U.S. patent application number 15/217659 was filed with the patent office on 2016-11-10 for methods and data structures for improved searchable formatted documents including citation and corpus generation.
The applicant listed for this patent is Kendyl A. Roman. Invention is credited to Kendyl A. Roman.
Application Number | 20160328374 15/217659 |
Document ID | / |
Family ID | 51790195 |
Filed Date | 2016-11-10 |
United States Patent
Application |
20160328374 |
Kind Code |
A1 |
Roman; Kendyl A. |
November 10, 2016 |
Methods and Data Structures for Improved Searchable Formatted
Documents including Citation and Corpus Generation
Abstract
Computer searchable annotated formatted documents are produced
by correlating documents stored as a photographic or scanned
graphic representations of an actual document (evidence, report,
court order, etc.) with textual version of the same documents. A
produced document will provide additional details in a computer
data structure that supports citation annotation as well as other
types of analysis of a document. The computer data structure also
supports generation of citation reports and corpus reports. A
computer method of creating searchable annotated formatted
documents including citation and corpus reports by correlating and
correcting text files with photographic or scanned graphic of the
original documents. Data structures for correlating and correcting
text files with graphic images. Generation of citation reports,
concordance reports, and corpus reports. Data structures for
citation reports, concordance reports, and corpus reports
generation.
Inventors: |
Roman; Kendyl A.;
(Sunnyvale, CA) |
|
Applicant: |
Name |
City |
State |
Country |
Type |
Roman; Kendyl A. |
Sunnyvale |
CA |
US |
|
|
Family ID: |
51790195 |
Appl. No.: |
15/217659 |
Filed: |
July 22, 2016 |
Related U.S. Patent Documents
|
|
|
|
|
|
Application
Number |
Filing Date |
Patent Number |
|
|
13872907 |
Apr 29, 2013 |
9405749 |
|
|
15217659 |
|
|
|
|
12586130 |
Sep 16, 2009 |
8433708 |
|
|
13872907 |
|
|
|
|
61192169 |
Sep 16, 2008 |
|
|
|
Current U.S.
Class: |
1/1 |
Current CPC
Class: |
G06F 40/289 20200101;
G06F 40/30 20200101; G06F 40/169 20200101; G06F 16/93 20190101;
G06F 40/103 20200101; G06F 40/205 20200101; G06K 9/18 20130101;
G06K 9/00463 20130101 |
International
Class: |
G06F 17/24 20060101
G06F017/24; G06F 17/27 20060101 G06F017/27; G06K 9/00 20060101
G06K009/00; G06F 17/21 20060101 G06F017/21; G06F 17/30 20060101
G06F017/30; G06K 9/18 20060101 G06K009/18 |
Claims
1. A non-transitory computer readable medium encoded with program
instructions which are executed by a computer to provide a method
of generating internal citations for a formatted document, the
instructions comprising the steps of: a) obtaining graphic
representations of each page of the formatted document, wherein the
formatted document is a patent document, b) optically recognizing
characters from the graphic representations, and determining the
position of the characters on each page, c) obtaining a separate
and distinct text version of the textual content of the formatted
document from a source separate and distinct from the formatted
document and the graphic representations obtained therefrom, d)
parsing text from the text version, the parsed text being separate
and distinct from the recognized characters, e) correlating the
recognized characters with the parsed text to determine an internal
citation for each sentence, wherein the internal citation
identifies the document and a citation location inside the document
where the corresponding sentence is found, f) creating a data
structure storing data determined in the correlating step.
2. The computer readable medium of claim 1 wherein the citation
location comprises: i) an internal citation column number; and ii)
an internal citation line number;
3. The computer readable medium of claim 1 further comprising a
step of: using the parsed text to correct the recognized
characters, yielding a corrected formatted document.
4. The computer readable medium of claim 2 further comprising a
step of: attaching the data structure to the formatted document,
wherein, when a portion of text is copied from the formatted
document, a corresponding internal citation is included with the
copied portion.
5. The computer readable medium of claim 4 wherein the attaching
step yields a searchable annotated formatted document.
6. The computer readable medium of claim 2 wherein the data
structure comprises citation start data, comprising a start column
number and a start line number.
7. The computer readable medium of claim 6 wherein the data
structure comprises citation end data, comprising an end column
number and an end line number.
8. The computer readable medium of claim 1 wherein the data
structure comprises: a) internal citation page number data or
internal citation paragraph number data and b) internal citation
sentence number data.
9. The computer readable medium of claim 1 wherein the parsing the
text step further includes at least one of the group of: a)
determining new paragraphs, and b) determining paragraph types.
10. The computer readable medium of claim 1 wherein the parsing the
text step further includes determining document parts, said
document parts each comprising a distinct set of pages, wherein the
document parts includes at least one of the group of: a) abstract,
b) drawing, c) specification, and d) claims.
11. The computer readable medium of claim 1 wherein the parsing the
text step further includes determining document sections, said
document sections each comprising a distinct group of paragraphs,
under one or more headings, wherein the document sections includes
at least one of the group of: a) field of invention, b) background
of invention, c) summary of invention, d) description of drawings,
and e) description of invention.
12. The computer readable medium of claim 1 wherein the determining
the position of the characters substep further includes at least
one of the group of: a) assembling lines, b) allocating lines to
columns, and c) calculating line numbers.
13. The computer readable medium of claim 1 wherein the correlating
step further includes at least one of the group of: a) determining
column numbers, and b) determining line numbers.
14. The computer readable medium of claim 1 wherein the formatted
document contains patent drawing figures, and wherein the
correlating step further includes at least one of the group of: a)
determining figure numbers in the drawing figures, and b)
determining figure item numbers in drawings figures.
15. The computer readable medium of claim 1 wherein the correlating
step further includes at least one of the group of: a) determining
patent claim numbers, and b) parsing patent clauses and determining
patent clause numbers.
16. The computer readable medium of claim 1 further comprising a
step of: generating a citation document using the correlation data
structure.
17. The computer readable medium of claim 1 further comprising a
step of: generating a concordance report using the correlation data
structure, the concordance report comprising rows comprising: a) a
word or phrase, and b) one or more internal citations, indicating
where the word or phrase occurs in the formatted document.
18. The computer readable medium of claim 1 further comprising a
step of: generating a patent corpus report using the correlation
data structure, the patent corpus report comprising rows
comprising: a) prior context comprising the entire prior portion of
the parsed sentence, b) a word or phrase, c) subsequent context
comprising the entire subsequent portion of the parsed sentence,
and d) an internal citation.
19. The computer readable medium of claim 18 wherein the patent
corpus report is based on a single word root.
20. The computer readable medium of claim 18 wherein the patent
corpus report is based on one of the group of: a) a phrase, and b)
a set of words having similar meaning or usage.
Description
RELATED APPLICATION
[0001] This application claims priority under 35 U.S.C.
.sctn.199(e) of the co-pending U.S. provisional application Ser.
No. 61/192169, filed Sep. 16, 2008, entitled "METHODS AND DATA
STRUCTURES FOR IMPROVED SEARCHABLE FORMATTED DOCUMENTS INCLUDING
CITATION AND CORPUS GENERATION," which is hereby incorporated by
reference.
FIELD OF THE INVENTION
[0002] This invention relates to improved searchable formatted
electronic documents and analysis tools, such as citation and
corpus generation. Examples of documents include patents, patent
applications, evidence files, and other documents which are
available in graphic form and optionally also available in a text
form.
BACKGROUND OF THE INVENTION
[0003] In the field of electronic document management there are
many situations where a document is stored electronically on a
computer system as a photographic or scanned graphic of the actual
document. For example, in a litigation document management system
example documents may represent evidence, reports, court orders,
patent documents, etc. The graphic image of the page is critical in
many cases and needs to be preserved. However, there is also a need
to electronically search the document using a computer.
Additionally, there has been a long felt need to be able to cut the
text from a document and have an accurate internal citation, or
location identification, automatically pasted into a new document
(e.g. report, brief, etc.) with the text that was cut. In
litigation, having analysis, reports, and arguments error free is
very important and a significant amount of time spent creating
quotes and internal citations and then in checking them to ensure
accuracy.
[0004] What is needed is a way to analyze documents in its graphic
format and then be able to generate quotations with accurate
internal citations using a computer. Also in patent analysis, for
example, what also is needed is a way to thoroughly review all
occurrences of certain terms in context to be able to thoroughly
and accurately determine the meaning of those terms.
SUMMARY OF THE INVENTION
[0005] The current invention provides the ability to produce
computer searchable annotated formatted documents by correlating
documents stored as a photographic or scanned graphic
representations of an actual document (evidence, report, court
order, etc.) with textual version of the same documents. A produced
document will provide additional details using computer data
structure(s) that would support the above described citation
annotation as well as other types of analysis of a document. The
data structure(s) also support computer generation of citation and
corpus reports.
Objects and Advantages
[0006] Accordingly, beside the objects and advantages described
above, some additional objects and advantages of the present
invention are: [0007] 1. To provide a quicker and more effective
method analyzing documents. [0008] 2. To provide a highly accurate,
electronically searchable document from graphic images of the
document pages. [0009] 3. To provide means and methods of document
analysis that are easy to use. [0010] 4. To reduce the cost of
document analysis. [0011] 5. To reduce the cost of evidence
analysis. [0012] 6. To improve the thoroughness of documents
analysis. [0013] 7. To improve the thoroughness of patent claim
term analysis. [0014] 8. To identify inconsistencies in the use of
document terms, e.g. usage of disputed patent claim terms.
BRIEF DESCRIPTION OF THE DRAWINGS
[0015] FIG. 1A illustrates a process by which remote documents are
obtained and collected and then subsequently converted to other
formats.
[0016] FIG. 1B illustrates a process by which a searchable
annotated formatted document is produced.
[0017] FIG. 2A illustrates data structures that are used to verify,
correlate and correct meta-data data, for example, patent OCR
data.
[0018] FIG. 2B illustrates an embodiment of a text data
structure.
[0019] FIG. 2C illustrates an embodiment of a meta-data data
structure.
[0020] FIG. 3A illustrates a text parse routine.
[0021] FIG. 3B illustrates a searchable formatted document parse
routine.
[0022] FIG. 4 illustrates a correlate and correct routine.
[0023] FIG. 5 illustrates an annotation routine.
[0024] FIG. 6 illustrates generation of a concordance.
[0025] FIG. 7A illustrates generation of a word corpus.
[0026] FIG. 7B illustrates an embodiment of word corpus.
[0027] FIG. 8A illustrates the generation of a citation report.
[0028] FIG. 8B illustrates an embodiment of a citation report.
TABLE-US-00001 REFERENCE NUMERALS IN DRAWINGS 1 local computer 2
network 3 remote computer 4 network accessible file collection 5
converted file 6 output device 10 graphic images 20 text file 30
formatted document 40 searchable formatted document 50 corrected
formatted document 60 searchable annotated formatted document 62
parse routine 64 correlate and correct routine 66 attach data
routine 68 tag words and sentences routine 71 relative fields 72
citation fields 73 part fields 74 section fields 75 text for
sentences fields 76 doc ID field 80 text field 81 sequential
paragraph number 82 sequential sentence number 83 new paragraph
start 84 citation start column # field 85 citation start line #
field 86 citation start word # field 87 citation end column # field
88 citation end line # field 89 citation end word # field 90 part
field 91 section field 92 specification part 98 doc ID field 102
OCR page # 104 OCR column # 106 OCR line # 108 OCR line coordinates
110 OCR new paragraph start 112 OCR line font size 114 OCR line
text 116 OCR line top coordinate 118 OCR line bottom coordinate 120
OCR line left coordinate 122 OCR line right coordinate 200 data
structure(s) 200 a first (text) data structure 200 b second
(meta-data) data structure 302 read text 304 determine document
parts 306 determine document sections 308 determine new paragraphs
310 determine paragraph types 312 apply relative numbers to each
section 314 read text output 316 determine parts output 318
determine sections output 320 determine new paragraph output 322
determine paragraph type output 324 re-apply relative numbers to
each section 326 read searchable formatted document 328 assemble
lines 330 allocate lines to columns 332 calculate line numbers 334
read searchable formatted document output 336 assemble lines output
338 allocate lines to columns output 402 read text file 404 read
searchable formatted document 406 match text 408 determine column
line and word values 410 contains figures? 412 determine figure #
and item # 414 contains claims? 416 determine claim and clause #
420 read text file output 422 read searchable formatted document
output 424 match text output 426 determine column line and word
values output 428 determine figure # and item # output 600
concordance program 602 concordance 700 corpus program 702 word or
phrase corpus 710 prior content column 712 word (or phrase) column
714 subsequent content column 716 citation column 720 heading row
722 corpus sample row 1 724 corpus sample row 2 728 corpus last
sample row 800 citation program 802 citation document 804 citation
document title 810 (a-d) citation document section title 812 (a-d)
citation sentence 814 (a-d) citation annotation
SPECIAL DEFINITIONS
[0029] corpus--a collection of recorded statements used as a basis
for the descriptive analysis of language in a written document
[0030] concordance--an index of the important words used in a
written document
[0031] annotation--extra information which is not normally
displayed, such as citation information from a data structure that
provides citations for text cut from a formatted document
DETAILED DESCRIPTION OF THE INVENTION
[0032] FIG. 1A illustrates a process by which remote documents are
obtained and collected and then subsequently converted to other
formats. Local computer 1 connects to a remote computer 3 via a
network 2. Then it accesses file data from a network accessible
file collection 4 and retrieves the desired files onto the local
computer 1 over the network 2. Once on the local computer 1, the
files can be converted into a single converted file 5. Once
converted, the file can be output to a peripheral device 6, such as
a display or a printer (as shown). Peripheral devices 6 are well
known to include hard disk drives, floppy disk drives, tape drives,
flash drives, CD drives, DVD drives. Peripheral devices 6 are well
known to accept various types of computer readable media such as
hard disk platters, floppy disks, tapes, memory chips, CDs, DVDs,
and similar media. The computer readable medium may store the
program instructions that make up a computer software program or
routine, computer data, and/or computer data structures.
[0033] For example, the United States Patent and Trademark Office
(USPTO) has a service (remote computer 3) which provides patent
publications as TIFF files, one file for each page.
[0034] A patent related embodiment performs the following steps on
local computer 1: [0035] a) Input a patent number [0036] b) Access
the USPTO World Wide Web site (remote computer 3) over the Internet
(network 2) to obtain the HTML text version of a patent and to use
that HTML to determine the number of pages represented by graphic
images (e.g. TIFF) [0037] c) Download each page's graphic image
from the USPTO World Wide Web site (remote computer 3) over the
Internet (network 2) [0038] d) Convert the collection of graphic
images into a single document (e.g. PDF or multipart TIFF). [0039]
e) Optically recognize (via OCR) the page graphic images.
Alternatively, each page's graphic image can be processed on the
fly (e.g. recognized as each is downloaded in step c above).
[0040] These steps run on one computer or on a group of computers.
These steps could be implemented in computer software. Example
embodiments include an Acrobat plug-in or a World Wide Web browser
plug-in. Good results have also been obtained implementing these
steps as a script running on a group of computers including one
computer running an OCR engine (such as OmniPage, TextBridge, or
other commercially available OCR engine) and another computer
running an Oracle database.
[0041] A novel improvement in the system illustrated in FIG. 1A is
that an HTML (or plain text) version of the same subject matter can
be used to correct spelling in an OCR document created from graphic
page images (e.g. using Acrobat OCR capture). For example, in the
embodiment for U.S. patents, the USPTO also provides an HTML
version of the patent that can be used to correct and correlated
the OCR text. It is well known that OCR is not 100% perfect and
human comparison and correction is costly. This aspect of the
system can significantly improve the value of searchable documents
created by OCR.
[0042] Once the graphic images are converted (and optionally
corrected), the document 5 can be printed, for example on printer
6.
[0043] FIG. 1B illustrates production of a searchable annotated
formatted document 60. The U.S. patent example will be used to
illustrate this aspect of the system. The USPTO provides a graphic
image 10 for each page of a published patent. These graphic images
10 (e.g. TIFF files) are not text searchable (they are like
photographs). A formatted document 30, such as an Adobe Acrobat
PDF, can be created as a binder holding every graphic image 10
(routine A). See also FIG. 1A. The formatted document 30 (e.g. PDF
file) can be processed with OCR (routine B) to convert the images
to searchable text forming a searchable formatted document 40.
Alternatively, an OCR engine can provide structured data which
describes the text elements found in the document with the graphic
location of each element, and which can be used instead of
searchable formatted document 40. The USPTO also provides a
separate and distinct text file 20 in HTML format for many patents.
Unlike the graphic images 10 which are not text searchable, the
text file 20 is electronically searchable. The searchable formatted
document 40 is correlated with the text file 20 to correct the OCR
text (routine C) resulting in a corrected formatted document
50.
[0044] The corrected formatted document 50 is a valuable tool for
analysis.
[0045] Further, the corrected formatted document 50 can be used to
add various annotations (routine D) to produce a searchable
annotated formatted document 60. "Annotated" is not used in the
general sense as would be understood by one of skill in the art.
The word is used here in a new limited way to refer to the
annotations from the data structure(s) 200 that provide the
internal citations for quotes pasted from the searchable annotated
formatted document 60. Again using the U.S. patent example, the
searchable annotated formatted document 60 version of a patent can
be used as the primary analysis document in a patent litigation or
evaluation. When an expert or attorney wants to refer to a
particular section of the patent, the user simply selects the
desired words and a highly accurate quote and citation, including
for example, column and line numbers, is automatically generated
and can be pasted into a report or brief.
[0046] FIG. 2A illustrates computer data structures that are used
to verify, correlate and correct document data, for example, patent
OCR data. The computer data structures 200 can be implemented in
various ways. Good results have been obtained by implementing them
as Microsoft Excel spreadsheets, per1 data structures, XML files,
or Oracle database data tables.
[0047] The first data structure 200a generally contains each unit
with information used to provide an internal citation, e.g. "('498
Patent 12:47-13:5)". An example of a unit would be a sentence or a
title. This data is substantially obtained from the HTML (or text
version of the document) and may contain estimated values. In one
embodiment, a spreadsheet document (such as Microsoft Excel) with
functions is used to estimate the citations (to improve correlation
ease or accuracy).
[0048] The second data structure 200b generally contains each line
(or column line). Each column line has only one column and/or line
number associated with it. This data preferably comes from OCR data
(but may be input by people).
[0049] The data between these two structures are correlated with a
unique ID for each entry in the tables. The sentence data has an ID
for each sentence. The column line data has an ID for each line.
The line data is matched, if possible, to each sentence's unique ID
and give a sentence a relative ordinal number. The actual citation
data, such as start column, start line, end column, and end line,
is filled in the first data structure 200a based on matching first
and last line data. The OCR text is corrected by verification
against the HMTL data. Missing data in the data structure is
flagged and estimates are used.
[0050] FIG. 2B illustrates an exemplary text data structure 200a
that holds information that will be used to provide the citations
for quotes pasted from the searchable annotated formatted document
60 or citations on a word or phrase corpus 702. The text data
structure 200a also provides an alternative output that can be used
independently from the formatted documents (30, 40, 50 or 60). The
text data structure 200a is first instantiated by a parsing routine
(FIG. 3) and updated by a correlation and correction routine (FIG.
4).
[0051] FIG. 2C illustrates an exemplary meta-data data structure
200b that holds information that will be used to as an interim
repository for the data parsed from the searchable formatted
document 40, or alternatively obtained from OCR.
[0052] FIG. 3A illustrates a text parse computer routine (which
could be implemented in Perl, for example) which reads the text
from the text file 20 or another source of text (step 302). It
determines the document parts (step 304). For example, in a U.S.
patent embodiment, parts could be one of Abstract, Drawing,
Specification, or Claims. Then it determines document sections
(step 306). For example, in a U.S. patent embodiment, document
sections include Background of the Invention, Summary of the
Invention section, and so forth. Then it determines new paragraphs
(step 308). Then it determines paragraphs types (step 310). For
example, regular paragraph text, tables, equations, etc. In step
312, the text data structure 200a in FIG. 2B would be filled in
except for the citation fields (72). The text for each heading or
sentence would be filled in the text field (80). Relative numbers
would be filled in, for example in the specification part (92),
each paragraph would be assigned a sequential paragraph number (81)
and sequential sentence number (82). Sentence numbers would restart
at 1 for each paragraph. In addition, any new paragraph starts (83)
would be identified. In the patent drawings, for example, any text
from the figures would be indexed. In the claims, each clause would
be relative to the claim. The part fields (73, 90) and the section
fields (74, 91) would be filled in.
[0053] At this point the text data structure 200a (without the
citations) has an alternative use. It can be used to produce a text
version of the subject document (e.g. patent) with numbered
paragraphs (e.g. for electronic filing) e.g. "[103] Referring to
FIG. 1B . . . . "
[0054] FIG. 3B illustrates the searchable formatted document parse
computer routine (which could be implemented in Perl, for example)
which reads the searchable formatted document 40 (step 326). Then
it takes the words and assembles the lines (step 328). Then it
allocates each line to either the left or right column (step 330).
Then it calculates the line numbers for each line (step 332). At
this point, the meta-data data structure 200b in FIG. 2C would be
filled in completely.
[0055] FIG. 5 illustrates the annotation computer routine which is
an implementation of routine D in FIG. 1B. First the text (20 or
50) is parsed by the parse routine 62 (e.g. using the text parse
routine shown in FIG. 3A) and the searchable formatted document
parse routine (e.g. see FIG. 3B). Next the text data structure 200a
is updated with correlation and correction information by the
correlate and correct routine 64. For example, see FIG. 4.
[0056] The correlation information in the patent example would be
the citation column, line and word numbers for the start and end of
each sentence. The end could be optional because, for example, it
could be determined by looking at the next record's start. The next
step is to attach the data (e.g. FIG. 2B) by the attach data
routine 66 to the document. In some embodiments, the data would be
meta-data that is not normally viewable in normal display modes. In
other embodiments, the data in each record could be stored as a
user displayable "annotation" (in the general sense) at the user's
option. In other embodiments, the data is simply appended to the
end of the document e.g. as a table.
[0057] Next the words and/or sentences are tagged by the tag words
and sentences routine (68) in the formatted document so that when a
portion is cut or copied, the citation associated with the text is
cut or copied with it. Special software associated with the
document viewer would handle the "cut" and/or "copy" operation. For
example, in PDF or FrameMaker or Microsoft Word a plug-in could be
added (either by the publisher or as a 3.sup.rd party plug-in) to
provide text plus the citation (e.g. "U.S. Pat. No. 8,888,523
17:40-18:3") in the paste buffer.
[0058] Note that step 68 in FIG. 5 could be optional because the
plug-in could use the attached data (step 66) to provide the
citation in the paste buffer. However, by tagging the documents,
the determination of the citation for a particular set of words is
more straightforward at the time of the cut and/or copy
operation.
[0059] FIG. 4 illustrates an example of the correlate and correct
step 64 of FIG. 5. A text file 20 is read (step 402). A searchable
formatted document 40 is read (step 404). The order of steps 402
and 404 is not significant, and may be performed in parallel. In
step 406, the text file 20 in the form of the text data structure
200a is compared with a searchable formatted document 40 in the
form of the meta-data data structure 200b. The relative text is
matched to the positional information in the graphic version of the
same document to determine page or column number, line number, and
word number (92) or the line (step 408). If the document contains
figures (decision 410), then the figure numbers and item numbers
(94) are determined in step 412. If the document contains claims
(decision 414), then the claim numbers and clause numbers are also
determined in step 416. Thus the correlation routine updates the
text data structure 200a (FIG. 2B) with the start (and optionally
end) citation information 72 (84-89).
[0060] If the document or page has a document ID (e.g. a BATES
number), it is added in (76, 98). After the correlate and correct
routine (e.g. for example, see FIG. 4) is complete the data
structure can be used in many ways including but not limited to:
[0061] 1. Annotation of the corrected formatted document 50 to form
the searchable annotated formatted document 60. [0062] 2.
Production of a spreadsheet containing each text element and
citation where by a reader can find text in the spreadsheet by
electronically searching and quickly find the internal citation or
location of the text in a paper version. [0063] 3. Input to a
concordance program 600 that identifies each occurrence of a word
or phrase and produces a concordance 602 with each sentence with
the specified word or phrase and its citation (FIG. 6). [0064] 4.
Input to a corpus program 700 that identifies each occurrence of a
word or phrase and produces a corpus 702 with each sentence and
with the specified word or phrase and its citation (FIG. 7A). FIG.
7B shows an embodiment of word or phrase corpus 702. [0065] 5.
Input to a citation program 800 that produces a citation document
802 with each sentence and with its citation (FIG. 8A). FIG. 8B
shows an embodiment of a citation document 802.
[0066] Note the word "corpus" generally means "entire body." In the
context of the field of language analysis it refers to a collection
of recorded utterances used as a basis for the descriptive analysis
of a language. In the context of this invention the word "corpus"
(as well as "concordance") is limited to the language used in a
single document (e.g. a patent) or a small group of related
documents (e.g. a set of related patents with common inventorship
or subject matter).
[0067] The corpus 702 could be corpus for a single word or phrase
i.e. a "word corpus" or for all the words in the document i.e. a
"document corpus" or for a set of key words (e.g. a) similar to
those selected for a concordance, or b) a set of disputed terms).
This invention is not limited to patent analysis. It is also useful
for analysis of other evidence, such as e-mail, source code,
contracts, discovery documents, Web pages, contracts, etc.
[0068] The methods for:
[0069] generating a corrected formatted document 50 (e.g. FIG.
1B),
[0070] generating a searchable annotated formatted document 60
(e.g. FIG. 1B),
[0071] generating a concordance 602 (e.g. FIG. 6),
[0072] generating a corpus 702 (e.g. FIG. 7A), and
[0073] generating a citation document 802 (e.g. FIG. 8A), can each
be implemented on a web site computer enabled by a database for
storing the correlation data and various generated documents.
ADVANTAGES
Searchable
[0074] Text searches can be performed within the document or across
multiple documents.
Copy and Paste Citation
[0075] Each sentence in the document can be copied and pasted into
other documents. The internal citation for each sentence (or group
of sentences) is copied and pasted along with the sentence.
Original Document Format Maintained
[0076] The original format of the document is maintained after it
has been converted to a searchable format.
Identify Location in Original Document
[0077] As formats in the text file 20 may differ from the original
formatted document 30, it is important that the annotations or
citations correlate with each sentence and/or line and/or heading
to identify its relative location with a document.
Accuracy
[0078] Optical Character Recognition (OCR) systems have struggled
for decades to improve accuracy. OCR does very poorly on some
documents. Further, for some situations, such as an expert witness
report or Federal Court brief, errors can have severe impact on the
perception of the facts in the case and ultimately cost millions of
dollars. The present invention provides a way to improve accuracy
of the computer generated data as well as improve the accuracy of
human construction of reports and briefs.
Thorough
[0079] The concordance and corpus features of present invention
provide a means for through analysis of every instance of a word or
phase, or related phrases, in a document of related set of
documents. For example, in a patent case, the corpus documents on
disputed claim terms will not only make claim term analysis more
efficient, but will also ensure that all usage is considered and
inconsistencies are understood. The use of corpus reports will
improve the rigorousness of claim term analysis and provide judges
with stronger input upon which to base their judgments.
Cost Saving
[0080] The present invention provides several aspects that will
save time and improve the performance of highly paid experts,
analysis, and attorneys. Thus, significant savings will result from
use of the present invention.
Conclusion, Ramification, and Scope
[0081] Accordingly, an aspect of the present invention provides
searchable annotated formatted documents that are produced by
correlating documents stored as a photographic or scanned graphic
representations of an actual document (evidence, report, court
order, etc.) with textual version of the same documents. A produced
document will provide additional details in a data structure that
supports citation annotation as well as other types of analysis of
a document. The data structure also supports generation of citation
reports and corpus reports. A method aspect includes creating
searchable annotated formatted documents including citation and
corpus reports by correlating and correcting text files with
photographic or scanned graphic of the original documents. Data
structures for correlating and correcting text files with graphic
images are valuable output by themselves. Another method aspect
includes generation of citation reports, concordance reports, and
corpus reports. Data structures provide for citation reports,
concordance reports, and corpus reports generation.
[0082] While the above descriptions contain several specifics these
should not be construed as limitations on the scope of the
invention, but rather as examples of some of the preferred
embodiments thereof. Many other variations are possible. For
example, although U.S. patent data has been used in the examples,
the document system could be applied to other categories of
documents. Some embodiments would target specific types of
documents. The routines could be implemented in hardware or using
various software platforms. Additionally, the system could have
additional features, or be used in different countries without
departing from the scope and spirit of the novel features of the
present invention.
[0083] Accordingly, the scope of the invention should be determined
not by the illustrated embodiments, but by the appended claims and
their legal equivalents.
* * * * *