Ex Parte Spellward et al

Patent Trial and Appeal Board

May 25, 2018

13827688 (P.T.A.B. May. 25, 2018)

UNITED STA TES p A TENT AND TRADEMARK OFFICE
APPLICATION NO. FILING DATE
13/827,688 03/14/2013
106095 7590 05/30/2018
Baker Botts LLP/CA Technologies
2001 Ross Avenue
SUITE 900
Dallas, TX 75201
FIRST NAMED INVENTOR
Peter C. Spellward
UNITED STATES DEPARTMENT OF COMMERCE
United States Patent and Trademark Office
Address: COMMISSIONER FOR PATENTS
P.O. Box 1450
Alexandria, Virginia 22313-1450
www .uspto.gov
ATTORNEY DOCKET NO. CONFIRMATION NO.
063170.9886 3032
EXAMINER
CHU, JENQ-KANG
ART UNIT PAPER NUMBER
2176
NOTIFICATION DATE DELIVERY MODE
05/30/2018 ELECTRONIC
Please find below and/or attached an Office communication concerning this application or proceeding.
The time period for reply, if any, is set in the attached communication.
Notice of the Office communication was sent electronically on above-indicated "Notification Date" to the
following e-mail address(es):
PTOmaill@bakerbotts.com
PTOL-90A (Rev. 04/07)
UNITED STATES PATENT AND TRADEMARK OFFICE
BEFORE THE PATENT TRIAL AND APPEAL BOARD
Ex parte PETER C. SPELL WARD and
HOWARD C. SNART WOODHOUSE
Appeal 2017-011054
Application 13/827 ,688
Technology Center 2100
Before JUSTIN BUSCH, JAMES W. DEJMEK, and JASON M. REPKO,
Administrative Patent Judges.
REPKO, Administrative Patent Judge.
DECISION ON APPEAL
STATEMENT OF THE CASE
Appellants 1 appeal under 35 U.S.C. § 134(a) from the Examiner's
rejection of claims 1-19 and 21. Br. 8. 2 Claim 20 was canceled. We have
jurisdiction under 35 U.S.C. § 6(b ).
We affirm-in-part.
1 Appellants identify the real party in interest as CA, Inc. Br. 3.
2 Throughout this opinion, we refer to the Final Rejection ("Final Act.")
mailed August 25, 2016; the Appeal Brief ("Br.") filed January 23, 2017;
and the Examiner's Answer ("Ans.") mailed May 19, 2017.
Appeal2017-011054
Application 13/827,688
THE INVENTION
Appellants' invention allows a user to define a classifier using a
graphical user interface. Spec. ,r 3. Classifiers are tools for content
analysis, such as identifying characteristics of a document's text. Id. ,r 4.
One embodiment highlights text that satisfies a classifier's parameters.
Id. ,r 20. The interface can use different colors to highlight text associated
with different classifiers. Id. ,r 30. For example, text associated with a
street classifier ("123 PARK STREET") could be highlighted in a different
color from text associated with a zip-code classifier. Id.
Claim 1 is reproduced below with our emphasis:
1. A method for generating a classifier, comprising:
displaying a document;
enabling a user to define a first classifier by specifying one
or more parameters using a graphical user interface;
storing the one or more parameters for the first classifier;
identifying one or more first portions of text of the
document that satisfy the one or more user-defined parameters
for the first classifier; and
in response to identifying one or more first portions of text
that satisfy the one or more user-defined parameters,
highlighting the first portions of text displayed in the document
to indicate that the first portions of text satisfy the user-defined
parameters for the first classifier.
THE EVIDENCE
The Examiner relies on the following as evidence:
Wang et al.
Baker et al.
US 2012/0072859 Al
US 2013/0315480 Al
2
Mar. 22, 2012
Nov. 28, 2013
Appeal2017-011054
Application 13/827,688
THE REJECTION
Claims 1-19 and 21 stand rejected under 35 U.S.C. § 103(a) as
unpatentable over Wang and Baker. Final Act. 7-16.
CLAIMS 1, 7-11, 17-19, AND 21
The Examiner's Findings
The Examiner finds that Wang teaches every limitation recited in
representative 3 claim 1 except for performing the highlighting step in
response to the identifying step. Final Act. 7-10. The Examiner cites Baker
as teaching iterating over the classification process to refine the results.
Id. at 9. The Examiner concludes that it would have been obvious to iterate
over Wang's identifying and highlighting to arrive at a method that performs
a highlighting step in response to an identifying step. Id. at 9-10.
Appellants ' Contentions
Appellants argue that Wang does not teach the recited identifying
step. Br. 10. According to Appellants, Wang's user provides images of
recognized words to train the text recognition. Id. In Appellants' view,
Wang's user-provided image is not a user-defined parameter. Id.
Appellants further argue that Wang's user-provided labels are not the recited
user-defined parameters because Wang does not identify text that satisfies
the labels. Id. Appellants contend that "the user does the identification
himself." Id.
3 Appellants present the same argument for claims 1, 11, and 21. See Br. 9-
11. We select independent claim 1 as representative of claims 1, 11, and 21.
See 37 C.F.R. § 4I.37(c)(l)(iv).
3
Appeal2017-011054
Application 13/827,688
Appellants also argue that Baker does not highlight text in response to
the recited identifying. Id. at 11. In Appellants' view, Baker only identifies
and classifies text based on images or other objects. Id.
Analysis
Claim 1 recites, in part, "identifying one or more first portions of text
of the document that satisfy the one or more user-defined parameters for the
first classifier" ( emphasis added).
Regarding the recited parameters, the Specification discloses an
application that highlights, or otherwise visually differentiates, text that
satisfies the parameters of one or more classifiers. Spec. ,r 20, cited in Br. 5
(Summary of Claimed Subject Matter). The classifier's parameters may
include stored text, and the application may identify matching text in a
document. Spec. ,r 21. Parameters may also include Boolean or logical
operations connecting different text strings, and the application may identify
text that satisfies the Boolean or logical expression. Id. In another example,
parameters include other classifiers-i.e., sub-classifiers-and the
application may identify text that satisfies the sub-classifier's parameters.
Id. In one example, the user can define these parameters by selecting text.
Id. ,r 25. On this record, a broad, but reasonable, interpretation of the term
"user-defined parameters for the first classifier," consistent with the
Specification, includes data----e.g., text, logical expressions, or classifiers-
that are used by the classifier to identify text in the document.
The Examiner finds that Wang teaches the recited user-defined
parameters. Final Act. 8 (citing Wang ,r,r 87, 158). We agree. Like the
4
Appeal2017-011054
Application 13/827,688
above discussed examples (Spec. ,r,r 20, 21, 25), Wang's parameters are for a
classifier and are used to identify text in a document (Wang ,r,r 87, 158).
Specifically, Wang teaches a method for reducing OCR 4 errors in
document processing. Wang ,r 158. Wang uses classifiers to segment the
electronic document. Id. ,r 56. Wang's segmentation identifies one or more
logical parts from the document----e.g., words, phrases sentences, or
paragraphs. Id. ,r 87. The method then processes images of individual
words. Id. To improve recognition accuracy, Wang's user trains a classifier
by providing images of recognized words. Id. ,r 158. According to the
Examiner, Wang's user-provided images correspond to the recited user-
defined parameter. Final Act. 8. We agree because Wang's classifier uses
the image to carry out an improved text identification. See Wang ,r,r 87, 158.
Contrary to Appellants' argument (Br. 9--10), Wang's identified
portions of text "satisfy" these parameters (Wang ,r,r 87, 158). In Wang, the
user-provided images are of recognized words. Wang ,r 158. Wang's
method identifies words in the document using classifiers trained on the
provided images. See id.; see also id. ,r,r 60-62 ( discussing "word finding").
Because the user provides images of recognized words, we agree that Wang
at least suggests the documents contain the words in the user-provided
image, and the classifiers would then be trained to identify where those
words occur. See id. ,r,r 60-62, 158. Appellants argument that the "user
does the identification himself' (Br. 10) disregards Wang's teaching that the
user-provided images are used to "train the classifier," not to perform the
identification itself (Wang ,r 158). On this record, the weight of the
4 Optical Character Recognition. Wang ,r 58.
5
Appeal2017-011054
Application 13/827,688
evidence favors the Examiner's finding that Wang teaches or suggests the
recited identifying step. Final Act. 8.
Appellants' arguments regarding the highlighting step do not address
the Examiner's proposed combination of Wang and Baker. See Br. 11. In
particular, these arguments focus on Baker's lack of highlighting in response
to the identifying step. Id. The Examiner, however, relied upon Wang for
the highlighting step. Final Act. 9. The Examiner then concluded that based
on the combined teachings of Wang and Baker it would have been obvious
to highlight the identified text in response to Baker's identification. Id. at 8-
10.
We agree with the Examiner that Baker, for its part, teaches iterating
on the classification process to refine the results. Baker ,r 3, cited in
Final Act. 9. Therefore, in combination with Wang's classifier training and
highlighting, we agree that it would have been obvious to highlight the text
in response to the identification when using Baker's iterative process.
Final Act. 9-10. In this way, the Examiner's proposed enhancement
predictably uses the prior art elements according to their established
functions, which is an obvious improvement. See KSR Int 'l Co. v. Teleflex,
Inc., 550 U.S. 398,417,421 (2007).
We, therefore, sustain the Examiner's rejection of representative claim
1 and claims 11 and 21, which fall with claim 1. We also sustain the
Examiner's rejection of dependent claims 7-10 and 17-19, which are not
separately argued with particularity. See Br. 9-15.
6
Appeal2017-011054
Application 13/827,688
CLAIMS 2--4, 6, 12-14, AND 16
Claim 2 recites, in part, "displaying to the user a hierarchy of a
plurality of classifiers that have been selected by the user."
The Examiner finds that Wang displays the recited classifier
hierarchy. Final Act. 10 (citing Wang ,r,r 50, 178). According to the
Examiner, Wang highlights a part of a document, and this highlighting "is
the indication of classifiers." Ans. 5 (citing Wang Fig. 4B). Furthermore,
the Examiner finds that Wang's documents have a textual hierarchy----e.g.,
words, phrases, sentences, or paragraphs. Ans. 6.
Appellants argue that Wang does not display a hierarchy of classifiers.
Br. 11-12. According to Appellants, Wang only shows that a document
contains logical parts of a textual hierarchy. Id. at 12. We agree.
Regarding the recited displaying step, the Specification shows an
example classifier hierarchy window 230 in Figure 2. Spec. ,r 29. This
figure is reproduced below.
7
Appeal2017-011054
Application 13/827,688
2f}Q
244
25{)
)
r17.,A ...------.....-------;i. ---j,----e,---->
Lv'U''\J.. W5~fflffi HPARCW 242 ( NtW u.AS)flffi
1 IP ~1'\:r l LJ.t
""ll"' T'' UA ~,\ A
)
Lt1TER TC JOt J 213
-
~7~~11~----1.,1212 1'1}U !
I
·~r251
~:Mf., ··.1 ,ir,,k., ---
0
0t;?
[I)(RiPTIGrt ··········· .. Ir~,..,,_
Dr?G~ "-'V'V P~~-AM~JQ( , :·,:1-t~a.~. C..~ ..... ,
,51210
5214
Classifier-hierarchy window 230, shown above, includes street
classifier 232 and zip-code classifier 233, which are sub-classifiers to
address classifier 231. Spec. ,r 32. Although the claim is not limited to this
example, it nevertheless informs our analysis of whether the Examiner's
interpretation of the claimed "hierarchy of a plurality of classifiers" is
reasonable.
Although the Examiner has identified some hierarchy---e.g., a textual
hierarchy-we agree with Appellants that this is not a classifier hierarchy, as
8
Appeal2017-011054
Application 13/827,688
required by the claim. Br. 12. That is, Wang segments an electronic
document, and the document's text content includes words, phrases,
sentences, and paragraphs. Wang ,r 70. In this way, Wang's text content has
"several logical parts (or sections) with a textual hierarchy." Id. To the
extent that Wang's textual hierarchy indicates that some classifiers are
present (Ans. 5), the claim, nevertheless, requires displaying those classifiers
in their own hierarchy. But here, the Examiner has not shown that Wang
teaches or suggests displaying a separate classifier hierarchy. Br. 12.
Accordingly, we do not sustain the Examiner's rejection of claim 2,
and the rejection of claim 12, which also recites a similar limitation-i.e.,
code configured to display the classifier hierarchy. For the same reason, we
also do not sustain the Examiner's rejection of claims 3, 4, 6, 13, 14, and 16,
which depend from claims 2 and 12.
CLAIMS 5 AND 15
Claim 5 recites, in part, "displaying a list of classifiers that each have
a parameter satisfied by text in the document." Claim 15 recites a similar
limitation: computer-readable program code configured to perform the
display.
The Examiner finds that Wang displays the recited classifier list.
Final Act. 13-14 (citing Wang ,r,r 62, 178). According to the Examiner,
Wang's user labels document segments with a concept from a concept-
classifier list. Final Act. 13. The Examiner also explains that Wang's
classifier list is shown in a hierarchical document, and in this way, Wang
displays a classifier hierarchy. Ans. 8.
9
Appeal2017-011054
Application 13/827,688
Appellants argue that Wang does not display a list of classifiers.
Br. 13. According to Appellants, Wang only displays a user-selected portion
of a document. Id. We agree.
Wang uses concept classifiers to associate a user-selected document
segment with a concept. Wang ,r 176. The user can then search the text to
see which parts belong to a particular concept. Id. ,r 175. For example, the
user can search for the concept "Payment Terms," and the system will return
the corresponding document segments with that label. Id. ,r 177.
To be sure, Wang teaches that a reviewer can label a document
segment as belonging to a particular concept. Id. ,r 178, cited in
Final Act. 13. We, however, agree with Appellants that Wang's list does not
contain the classifiers having a satisfied parameter. Br. 13. Rather, Wang
uses these labels to train the classifiers. Wang ,r 178. After a sufficient
number of segments are labeled, Wang trains a concept classifier to
associate additional segments with the concept. Id. So, to the extent Wang
teaches some list, the Examiner has not shown the particular classifier list
recited in claims 5 and 15. Accord Br. 13.
As discussed above, although the Examiner has identified some
hierarchy----e.g., a textual hierarchy (Ans. 8}-this is neither a classifier
hierarchy, nor a list, as required by the claim. See Wang ,r 13, Fig. 4B, cited
in Ans. 8.
Accordingly, we do not sustain the Examiner's rejection of claims 5
and 15.
10
Appeal2017-011054
Application 13/827,688
CONCLUSION
We sustain the Examiner's rejection of claims 1, 7-11, 17-19, and 21.
We do not sustain the Examiner's rejection of claims 2-6 and 12-16.
DECISION
We affirm-in-part the Examiner's rejection of claims 1-19 and 21.
No time period for taking any subsequent action in connection with
this appeal may be extended under 3 7 C.F .R. § 1.13 6( a)( 1 )(iv )(2016).
AFFIRMED-IN-PART
11