Method and system for text understanding in an ontology driven platform
Embodiments of methods and systems for informatics systems are disclosed. Such informatics systems may utilize a unifying format to represent text to facilitate linking between data from the text and one or more ontologies, and the commensurate ability to mine such data.
1. A system, comprising:
an informatics system coupled to one or more data sources, the informatics system comprising a hardware processor and a non-transitory computer readable memory comprising instructions for:
receiving unstructured clinical text as input from a data source, the unstructured clinical text in a natural language;
creating a parse graph from the unstructured clinical text by:
segmenting the unstructured clinical text into a set of evidence spaces based on one or more identifiers within the natural language, each evidence space comprising a distinct portion of the natural language of the clinical text;
parsing each evidence space individually based on one or more permutations of tokens within the corresponding text of the evidence space to determine relationships between the tokens of the evidence space; and
constructing a non-ontological graph for each of the set of evidence spaces based upon the parsing of the corresponding evidence space, the graph for each evidence space representing one or more concepts and relations between those concepts in the corresponding portion of the unstructured clinical text corresponding to that evidence space, wherein the sequenced graphs for each of the set of evidence spaces form the parse graph;
obtaining a graph representation of a domain ontology, wherein the domain ontology comprises a set of concepts and a set of relationships;
mapping the parse graph to the graph representation of the domain ontology and a semantic ontology to create a unified graph comprising the graph representation of the input, the graph of the domain ontology, and a graph representation of the semantic ontology by matching one or more concepts in the parse graph to one or more corresponding concepts in the domain ontology using a mapping algorithm, and creating relationships between any concepts in the parse graph that matches a corresponding concept in the domain ontology over a corresponding threshold;
constructing a query based on at least one concept or at least one relationship of the domain ontology; and
searching the unified graph based on the query to obtain data of the input associated with the at least one concept or the at least one relationship.
2. The system of claim 1 , wherein the input is parsed based on a graph representation of a syntax ontology and the syntax ontology comprises the one or more identifiers.
3. The system of claim 1 , wherein the instructions are further for chunking each of the set of evidence spaces into one or more chunks, each chunk comprising a permutation of tokens of the evidence space within a distance of one another in the evidence space.
4. The system of claim 3 , wherein the distance is five plus or minus two.
5. The system of claim 3 , wherein creating the parse graph further comprises ordering the set of evidence spaces.
6. The system of claim 5 , wherein mapping the parse graph to the graph representation of the domain ontology comprises matching a set of concepts represented by the parse graph to one or more of the set of concepts of the domain ontology.
7. A method, comprising:
receiving unstructured clinical text as input from a data source at an informatics system comprising a hardware processor and a non-transitory computer readable memory, the unstructured clinical text in a natural language;
creating a parse graph from the unstructured clinical text by:
segmenting the unstructured clinical text into a set of evidence spaces based on one or more identifiers within the natural language, each evidence space comprising a distinct portion of the natural language of the clinical text;
parsing each evidence space individually based on one or more permutations of tokens within the corresponding text of the evidence space to determine relationships between the tokens of the evidence space; and
constructing a non-ontological graph for each of the set of evidence spaces based upon the parsing of the corresponding evidence space, the graph for each evidence space representing one or more concepts and relations between those concepts in the corresponding portion of the unstructured clinical text corresponding to that evidence space, wherein the sequenced graphs for each of the set of evidence spaces form the parse graph;
obtaining a graph representation of a domain ontology, wherein the domain ontology comprises a set of concepts and a set of relationships;
mapping the parse graph to the graph representation of the domain ontology and a semantic ontology to create a unified graph comprising the graph representation of the input, the graph of the domain ontology, and a graph representation of the semantic ontology by matching one or more concepts in the parse graph to one or more corresponding concepts in the domain ontology using a mapping algorithm, and creating relationships between any concepts in the parse graph that matches a corresponding concept in the domain ontology over a corresponding threshold;
constructing a query based on at least one concept or at least one relationship of the domain ontology; and
searching the unified graph based on the query to obtain data of the input associated with the at least one concept or the at least one relationship.
8. The method of claim 7 , wherein the input is parsed based on a graph representation of a syntax ontology and the syntax ontology comprises the one or more identifiers.
9. The method of claim 7 , further comprising chunking each of the set of evidence spaces into one or more chunks, each chunk comprising a permutation of tokens of the evidence space within a distance of one another in the evidence space.
10. The method of claim 9 , wherein the distance is five plus or minus two.
11. The method of claim 9 , wherein creating the parse graph further comprises ordering the set of evidence spaces.
12. The method of claim 11 , wherein mapping the parse graph to the graph representation of the domain ontology comprises matching a set of concepts represented by the parse graph to one or more of the set of concepts of the domain ontology.
13. A non-transitory computer readable medium comprising instructions for:
receiving unstructured clinical text as input from a data source, the unstructured clinical text in a natural language;
creating a parse graph from the unstructured clinical text by:
segmenting the unstructured clinical text into a set of evidence spaces based on one or more identifiers within the natural language, each evidence space comprising a distinct portion of the natural language of the clinical text;
parsing each evidence space individually based on one or more permutations of tokens within the corresponding text of the evidence space to determine relationships between the tokens of the evidence space; and
constructing a non-ontological graph for each of the set of evidence spaces based upon the parsing of the corresponding evidence space, the graph for each evidence space representing one or more concepts and relations between those concepts in the corresponding portion of the unstructured clinical text corresponding to that evidence space, wherein the sequenced graphs for each of the set of evidence spaces form the parse graph;
obtaining a graph representation of a domain ontology, wherein the domain ontology comprises a set of concepts and a set of relationships;
mapping the parse graph to the graph representation of the domain ontology and a semantic ontology to create a unified graph comprising the graph representation of the input, the graph of the domain ontology, and a graph representation of the semantic ontology by matching one or more concepts in the parse graph to one or more corresponding concepts in the domain ontology using a mapping algorithm, and creating relationships between any concepts in the parse graph that matches a corresponding concept in the domain ontology over a corresponding threshold;
constructing a query based on at least one concept or at least one relationship of the domain ontology; and
searching the unified graph based on the query to obtain data of the input associated with the at least one concept or the at least one relationship.
14. The computer readable medium of claim 13 , wherein the input is parsed based on a graph representation of a syntax ontology and the syntax ontology comprises the one or more identifiers.
15. The computer readable medium of claim 13 , wherein the instructions are further for chunking each of the set of evidence spaces into one or more chunks, each chunk comprising a permutation of tokens of the evidence space within a distance of one another in the evidence space.
16. The computer readable medium of claim 15 , wherein the distance is five plus or minus two.
17. The computer readable medium of claim 15 , wherein creating the parse graph further comprises ordering the set of evidence spaces.
18. The computer readable medium of claim 17 , wherein mapping the parse graph to the graph representation of the domain ontology comprises matching a set of concepts represented by the parse graph to one or more of the set of concepts of the domain ontology.