Visual structure of documents in question answering

1. A method comprising:

constructing a knowledge base for a natural-language question-answering system,

wherein constructing said knowledge base for said natural-language question-and-answering system comprises:

ingesting a document that comprises first, second, and third visual segments, each of which comprises semantic content, and

pre-processing said document to enable said question-answering system to access said semantic content in response to a natural-language question that has been posed by a user of said natural-language question-answering system,

wherein pre-processing said document comprises:

using a visual structure of said document to extract structural information from said document and

incorporating said structural information in said knowledge base,

wherein said visual structure of said document comprises

a spatial distribution of said visual segments in said document when said document is in a form that renders said document visible,

wherein using said visual structure to extract structural information from said document comprises:

based at least in part on locations of said first and second visual segments relative to each other within said document, determining that a context relationship is to exist between said first and second visual segments,

based at least in part on locations of said first and third visual segments relative to each other within said document, determining that no context relationship is to exist between said first and third visual segments, and

establishing said context relationship between said first and second visual segments.

2. The method of claim 1 , wherein determining that a context relationship is to exist between said first and second visual segments comprises determining that said context relationship is to exist based at least in part on a distance between said first and second visual segments and wherein determining that no context relationship is to exist between said first and third visual segments comprises determining that no context relationship is to exist based at least in part on a distance between said first and third visual segments.

3. The method of claim 1 , wherein determining that a context relationship is to exist between said first and second visual segments comprises determining that said context relationship is to exist based at least in part on a using semantic information in addition to using said structural information.

4. The method of claim 1 ,

wherein determining that a context relationship is to exist between said first and second visual segments comprises determining that said context relationship is to exist based at least in part on a reference in said first visual segment to said second visual segment and

wherein determining that no context relationship is to exist between said first visual segment and said third visual segment comprises determining that no context relationship is to exist based at least in part on an absence of a reference in said first visual segment to said third visual segment.

5. The method of claim 1 ,

wherein using said visual structure to extract structural information from said document comprises classifying said visual segments into different classes based on differences in visual appearances of said visual segments,

wherein determining that a context relationship is to exist between said first and second visual segments comprising determining that said context relationship is to exist based at least in part on classes of said first and second visual segments, and

wherein determining that no context relationship is to exist between said first and third visual segments comprises determining that no context relationship is to exist between said first and second sematic segments based at least in part on classes of said first and third visual segments.

6. The method of claim 1 ,

wherein determining that a context relationship is to exist between said first and second visual segments comprises determining that said context relationship is to exist based at least in part on a using semantic information in addition to using said structural information and

wherein determining that no context relationship is to exist between said first and third visual segments comprises determining that no context relationship is to exist based at least in part on a using semantic information in addition to using said structural information.

7. The method of claim 1 , wherein determining that a context relationship is to exist between said first and second visual segments comprises determining that said context relationship exists based at least in part on having determined that said first visual segment is a figure and that said second visual segment is a caption for said figure.

8. The method of claim 1 , wherein determining that a context relationship is to exist between said first and second visual segments comprises determining that said context relationship exists based at least in part on having determined that said first and second visual segments have a common location.

9. The method of claim 1 , wherein determining that a context relationship is to exist between said first and second visual segments comprises determining that said context relationship exists based on having determined that a first visual segment is a figure and that a second visual segment is text that is superimposed on said figure.

10. The method of claim 1 , wherein said ingested document comprises instructions for causing said document to be rendered visible and wherein said method further comprises determining said visual structure based on said instructions.

11. The method of claim 1 , wherein said ingested document is in a portable document format and wherein said method further comprises determining said visual structure of said document based on rendering instructions that are expressed in said portable document format.

12. The method of claim 10 , wherein said ingested document is an HTML file and wherein said method further comprises determining said visual structure of said document based on tags in said HTML file.

13. The method of claim 1 , further comprising determining that said context relationship exists based at least in part on having determined that said first visual segment comprises text arranged in rows and columns and on having determined that said second visual segment includes text at an intersection of one of said rows and one of said columns.

14. The method of claim 1 , wherein using said visual structure to extract structural information from said ingested document comprises inspecting tags associated with said visual segments, said tags indicating that each of said visual segments belongs to a class in a set of classes, said set of classes comprising a list, a paragraph, a heading, and an image.

15. The method of claim 1 ,

wherein determining that a context relationship is to exist between said first and second visual segments and that no context relationship is to exist between said first and third visual segments comprises using both said semantic content in said first, second, and third visual segments and said structural information to determine that said context relationship is to exist between said first and second visual segments and that no context relationship is to exist between said first and third visual segments,

said structural information having been extracted based on instructions that are embedded in said document and that affect said visual structure of said document,

said document being a PDF document.

16. The method of claim 1 , further comprising:

wherein incorporating the structural information in the knowledge base comprises storing a first item of composite content from a combination of content of the first and second visual segments in a computer storage;

receiving a natural language question that has been posed by a user of said natural-language question-answering system;

processing said question using a computer-implemented question answering procedure accessing content in the computer storage, including forming an answer to the question by retrieving the first item of composite content and processing the question and composite content using a natural language processing procedure; and

presenting the answer to the user.

17. The method of claim 16 , wherein ingesting the document comprises forming a retrieval view of the first visual segment, said retrieval view including text content of the first visual segment, and encoding formatting properties of said segment determined from a visual representation of said segment, and wherein presenting the answer to the user comprises presenting a visual representation of the retrieval view.

18. A non-transitory machine-readable media comprising instructions stored thereon for execution by a data-processing system that has access to a knowledge base that has been constructed for a natural-language question-and-answering system and that comprises ingested documents, among which is an ingested document that comprises first, second, and third visual segments, each of which comprises content, wherein, when executed by said data-processing system, said instructions cause said data-processing system to pre-process said document to enable said question-answering system to extract said content in response to a natural-language question that has been posed by a user of said natural-language system, wherein pre-processing said document comprises: using a visual structure of said document to extract structural information from said document and incorporating said structural information in said knowledge base, wherein said visual structure of said document comprises a spatial distribution of said visual segments in said document when said document is in a form that renders said document visible, wherein using said visual structure to extract structural information from said document comprises: based at least in part on locations of said first and second visual segments relative to each other within said document, determining that a context relationship is to exist between said first and second visual segments, based at least in part on locations of said first and third visual segments relative to each other within said document, determining that no context relationship is to exist between said first and third visual segments, and establishing said context relationship between said first and second visual segments.