Automated classification and interpretation of life science documents
A computer-implemented tool for automated classification and interpretation of documents, such as life science documents supporting clinical trials, is configured to perform a combination of raw text, document construct, and image analyses to enhance classification accuracy by enabling a more comprehensive machine-based understanding of document content. The combination of analyses provides context for classification by leveraging relative spatial relationships among text and image elements, identifying characteristics and formatting of elements, and extracting additional metadata from the documents as compared to conventional automated classification tools.
1. A computer-implemented method for classifying and interpreting one or more documents, the method comprising:
receiving a digitized representation of the one or more documents that include information on life sciences, text, and images;
analyzing the text for each of the one or more documents to identify one or more root words within the text;
performing a construct analysis for each of the one or more documents after the text analysis is performed to identify document elements and spatial positioning of the document elements in the one or more documents, wherein performing the construct analysis for each of the one or more documents comprises analyzing the text, analyzing the images, and analyzing context connections of each of the one or more documents;
classifying each of the one or more documents into at least one predefined class based on the analyzed text and construct analysis; and
tagging each of the one or more documents with a class tag to identify the least one predefined class and an event tag to identify one or more events within the one or more documents.
2. The computer-implemented method of claim 1 , wherein analyzing the images of each of the one or more documents comprises identifying one or more images and extracting characteristics from the images.
3. The computer-implemented method of claim 1 , wherein the image analysis is utilized along with the analyzed text and the construct analysis to classify the one or more documents into the predefined classes.
4. The computer-implemented method of claim 1 , wherein analyzing the text for each of the one or more documents comprises analyzing one or more words of the respective document.
5. The computer-implemented method of claim 1 , further comprising:
creating context-based representations of the one or more documents after the text and construct analysis to identify spatial relationships among the document elements.
6. The computer-implemented method of claim 1 , further comprising:
creating context-based representations of the one or more documents after the text and construct analysis to enable the classification of the one or more documents into the predefined classes.
7. A computer program product comprising a tangible non-transitory storage medium encoded with processor-readable instructions that, when executed by one or more processors, enable the computer program product to:
receive a digitized representation of one or more documents that include information on life sciences, text, and images;
analyze the text for each of the one or more documents to identify one or more root words within the text;
perform a construct analysis for each of the one or more documents after the text analysis is performed to identify document elements and spatial positioning of the document elements in the one or more documents, wherein performing the construct analysis for each of the one or more documents comprises analyzing the text, analyzing the images, and analyzing context connections of each of the one or more documents;
classify each of the one or more documents into at least one predefined class based on the analyzed text and construct analysis; and
tag each of the one or more documents with a class tag to identify the least one predefined class and an event tag to identify one or more events within the one or more documents.
8. The computer program product of claim 7 , wherein the one or more documents include one or more triggers.
9. The computer program product of claim 7 , wherein the event tag identifies significant events within the one or more documents.
10. The computer program product of claim 7 , wherein analyzing the images comprises identifying one or more images and extracting characteristics from the one or more images.
11. The computer program product of claim 7 , wherein the construct analysis for each of the one or more documents includes analyzing element characteristics including text font, text size, and format of the respective document.
12. The computer program product of claim 7 , wherein analyzing the text for each of the one or more documents comprises analyzing one or more text sequences within the document elements of the respective document.
13. The computer program product of claim 7 , wherein context-based representations are created after the text analysis and construct analysis to enable the one or more documents to be classified into the pre-defined classes.
14. A computer system connected to a network, the system comprising:
one or more processors configured to:
receive a digitized representation of one or more documents that include information on life sciences, text, and images;
analyze the text for each of the one or more documents to identify one or more root words within the text;
perform a construct analysis for each of the one or more documents after the text analysis is performed to identify document elements and spatial positioning of the document elements in the one or more documents, wherein performing the construct analysis for each of the one or more documents comprises analyzing the text, analyzing the images, and analyzing context connections of each of the one or more documents;
classify each of the one or more documents into at least one predefined class based on the analyzed text and construct analysis; and
tag each of the one or more documents with a class tag to identify the least one predefined class and an event tag to identify one or more events within the one or more documents.
15. The computer system of claim 1 , wherein the event tag identifies one or more triggers within the one or more documents.
16. The computer system of claim 14 , wherein document deconstruction is performed on the one or more documents to accommodate noise and unwanted artifacts to enable the one or more documents to be converted into the digitized representation and be received as the digitized representation.
17. The computing system of claim 14 , wherein the construct analysis for each of the one or more documents includes analyzing context connections by identifying text information directly before and directly after at least one of the document elements in the respective document.
18. The computing system of claim 14 , wherein analyzing the text for each of the one or more documents comprises analyzing at least one key word among the identified root words in the respective document.
19. The computing system of claim 14 , wherein the construct analysis analyzes the spatial positioning of the document elements in the one or more documents that is used to generate metadata to enable the one or more documents to be classified into the predefined classes.