US 11,775,763 B2Granted

Weakly supervised and explainable training of a machine-learning-based named-entity recognition (NER) mechanism

AssigneeRobert Bosch GmbH

InventorsJiacheng Li, Haibo Ding, Zhe Feng

Life of this patent

Estimating the term…

Loading patent…

Abstract

Systems and methods for weakly-supervised training a machine-learning model to perform named-entity recognition. All possible entity candidates and all possible rule candidates are automatically identified in an input data set of unlabeled text. An initial training of the machine-learning model is performed using labels assigned to entity candidates by a set of seeding rules as a first set of training data. The trained machine-learning model is then applied to the unlabeled text and a subset of rules from the rule candidates is identified that produces labels that most accurately match the labels assigned by the trained machine-learning model. The machine-learning model is then retrained using the labels assigned by the identified subset of rules as the second set of training data. This process is iteratively repeated to further refine and improve the performance of the machine-learning model for named-entity recognition.

Claims20

Dependent claims

A method of training a machine-learning model to perform named-entity recognition of unlabeled text data, the machine-learning model being configured to receive the unlabeled text data as input and to produce as output one or more text spans identified as named-entities and a category label assigned to each of the one or more text spans identified as named-entities, the method comprising:

identifying, by an electronic processor, a plurality of entity candidates from an input data set of unlabeled text, each entity candidate of the plurality of entity candidates including a text span from the input data set;
applying, by the electronic processor, a set of seeding rules to the input data set to assign a label to each entity candidate of the plurality of entity candidates;
training the machine-learning model using the labels assigned to the plurality of entity candidates using the set of seeding rules as a first set of training data;
applying the trained machine-learning model to the unlabeled text of the input data set to produce a first set of predicted data, the first set of predicted data including one or more text spans from the unlabeled text identified as named-entities and a category label assigned to each of the one or more text spans identified as named-entities;
automatically generating, by the electronic processor, a plurality of rule candidates from the input data set of unlabeled text, wherein each rule candidate of the plurality of rule candidates is generated using a set of rule templates;
applying the plurality of rule candidates to the unlabeled text of the input data set to produce a set of label candidate data;
identifying a subset of rules from the plurality of rule candidates that, when applied to the unlabeled text, assign labels to entity candidates that most accurately match the assigned labels in the first set of predicted data;
and
retraining the machine-learning model using the labels assigned by the identified subset of rules as a second set of training data.

6 more elements

A computer-based system for training a machine-learning model to perform named-entity recognition of unlabeled text data, the machine-learning model being configured to receive the unlabeled text data as input and to produce as output one or more text spans identified as named-entities and a category label assigned to each of the one or more text spans identified as named-entities, the system comprising an electronic processor configured to:

identify a plurality of entity candidates from an input data set of unlabeled text, each entity candidate of the plurality of entity candidates including a text span from the input data set;
apply a set of seeding rules to the input data set to assign a label to each entity candidate of the plurality of entity candidates;
train the machine-learning model using the labels assigned to the plurality of entity candidates using the set of seeding rules as a first set of training data;
apply the trained machine-learning model to the unlabeled text of the input data set to produce a first set of predicted data, the first set of predicted data including one or more text spans from the unlabeled text identified as named-entities and a category label assigned to each of the one or more text spans identified as named-entities;
automatically generate a plurality of rule candidates from the input data set of unlabeled text, wherein each rule candidate of the plurality of rule candidates is generated using a set of rule templates;
apply the plurality of rule candidates to the unlabeled text of the input data;
identify a subset of rules from the plurality of rule candidates that, when applied to the unlabeled text, assign labels to entity candidates that most accurately match the assigned labels in the first set of predicted data;
and
retrain the machine-learning model using the labels assigned by the identified subset of rules as a second set of training data.

6 more elements

Dependencies are inferred from the claim text.

Facts

Application: 17/185,633Filed 25 Feb 2021
Published: US 2022/0269862 A125 Aug 2022
Granted: 3 Oct 2023
Term ends: Not estimatedThe source records don’t support an estimate.

Classification10 codes

G06F 40/169
Editing, e.g. inserting or deleting: annotation, e.g. comment data or footnotes
G06F 40/295
Phrasal analysis, e.g. finite state techniques or chunking: named entity recognition
G06F 40/279
Natural language analysis: recognition of textual entities
G06N 3/042
Architecture, e.g. interconnection topology: knowledge-based neural networks; Logical representations of neural networks