate bounding boxes;
- selecting, by the processor, a key value from the candidate key values based on an output of a trained neural network;
- and
- extracting, by the processor, the key and the key value from the electronic file.
A computer program product, comprising a computer readable hardware storage device storing a computer readable program code, the computer readable program code comprising an algorithm that when executed by a computer processor of a computing system implements a method comprising:
- determining, by the processor, a key position of a key in an electronic file;
- isolating, by the processor, candidate key values based on a distance from the key position measured in at least three directions, wherein the isolating includes:
- searching for candidate bounding boxes that contain the candidate key values based on the distance from the key position, and
- filtering the candidate bounding boxes based on a relative position of a candidate bounding box with respect to the key position and other candidate bounding boxes;
- selecting, by the processor, a key value from the candidate key values based on an output of a trained neural network;
- and
- extracting, by the processor, the key and the key value from the electronic file.
Dependencies are inferred from the claim text.