Ex Parte Stoke et al

Patent Trial and Appeal Board

Jul 30, 2018

13708152 (P.T.A.B. Jul. 30, 2018)

UNITED STA TES p A TENT AND TRADEMARK OFFICE
APPLICATION NO. FILING DATE
13/708, 152 12/07/2012
26212 7590 08/01/2018
FISH & RICHARDSON P.C. (ACCENTURE)
P.O. BOX 1022
MINNEAPOLIS, MN 55440-1022
FIRST NAMED INVENTOR
Mark Peter Stoke
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.
12587-0099002 I
22139US3
CONFIRMATION NO.
9678
EXAMINER
SWARTZ, STEPHENS
ART UNIT PAPER NUMBER
3623
NOTIFICATION DATE DELIVERY MODE
08/01/2018 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):
P ATDOCTC@fr.com
PTOL-90A (Rev. 04/07)
UNITED STATES PATENT AND TRADEMARK OFFICE
BEFORE THE PATENT TRIAL AND APPEAL BOARD
Ex parte MARK PETER STOKE
and
CARL WARD
Appeal2017-004527
Application 13/708, 152 1
Technology Center 3600
Before CARLA M. KRIVAK, HUNG H. BUI, and JON M. JURGOV AN,
Administrative Patent Judges.
JURGOV AN, Administrative Patent Judge.
DECISION ON APPEAL
Appellants seek review under 35 U.S.C. § 134(a) from a Final
Rejection of claims 1, 5-7, 11-16, and 20-23, which are all the claims
pending in the application. We have jurisdiction under 35 U.S.C. § 6(b).
We affirm. 2
1 Appellants identify Accenture Global Services Limited, as the real party in
interest. (App. Br. 1.)
2 Our Decision refers to the Specification ("Spec.") filed December 7, 2012,
the Final Office Action ("Final Act.") mailed March 30, 2016, the Appeal
Brief ("App. Br.") filed September 23, 2016, the Examiner's Answer
("Ans.") mailed December 16, 2016, and the Reply Brief ("Reply Br.") filed
Appeal2017-004527
Application 13/708,152
CLAIMED INVENTION
The claims are directed to a method and system for "conducting an
assessment of the risk that [client's tax] data is either incomplete or
incorrect, and deciding [a] future action as a result of the outcome of the
assessment." (Spec. 1:15-17.) Appellants' invention "generates risk
assessment data," "receives the risk assessment data," and "selects
appropriate actions for subsequent processing of the client [tax] data
according to the assessment of risk." (Abstract.) Appropriate actions
include "[personalizing] any online interaction ... to force high risk clients
... to provide additional data that others are generally not required to
provide," and "having more checks applied [to high risk cases] throughout [a
tax preparation] process whilst low risk cases will generally proceed with
fewer checks." (Spec. 15:5-9, 15:16-18, 15:31-33; Abstract.)
Claims 1, 7, and 16 are independent. Claim 1, reproduced below, is
illustrative of the claimed subject matter:
1. A system comprising:
one or more computers and one or more storage devices
storing instructions that are operable, when executed by the one
or more computers, to cause the one or more computers to
perform operations comprising:
receiving historical tax return data that includes, for
each of a plurality of users, data indicating whether a respective
user submitted an improper tax return and data included in the
respective user's respective tax return;
training, using the data indicating whether the
respective user submitted an improper tax return and the data
included in the respective user's respective tax return as training
data, a neural network to predict a given user's predictive risk
January 17, 2017.
2
Appeal2017-004527
Application 13/708,152
score for the given user's likelihood to submit an improper tax
return based at least in part on data associated with the given
user's tax return;
generating, using the trained neural network, a
predictive risk score for a user, wherein the predictive risk score
reflects the user's predicted risk for submitting an improper tax
return;
after generating the predictive risk score, providing,
for output, a first interface including one or more initial fields of
an interactive tax return form, receiving one or more initial
values entered by the user through the first interface in the one
or more initial fields of the interactive tax return form;
before submitting the tax return for the user:
generating, using the neural network, a first
modified, predictive risk score based at least on one or more of
the initial values;
determining, based on comparing the first
modified predictive risk score to a threshold risk score, that the
user is classified as having a heightened risk for submitting an
improper tax return;
based on determining that the user is
classified as having heightened risk for submitting an improper
tax return, providing, for output, a second interface including one
or more additional fields of the interactive tax return form,
wherein the one or more additional fields of the interactive tax
return form are not provided to users that are not classified as
having a heightened risk for submitting an improper tax return
receiving one or more additional values
entered by the user through the second interface in the one or
more additional interactive tax return form fields; and
generating, using the neural network, a second
modified predictive risk score based at least on one or more of
the initial values and one or more of the additional values; and
submitting the tax return for further processing by a
tax return process that is associated with the second modified
predictive risk score.
3
Appeal2017-004527
Application 13/708,152
(App. Br. 19--20 (Claims App'x).)
REJECTIONS & REFERENCES
( 1) Claims 1, 5-7, 11-16, and 20-23 stand rejected under
35 U.S.C. § 101 as directed to non-statutory subject matter. (Final Act. 2-
7.)
(2) Claims 1, 6, 7, 12-16, 21, and 23 stand rejected under
35 U.S.C. § I03(a) based on Walker et al. (US 2001/0044734 Al, published
Nov. 22, 2001) ("Walker"), TurboTax (Taxes Made Easy. Taxes Done Right,
Internet Archive Wayback Machine, available at, http://turbotax.intuit.com,
1-5 (2005)) ("TurboTax"), and Fischthal (US 5,822,741, issued Oct. 13,
1998). 3 (Final Act. 8-21.)
(3) Claims 5, 11, 20, and 22 stand rejected under
35 U.S.C. § I03(a) based on Walker, TurboTax, Fischthal, and Quinn et al.
(US 7,860,763 Bl, issued Dec. 28, 2010) ("Quinn"). (Final Act. 21-24.)
ANALYSIS
Rejection of Claims 1, 5-7, 11-16, and 20-23
under 35 US.C. § 101
Patent eligibility is a question of law that is reviewable de nova.
Dealertrack, Inc. v. Huber, 674 F.3d 1315, 1333 (Fed. Cir. 2012).
Patentable subject matter is defined by 35 U.S.C. § 101 as follows:
[ w ]hoever invents or discovers any new and useful process,
machine, manufacture, or composition of matter, or any new
3 The TurboTax document has five consecutive pages. Although these
pages are not consecutively numbered inside the document, the Examiner
refers to TurboTax's five consecutive pages as pages 1-5; we do the same.
(See Final Act. 8.)
4
Appeal2017-004527
Application 13/708,152
and useful improvement thereof, may obtain a patent therefor,
subject to the conditions and requirements of this title.
In interpreting this statute, the Supreme Court emphasizes that patent
protection should not preempt "the basic tools of scientific and technological
work." Gottschalk v. Benson, 409 U.S. 63, 67 (1972); Mayo Collaborative
Servs. v. Prometheus Labs., Inc., 566 U.S. 66, 71 (2012); Alice Corp. Pty.
Ltd. v. CLS Bankint'l, 134 S. Ct. 2347, 2354 (2014). The rationale is that
patents directed to basic building blocks of technology would not "promote
the progress of science" under the U.S. Constitution, Article I, Section 8,
Clause 8, but instead would impede it. Accordingly, laws of nature, natural
phenomena, and abstract ideas are not patent-eligible subject matter. Thales
Visionix Inc. v. US., 850 F.3d 1343, 1346 (Fed. Cir. 2017) (citing Alice, 134
S. Ct. at 2354).
The Supreme Court set forth a two-part test for subject matter
eligibility in Alice. Alice, at 2355. The first step is to determine whether the
claim is directed to a patent-ineligible concept. Alice, at 2355 ( citing Mayo,
566 U.S. at 76-77). If so, then the eligibility analysis proceeds to the second
step of the Alice/Mayo test, in which we "examine the elements of the claim
to determine whether it contains an 'inventive concept' sufficient to
'transform' the claimed abstract idea into a patent-eligible application."
Alice, at 2357 (quoting Mayo, at 72, 79). The "'inventive concept"' may be
embodied in one or more of the individual claim limitations or in the ordered
combination of the limitations. Alice, at 2355. The "inventive concept"
must be significantly more than the abstract idea itself, and cannot be simply
an instruction to implement or apply the abstract idea on a computer. Alice,
at 2358. "'[W]ell-understood, routine, [and] conventional activit[ies]'
5
Appeal2017-004527
Application 13/708,152
previously known to the industry" are insufficient to transform an abstract
idea into patent-eligible subject matter. Alice, at 2359 ( citing Mayo, at 73).
Turning to the first part of the Alice/ Mayo analysis, the Examiner
finds claim 1 is directed to "the abstract idea of analyzing information
provided by users to determine whether the data entered" is correct and "to
determine [whether] tax preparation is correct," which is a fundamental
economic practice that can also "'be performed in the human mind, or by a
human using a pen and paper."' (Final Act. 3--4.) The Examiner also finds
claim 1 is directed to an abstract idea of "using categories to organize, store
and transmit information," "comparing new and stored information and
using rules to identify options," and "collecting data, recognizing certain
data within the collected data set, and storing that recognized data," which is
analogous or similar to the abstract ideas of gathering and organizing
information discussed in Cybeifone, SmartGene, Content Extraction, and
Electric Power Group. (Final Act. 4 ( citing Cybeifone Sys., LLC v. CNN
Interactive Grp., Inc., 558 F. App'x 988 (Fed. Cir. 2014); SmartGene, Inc. v.
Advanced Biological Labs., SA, 852 F. Supp. 2d 42 (D.D.C. 2012), ajf'd 555
F. App'x 950 (Fed. Cir. 2014); Content Extraction & Transmission LLC v.
Wells Fargo Bank, Nat'! Ass 'n, 776 F.3d 1343 (Fed. Cir. 2014)); Ans. 10-11
(citing Electric Power Grp, LLC v. Alstom S.A., 830 F.3d 1350 (Fed. Cir.
2016)).)
Appellants contend the Examiner erred in rejecting claim 1, and
similarly claims 7 and 16, as directed to non-statutory subject matter because
the claims are not directed to an abstract idea, and the Examiner
overgeneralized the claims and "ignored various claim features." (App. Br.
14, 17-18.) However, the Examiner concludes, and we agree, claim 1 is
6
Appeal2017-004527
Application 13/708,152
abstract because it is directed to a fundamental economic business practice
of checking tax filing information to ensure tax documents are correct,
helping tax filers avoid audits by revenue agencies ( e.g., IRS), and helping
revenue agencies identify tax reporting errors. (Final Act. 3--4.)
The Examiner's finding is supported by Appellants' Specification,
which provides "[ a ]n embodiment of the invention continuously predicts
compliance risk for each taxpayer and such a risk assessment may be used to
intervene proactively with taxpayers to avoid lodgement of a non complying
return," thus "requir[ing] less effort for the tax payer to lodge a compliant
return." (See Spec. 14:29-15:2.) The Specification further describes an
"embodiment of the invention" in which "[a]s the revenue agency processes
each return form[,] it may calculate interaction risk scores" "using the same
risk model that is used for calculating predictive risk scores," further
"enabl[ing] the tax agency to set hidden fields within the [electronic] form
that provide an indication if a risk condition has been reached." (See Spec.
26:3-7, 28:1-7.) Similar abstract ideas long prevalent in our system of
commerce, for implementing tasks to mitigate risks in financial and
transactional practices, have been identified in Alice, Bilski, buySAFE,
CyberSource, and Accenture Global Services. See Alice, at 2356-57
(intermediated settlement of traded or exchanged financial obligations to
mitigate the risk that one party will not perform); Bilski v. Kappas, 561 U.S.
593, 599, 611 (2010) (risk hedging); buySAFE, Inc. v. Google, Inc., 765 F.3d
1350, 1355 (Fed. Cir. 2014) (guaranteeing a party's performance of its
online transaction); CyberSource Corp. v. Retail Decisions, Inc., 654 F.3d
1366, 1370 (Fed. Cir. 2011) (verifying the validity of a credit card
transaction over the Internet); and Accenture Global Servs., GmbH v.
7
Appeal2017-004527
Application 13/708,152
Guidewire Software, Inc., 728 F.3d 1336, 1344--45 (Fed. Cir. 2013) (claims
reciting "generalized software components arranged to implement an
abstract concept [ of generating insurance-policy-related tasks based on rules
to be completed upon the occurrence of an event, e.g., for processing an
insurance claim] on a computer" not patent eligible).
As further recognized by the Examiner, Appellants' claim 1 recites
generic data manipulation steps of receiving, comparing, and organizing
data. (Final Act. 4; Ans. 10-11.) As the Specification itself observes, "[t]he
system and method of the present invention is particularly useful for
receiving data from clients, conducting an assessment of the risk that the
data is either incomplete or incorrect, and deciding future action as a result
of the outcome of the assessment." (See Spec. 1:14--20.) Such data
manipulation steps are abstract ideas similar to data manipulation techniques
identified in Cyberfone, SmartGene, Digitech, Content Extraction, Electric
Power Group, and Intellectual Ventures. (Final Act. 4; Ans. 10-11; see
SmartGene, Inc. v. Advanced Biological Labs., SA, 852 F. Supp. 2d 42
(D.D.C. 2012), aff'd 555 F. App'x 950 (Fed. Cir. 2014); Cyberfone, 558 F.
App'x at 988,992; Digitech Image Techs., LLC v. Elecs.for Imaging, Inc.,
758 F.3d 1344, 1351 (Fed. Cir. 2014) (employing mathematical algorithms
to manipulate existing information); PerkinElmer, Inc. v Intema Ltd., 496
Fed. Appx. 65 (Fed. Cir. 2012) (comparing data to determine a risk level);
Content Extraction, 776 F.3d at 1347--48 (finding that "[t]he concept of data
collection, recognition, and storage is undisputedly well-known," and
"humans have always performed these functions"); Electric Power Grp.,
830 F.3d at 1353-54 (collecting information and "analyzing information by
steps people go through in their minds, or by mathematical algorithms,
8
Appeal2017-004527
Application 13/708,152
without more, [are] essentially mental processes within the abstract-idea
category"); and Intellectual Ventures I LLC v. Capital One Bank (USA), 792
F.3d 1363, 1370 (Fed. Cir. 2015) (tailoring information presented to a user
based on particular information).)
We further agree with the Examiner the data manipulation steps
recited in claim 1 "can be performed ... by a human using a pen and paper."
(Final Act. 4.) "[A] method that can be performed by human thought alone
is merely an abstract idea and is not patent-eligible under§ 101."
CyberSource, 654 F.3d at 1373. That is, mental processes can be
unpatentable even when automated to reduce the burden on the user of what
once could have been done with pen and paper. See CyberSource, at 1375
("purely mental processes can be unpatentable, even when performed by a
computer, was precisely the holding of the Supreme Court in Gottschalk v.
Benson."). For example, the claimed steps of generating risk scores for a
user before submitting the user's tax return could be performed by a certified
public accountant (CPA) hired to prepare the user's taxes. A CPA could
determine or anticipate a user's possible tax reporting errors by reviewing
the user's tax filing history and previous user's interactions with IRS. The
claimed steps of requesting "initial" and "additional values" ( e.g., tax
information) from the user could be performed by the CPA requesting tax-
related information from the user based on the CPA's understanding of
user's previous reporting errors and oversights.
We recognize Appellants' claim 1 recites a "trained neural network,"
which is a computing system, not a pen and paper application. Appellants,
however, have not demonstrated their claimed neural network is able to
perform functions that are not merely generic for neural networks. See DDR
9
Appeal2017-004527
Application 13/708,152
Holdings, LLC, v. Hotels.com, L.P., 773 F.3d 1245, 1257-58 (Fed. Cir.
2014) (holding the claims at issue patent eligible because "they do not
broadly and generically claim 'use of the Internet' to perform an abstract
business practice (with insignificant added activity)," and "specify how
interactions with the Internet are manipulated to yield a desired result-a
result that overrides the routine and conventional sequence of events
ordinarily triggered by the click of a hyperlink"); see also Bancorp Servs.,
L.L.C. v. Sun Life Assurance Co. of Can. (US.), 687 F.3d 1266, 1278 (Fed.
Cir. 2012) ("[T]he fact that the required calculations could be performed
more efficiently via a computer does not materially alter the patent eligibility
of the claimed subject matter."); Dealertrack, 674 F.3d at 1333-34
("[ s ]imply adding a 'computer aided' limitation to a claim covering an
abstract concept, without more, is insufficient to render [a] claim patent
eligible" (internal citation omitted)).
Appellants further argue claim 1 is similar to the claims in Enfzsh
because "claim 1 recites several features that are tied to [a] specific solution"
that is "directed to using neural networks to identify likely improper tax
returns and only present additional user interfaces to the users who have a
heightened risk for submitting an improper tax return." (App. Br. 14--15
(citing Enfzsh, LLC v. Microsoft Corp., 822 F.3d 1327 (Fed. Cir. 2016)).)
Appellants' Specification, however, does not describe how the claimed
applications for identifying and correcting improper tax returns effect
specific improvements to the computers, neural networks, or user interfaces,
or to the way such systems operate. (Ans. 10, 12.) See Enfzsh, 822 F.3d at
1336. Additionally, Appellants do not present evidence to establish their
claim recites a specific improvement to the computers, or an improvement in
10
Appeal2017-004527
Application 13/708,152
the technical functioning of a neural network system. See Enfish, at 1336.
Appellants also have not demonstrated their claim "improve[ s] the way a
computer stores and retrieves data in memory," as the claims in Enfzsh did
via a "self-referential table for a computer database." See Enfzsh, at 1327,
1336. Additionally, Appellants have not demonstrated their claim provides
a "solution ... necessarily rooted in computer technology in order to
overcome a problem specifically arising in the realm of computer networks,"
as explained by the Federal Circuit in DDR, 773 F.3d at 1257, or an
"unconventional technological solution ... to a technological problem" that
"improve[ s] the performance of the system itself," as explained in Amdocs
(Israel) Ltd. v. Openet Telecom, Inc., 841 F.3d 1288, 1300, 1302 (Fed. Cir.
2016).
We are also unpersuaded by Appellants' argument that "claim 1 is
patent-eligible" because the claimed computer "perform[ s] a distinct process
to automate a task" that allows for specific advantages and improvements
realized. (App. Br. 15-16.) Particularly, Appellants argue claim 1
"improves the state of the art using a 'specific process for automatically'
using neural networks to analyze tax returns in real time" and automatically
"predicts the probability a taxpayer will be non-compliant based on the data
available at the time the prediction is to be made." (App. Br. 15-16 (citing
Spec. 4:24--5:20, 21: 19-22).) Appellants' arguments are unpersuasive
because Appellants are reading limitations from the Specification into the
claims. Although claims are interpreted in light of the Specification,
limitations from the Specification are not read into the claims. In re Van
Geuns, 988 F.2d 1181, 1184 (Fed. Cir. 1993). Here, claim 1 does not recite
or effect the advocated improvements of identifying or predicting non-
11
Appeal2017-004527
Application 13/708,152
compliant tax submissions to a revenue agency. (See App. Br. 15.) Claim
1 's predictive risk scores are generated "before submitting the tax return,"
and are used for "submitting the tax return for further processing by a tax
return process that is associated with the second modified predictive risk
score." (App. Br. 19-20 (Claims App'x) (emphasis added).) The claimed
"submitting the tax return for further processing by a tax return process"
does not recite a purpose or application of the risk score generating steps,
such as determining whether a filed tax return is non-compliant with a
revenue agency's tax reporting guidelines. In fact, claim 1 does not require
the tax return to be submitted, or even evaluated, by a revenue agency or
other official tax entity.
Further, with respect to Appellants' preemption argument (that
"'claim 1 is limited to a specific process"' and does not preempt "all ...
means of[] determining whether tax preparation is correct," see App. Br.
17), we note the McRO court explicitly "recognized that 'the absence of
complete preemption does not demonstrate patent eligibility."' See McRO,
Inc. v. Bandai Namco Games America Inc., 837 F.3d 1299, 1315 (Fed. Cir.
2016) (quoting Ariosa Diagnostics, Inc. v. Sequenom, Inc., 788 F.3d 1371,
1379 (Fed. Cir. 2015)).
Additionally, as explained supra, we are not persuaded Appellants'
stated advantages and solutions are caused by a specific technological
implementation or by any technical improvement to a computer's operation.
Rather, as recognized by the Examiner, Appellants' claim "utiliz[es] generic
computer equipment." (Ans. 12.) "[T]he use of generic computer elements
like a microprocessor" to perform conventional computer functions does
"not alone transform an otherwise abstract idea into patent-eligible subject
12
Appeal2017-004527
Application 13/708,152
matter." FairWarning IP, LLC v. Iatric Sys., Inc., 839 F.3d 1089, 1096
(Fed. Cir. 2016) (citing DDR Holdings, at 1256).
Accordingly, we agree with the Examiner that independent claims 1,
7, and 16 are directed to an abstract idea and amount to nothing more than
an attempt to patent the abstract idea embodied in the steps of the claims.
See Alice, at 2355 (quoting Mayo, at 78).
Because we agree with the Examiner's analysis and find Appellants'
arguments insufficient to show error, we sustain the rejection of claims 1, 7,
and 16 under 35 U.S.C. § 101. No separate arguments are presented for the
dependent claims 5, 6, 11-15, and 20-23, which fall with independent
claims 1, 7, and 16. 37 C.F.R. § 4I.37(c)(l)(iv). We, therefore, sustain the
rejection under 35 U.S.C. § 101 of claims 5, 6, 11-15, and 20-23.
Rejection of Claims 1, 5-7, 11-16, and 20-23
under 35 US.C. § 103(a)
Claim 1 ( and similarly, claims 7 and 16) recites:
generating, using the trained neural network, a predictive
risk score for a user, wherein the predictive risk score reflects
the user's predicted risk for submitting an improper tax return;
after generating the predictive risk score, providing, for
output, a first interface including one or more initial fields of an
interactive tax return form, receiving one or more initial values
entered by the user through the first interface in the one or more
initial fields of the interactive tax return form;
before submitting the tax return for the user:
generating, using the neural network, a first
modified, predictive risk score based at least on one or more of
the initial values;
determining, based on comparing the first modified
predictive risk score to a threshold risk score, that the user is
classified as having a heightened risk for submitting an improper
tax return;
13
Appeal2017-004527
Application 13/708,152
based on determining that the user is classified as
having heightened risk for submitting an improper tax return,
providing, for output, a second interface including one or more
additional fields of the interactive tax return form, wherein the
one or more additional fields of the interactive tax return form
are not provided to users that are not classified as having a
heightened risk for submitting an improper tax return
receiving one or more additional values entered by
the user through the second interface in the one or more
additional interactive tax return form fields; and
generating, using the neural network, a second modified
predictive risk score based at least on one or more of the initial
values and one or more of the additional values.
(App. Br. 19--20 (Claims App'x) (emphasis added).)
Thus, claims 1, 7, and 16 require a trained neural network to generate
the following three risk scores reflecting a user's predicted risk for
submitting an improper tax return: (1) an initial predictive risk score for the
user, (2) a first modified, predictive risk score based on user's initial value(s)
provided in initial field(s) of an interactive tax return form, and (3) a second
modified predictive risk score based on user's initial value(s) and additional
value(s) requested via additionalfield(s) of the interactive tax return form,
the additional value(s) being requested from the user after "comparing the
first modified predictive risk score to a threshold risk score" and
"determining, based on [the] comparing ... that the user is classified as
having a heightened risk for submitting an improper tax return." (App. Br.
19-20 (Claims App'x).) The additionalfield(s) of the interactive tax return
form "are not provided to users that are not classified as having a heightened
risk for submitting an improper tax return." (App. Br. 19--20.)
With respect to claims 1, 7, and 16, the Examiner finds the
combination of Walker, Fischthal, and Turbo Tax teaches determining a
14
Appeal2017-004527
Application 13/708,152
user's predicted risk for submitting an improper tax return and generating
corresponding predictive risk scores, as claimed. (Final Act. 8-13; Ans. 4--
7.) Particularly, the Examiner finds Walker's paragraph 105 discloses
"based on the audit risk providing additional documents that need to be
addressed (i.e.[,] additional forms)" or "not providing additional documents
(i.e.[,] additional forms)," thereby teaching the claimed "based on
determining that the user is classified as having heightened risk for
submitting an improper tax return, providing, for output, a second interface
including one or more additional fields of the interactive tax return form"
although "the one or more additional fields of the interactive tax return form
are not provided to users that are not classified as having a heightened risk
for submitting an improper tax return." (Final Act. 8-9 ( citing Walker
,r 105) (emphasis omitted).) The Examiner further finds TurboTax teaches
receiving a user's initial values and additional values provided in respective
initial fields and additional fields of an interactive tax return form. (Ans. 5-
6 (citing TurboTax 5); Final Act. 9.) The Examiner also finds Fischthal
teaches a trained neural network generates the claimed predictive risk scores
by determining "a measure of error likelihood (i.e.[,] score) of fraudulent tax
return" and "an iteratively calculated error likelihood (i.e.[,] score)" based
on "additional data inputted." (Final Act. 11 (citing Fischthal col. 5, 11. 5-
15, col. 5, 1. 49-col. 6, 1. 25, col. 7, 11. 29-42, col. 7, 1. 60-col. 8, 1. 33, col. 9,
11. 41-59); see also Ans. 5---6.) The Examiner finds Fischthal's "different
attributes that are summed up are ... [ the claimed] one or more initial
values," the "determined error likelihood of the individual attributes are ...
a first predictive score," and "the addition of the multiple other attributes
15
Appeal2017-004527
Application 13/708,152
contributes to ... a second modified predictive score which may cause an
error in the tax return." (Ans. 7.) We do not agree.
We agree with Appellants that Walker, Fischthal, and TurboTax,
alone or in combination, fail to teach or suggest a "'trained neural network"'
generating a "'second modified predictive risk score"' based on a "'first
modified, predictive risk score"' generated by the same trained neural
network, the "'second modified predictive risk score"' being generated only
for a user classified as having a heightened risk for submitting an improper
tax return, as claimed. (App. Br. 8-10 (emphasis added).) Walker,
Fischthal, and TurboTax also fail to teach or suggest a "trained neural
network" generates the "second modified predictive risk score" based on
additional value( s) requested from and provided by a user after "comparing
the first modified predictive risk score [generated by the trained neural
network] to a threshold risk score" and "determining, based on [the]
comparing ... that the user is classified as having a heightened risk for
submitting an improper tax return." (App. Br. 7-10, 19-20.) Rather,
Fischthal' s "error likelihoods," "sum of the error likelihoods," and "the total
error likelihood for an attribute" of historical data (see Ans. 5, 7) are
parameters that "create a set of conceptually cohesive classes, class 1
through class N" before "a separate and independent neural network is ...
created" for each class. (See Fischthal col. 5, 11. 5-15 and 11. 50-55, col. 6,
11. 18-33; App. Br. 8-9.) In other words, Fischthal's error likelihoods and
total error likelihood are used "to build a series of separate classes" (such as
"classes of taxpayers") from historical data attributes, before a "next task, [in
which] the neural network classifier 30 must be created and trained" to
establish "a neural network ... for each of the respective classes." (See
16
Appeal2017-004527
Application 13/708,152
Fischthal col. 6, 11. 45--46 and 11. 57-59, col. 7, 11. 45-52, col. 8, 11. 62---64.)
Thus, Fischthal's error likelihoods and total error likelihood do not teach a
"second modified predictive risk score" generated by a trained neural
network based on a "first modified, predictive risk score" generated by the
same trained neural network, as claimed. (App. Br. 8-9.)
The Examiner also refers to Fischthal' s iterations as generating the
claimed "second modified predictive risk score." (Final Act. 11.) As
Appellants explain, however, Fischthal's iterations merely create the classes
and train neural networks for respective classes, and do not teach
"' generating ... a second modified predictive risk score' 'using [a] [ trained]
neural network"' as claimed. (App. Br. 9; see Fischthal col. 7, 11. 29--42, col.
7, 1. 60-col. 8, 1. 33.)
The Examiner further cites to Fischthal' s column 9, lines 41-59 as
teaching the use of a trained neural network to measure a user's predicted
risk for submitting an improper tax return. (Final Act. 11 ( citing Fischthal
col. 9, 11. 41-59).) Fischthal, however, discloses a parser "assign[s] the
event [e.g., a particular tax return] to the correct class"-this classification
not using the neural network. (See Fischthal col. 9, 11. 33-35 and 11. 45-51,
col. 10, 11. 16-18; App. Br. 10.) "After the event has been assigned to a
class, a second set of predetermined vectors is provided as input to the
neural network 30 which corresponds to the class," "[t]he neural network
then processes the information" and "provide[ s] an output in the form of an
identification of whether there is or is not likely fraud in the tax return."
(See Fischthal col. 9, 11. 34--40, col. 10, 11. 24--27 (emphases added).) Thus,
Fischthal at most teaches the trained neural network generates one risk score
("an identification of whether there is or is not likely fraud in the tax return,"
17
Appeal2017-004527
Application 13/708,152
see Fischthal col. 10, 11. 25-27), but does not teach "'modified' versions of
'risk scores' that were 'generat[ed] using the trained neural network,' as
recited by claim 1." (App. Br. 10 (emphasis added).)
Fischthal therefore does not teach or suggest the claimed "second
modified predictive risk score" generated by a trained neural network based
on additional values requested from a user "classified as having a heightened
risk for submitting an improper tax return" based on "comparing the [user's]
first modified predictive risk score [generated by the same trained neural
network] to a threshold risk score." (App Br. 9-10.)
Walker does not make up for the above-noted deficiencies of
Fischthal. Walker's online risk evaluator generates one risk score based on
the user's answers to a fixed set of questions. (See Walker ,r,r 104--105, Figs.
6A---6C (questionnaire generates a risk of being audited by the IRS).) For
example, Walker's online risk evaluator requests a list of the user's filed
Schedules and Forms. (See Walker ,r,r 104--105, Figs. 6A---6B.) Walker does
not teach or suggest the claimed "second modified predictive risk score"
generated based on additional value(s) requested from a user after
"comparing [a] first modified predictive risk score to a threshold risk score"
and "'determining, based on [the] comparing ... that the user is classified as
having a heightened risk for submitting an improper tax return."' (App Br.
10, 19-20.) Walker also does not teach "one or more additional fields of the
interactive tax return form are not provided to users that are not classified as
having a heightened risk for submitting an improper tax return," as claimed.
(App. Br. 10.)
TurboTax similarly fails to teach the claimed "second modified
predictive risk score" generated based on additional value(s) requested from
18
Appeal2017-004527
Application 13/708,152
a user after "comparing [a] first modified predictive risk score to a threshold
risk score" and "determining, based on [the] comparing ... that the user is
classified as having a heightened risk for submitting an improper tax return."
(App Br. 9, 19-20.) Rather, TurboTax requests values from its users based
on the fields required by IRS tax forms. (See TurboTax 4 ("The interview
... ask[ s] simple questions in plain English. Your answers are put on the
right IRS and state tax forms," and "[a]s you answer simple tax preparation
questions, Turbo Tax automatically places your answers onto the right IRS-
approved forms").)
TurboTax then "double checks [the user's] return for accuracy" so the
user "[doesn't] have to worry about making a mistake." (See TurboTax 5;
App. Br. 9.) As Appellants explain, "'double check[ing] ... for accuracy'
does not generate a [second] 'predictive risk score' ... 'using [a] neural
network"' and using additional values requested from a user after
"determining ... that the user is classified as having a heightened risk for
submitting an improper tax return." (App. Br. 9, 19-20 (emphasis added).)
The Examiner also has not shown the additional teachings of Quinn make up
for the above-noted deficiencies of Walker, Fischthal, and Turbo Tax.
Thus, for the reasons set forth above, we do not sustain the
Examiner's obviousness rejection of independent claims 1, 7, and 16, and
claims 5, 6, 11-15, and 20-23 dependent therefrom. Because the above-
discussed issues are dispositive as to the obviousness rejections of all claims
on appeal, we do not reach additional issues raised by Appellants' arguments
as to the rejections of dependent claims 12, 14, and 15.
19
Appeal2017-004527
Application 13/708,152
DECISION
The Examiner's rejection of claims 1, 5-7, 11-16, and 20-23 under
35 U.S.C. § 101 is affirmed.
The Examiner's rejection of claims 1, 5-7, 11-16, and 20-23 under
35 U.S.C. § 103(a) is reversed.
Because we have affirmed at least one ground of rejection with
respect to each claim on appeal, we affirm the Examiner's decision rejecting
claims 1, 5-7, 11-16, and 20-23. See 37 C.F.R. § 4I.50(a)(l).
No time period for taking any subsequent action in connection with
this appeal may be extended under 37 C.F.R. § 1.136(a)(l )(iv).
AFFIRMED
20