Automated graph based information extraction

1. A computer-implemented method comprising:

parsing a set of sentences within a natural language text document, the parsing generating a word-level graph corresponding to a sentence in the set of sentences, the word-level graph modelling relationships between words within the sentence, a word-level node within the word-level graph corresponding to a word within the sentence;

identifying, within the word-level graph using a trained entity identification model, a set of entity candidates, an entity candidate in the set of entity candidates comprising at least one word-level node;

generating, from a set of graphs modelling relationships between portions of the set of sentences, a set of embeddings, an embedding in the set of embeddings comprising a multidimensional numerical representation of an element of a graph in the set of graphs;

extracting, from a set of pairs of embeddings in the set of embeddings using a set of deconvolution layers, a set of links between entity candidates within the set of entity candidates; and

generating, from the set of links and the set of entity candidates, an output graph modelling linkages between portions of the set of sentences within the natural language text document.

2. The computer-implemented method of claim 1 , wherein the trained entity identification model comprises a bidirectional long short-term memory and a graph convolution network trained to identify entity candidates.

3. The computer-implemented method of claim 1 , further comprising:

tagging, according to membership in the entity candidate, a plurality of word-level nodes within the word-level graph.

4. The computer-implemented method of claim 1 , wherein the set of graphs comprises a sentence-level graph modelling relationships between sentences in the set of sentences, an entity-level graph modelling relationships between a set of entity candidates including the entity candidate, and an interaction graph modelling relationships between the sentence-level graph and the entity-level graph.

5. The computer-implemented method of claim 1 , wherein the set of deconvolution layers comprises a relation edge deconvolution path and a coreference link deconvolution path, an output of a first deconvolution layer in the relation edge deconvolution path applied to an input of a second deconvolution layer in the relation edge deconvolution path and an input of a second deconvolution layer in the coreference link deconvolution path, an output of a first deconvolution layer in the coreference link deconvolution path applied to the input of the second deconvolution layer in the relation edge deconvolution path and the input of the second deconvolution layer in the coreference link deconvolution path.

6. The computer-implemented method of claim 1 , wherein a link in the set of links between entities within the set of entity candidates comprises a coreference link.

7. The computer-implemented method of claim 1 , wherein a link in the set of links between entities within the set of entity candidates comprises a relation link.

8. A computer program product for graph based information extraction, 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 instructions comprising:

program instructions to parse a set of sentences within a natural language text document, the parsing generating a word-level graph corresponding to a sentence in the set of sentences, the word-level graph modelling relationships between words within the sentence, a word-level node within the word-level graph corresponding to a word within the sentence;

program instructions to identify, within the word-level graph using a trained entity identification model, a set of entity candidates, an entity candidate in the set of entity candidates comprising at least one word-level node;

program instructions to generate, from a set of graphs modelling relationships between portions of the set of sentences, a set of embeddings, an embedding in the set of embeddings comprising a multidimensional numerical representation of an element of a graph in the set of graphs;

program instructions to extract, from a set of pairs of embeddings in the set of embeddings using a set of deconvolution layers, a set of links between entity candidates within the set of entity candidates; and

program instructions to generate, from the set of links and the set of entity candidates, an output graph modelling linkages between portions of the set of sentences within the natural language text document.

9. The computer program product of claim 8 , wherein the trained entity identification model comprises a bidirectional long short-term memory and a graph convolution network trained to identify entity candidates.

10. The computer program product of claim 8 , further comprising:

program instructions to tag, according to membership in the entity candidate, a plurality of word-level nodes within the word-level graph.

11. The computer program product of claim 8 , wherein the set of graphs comprises a sentence-level graph modelling relationships between sentences in the set of sentences, an entity-level graph modelling relationships between a set of entity candidates including the entity candidate, and an interaction graph modelling relationships between the sentence-level graph and the entity-level graph.

12. The computer program product of claim 8 , wherein the set of deconvolution layers comprises a relation edge deconvolution path and a coreference link deconvolution path, an output of a first deconvolution layer in the relation edge deconvolution path applied to an input of a second deconvolution layer in the relation edge deconvolution path and an input of a second deconvolution layer in the coreference link deconvolution path, an output of a first deconvolution layer in the coreference link deconvolution path applied to the input of the second deconvolution layer in the relation edge deconvolution path and the input of the second deconvolution layer in the coreference link deconvolution path.

13. The computer program product of claim 8 , wherein a link in the set of links between entities within the set of entity candidates comprises a coreference link.

14. The computer program product of claim 8 , wherein a link in the set of links between entities within the set of entity candidates comprises a relation link.

15. The computer program product of claim 8 , wherein the stored program instructions are stored in the at least one of the one or more storage media of a local data processing system, and wherein the stored program instructions are transferred over a network from a remote data processing system.

16. The computer program product of claim 8 , wherein the stored program instructions are stored in the at least one of the one or more storage media of a server data processing system, and wherein the stored program instructions are downloaded over a network to a remote data processing system for use in a computer readable storage device associated with the remote data processing system.

17. The computer program product of claim 8 , wherein the computer program product is provided as a service in a cloud environment.

18. A computer system comprising one or more processors, one or more computer-readable memories, and one or more computer-readable storage devices, and program instructions stored on at least one of the one or more storage devices for execution by at least one of the one or more processors via at least one of the one or more memories, the stored program instructions comprising:

program instructions to parse a set of sentences within a natural language text document, the parsing generating a word-level graph corresponding to a sentence in the set of sentences, the word-level graph modelling relationships between words within the sentence, a word-level node within the word-level graph corresponding to a word within the sentence;

program instructions to identify, within the word-level graph using a trained entity identification model, a set of entity candidates, an entity candidate in the set of entity candidates comprising at least one word-level node;

program instructions to generate, from a set of graphs modelling relationships between portions of the set of sentences, a set of embeddings, an embedding in the set of embeddings comprising a multidimensional numerical representation of an element of a graph in the set of graphs;

program instructions to extract, from a set of pairs of embeddings in the set of embeddings using a set of deconvolution layers, a set of links between entity candidates within the set of entity candidates; and

program instructions to generate, from the set of links and the set of entity candidates, an output graph modelling linkages between portions of the set of sentences within the natural language text document.

19. The computer system of claim 18 , wherein the trained entity identification model comprises a bidirectional long short-term memory and a graph convolution network trained to identify entity candidates.

20. The computer system of claim 18 , further comprising:

program instructions to tag, according to membership in the entity candidate, a plurality of word-level nodes within the word-level graph.