Reasonable language model learning for text generation from a knowledge graph
AssigneeInternational Business Machines Corporation
InventorsThanh Lam Hoang, Dzung Tien Phan, Gabriele Picco, Lam Minh Nguyen, Vanessa Lopez Garcia
Life of this patent
Estimating the term…
Abstract
Embodiments are provided for generating a reasonable language model learning for text data in a knowledge graph in a computing system by a processor. One or more data sources and one or more triples may be analyzed from a knowledge graph. Training data having one or more candidate labels associated with one or more of the triples may be generated. One or more reasonable language models may be trained based on the training data.
Claims20
A method for providing reasonable language model learning for text data in a computing system by a processor, comprising:
- analyzing content from a plurality of data sources and a plurality of triples from a knowledge graph;
- generating training data having a plurality of candidate labels derived from the analyzed content, each candidate label associated with a corresponding triple of the plurality of triples from the knowledge graph, the association of a first candidate label with a first triple based on semantic similarity between content from of a first data source and a keyword from the first triple;
- training, by a convex continuous relation model, one or more reasonable language models based on the training data;
- and
- generating text data by the trained reasonable language models using the plurality of triples.
A system for providing reasonable language model learning in a computing environment, comprising:
- one or more computers with executable instructions that when executed cause the system to:
- analyze content from a plurality of data sources and a plurality of triples from a knowledge graph;
- generate training data having a plurality of candidate labels derived from the analyzed content, each candidate label associated with a corresponding triple of the plurality of triples from the knowledge graph, the association of a first candidate label with a first triple based on semantic similarity between content from of a first data source and a keyword from the first triple;
- train, by a convex continuous relation model, one or more reasonable language models based on the training data;
- and
- generate text data by the trained reasonable language models using the plurality of triples.
A computer program product for increasing trustworthiness of an accelerator in heterogenous systems in a computing environment, the computer program product comprising:
- one or more computer readable storage media, and program instructions collectively stored on the one or more computer readable storage media, the program instruction comprising:
- program instructions to analyze content from a plurality of data sources and a plurality of triples from a knowledge graph;
- program instructions to generate training data having a plurality of candidate labels derived from the analyzed content, each candidate label associated with a corresponding triple of the plurality of triples from the knowledge graph, the association of a first candidate label with a first triple based on semantic similarity between content from of a first data source and a keyword from the first triple;
- program instructions to train, by a convex continuous relation model, one or more reasonable language models based on the training data;
- and
- program instructions to generate text data by the trained reasonable language models using the plurality of triples.
Dependencies are inferred from the claim text.