System and method for enhancing neural sentence classification
AssigneeRobert Bosch GmbH
InventorsJun ARAKI, Khanh Linh HOANG
Life of this patent
Estimating the term…
Abstract
A system and method is disclosed for classifying natural language sentences by employing external knowledge to assist in constructing a knowledge base of sentences with a target meaning. The disclosed system and method provide a general sentence classification framework applicable for a knowledge-oriented domain (e.g., domain-specific knowledge). The system and method may be implemented in an intelligent automotive aftermarket assistance tool to assist with the identification of sentences describing specific problems and solutions for car repairs. In addition to the domain adaptability, the system and method is language-independent and could be applicable to any natural written language.
Claims20
A method for classifying a natural language sentence, comprising:
- receiving a textual sentence;
- identifying one or more links between a sentence representation and a synthesized document, wherein the sentence representation is generated using the textual sentence, and wherein the synthesized document is generated from a knowledge graph;
- and
- classifying the textual sentence using the one or more links.
A method for classifying a natural language sentence, comprising:
- receiving a textual sentence;
- encoding one or more relational triplets provided by a knowledge graph to produce a low-dimensional knowledge vector representation;
- encoding the textual sentence to produce a low-dimensional textual vector representation;
- identifying one or more links between the low-dimensional knowledge vector representation and the low-dimensional textual vector representation;
- and
- classifying the textual sentence using the one or more links.
A system operable to classify a natural language sentence, comprising:
- a controller operable to:
- receive a textual sentence;
- identify one or more links between a sentence representation, a knowledge entity, and a relational representation, wherein the sentence representation is generated using the textual sentence, and wherein the knowledge entity and relational representation is generated from a knowledge graph;
- and
- classify the textual sentence using the one or more links.
Dependencies are inferred from the claim text.