Li, Huajing et al.

Patent Trials and Appeals Board

Jul 2, 2020

2019003274 (P.T.A.B. Jul. 2, 2020)

UNITED STATES PATENT AND TRADEMARK OFFICE
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
APPLICATION NO. FILING DATE FIRST NAMED INVENTOR ATTORNEY DOCKET NO. CONFIRMATION NO.
13/297,117 11/15/2011 Huajing Li 26295-18983 1921
87851 7590 07/02/2020
Facebook/Fenwick
Silicon Valley Center
801 California Street
Mountain View, CA 94041
EXAMINER
SITTNER, MICHAEL J
ART UNIT PAPER NUMBER
3622
NOTIFICATION DATE DELIVERY MODE
07/02/2020 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):
fwfacebookpatents@fenwick.com
ptoc@fenwick.com
PTOL-90A (Rev. 04/07)
UNITED STATES PATENT AND TRADEMARK OFFICE
BEFORE THE PATENT TRIAL AND APPEAL BOARD
Ex parte HUAJING LI, YANXIN SHI, ROHIT DHAWAN,
RICHARD BILL SIM, RONG YAN, and DAVID DAWEI YE
Appeal 2019-003274
Application 13/297,117
Technology Center 3600
Before BRADLEY W. BAUMEISTER, PHILLIP A. BENNETT, and
IFTIKHAR AHMED, Administrative Patent Judges.
BENNETT, Administrative Patent Judge.
DECISION ON APPEAL
STATEMENT OF THE CASE
Pursuant to 35 U.S.C. § 134(a), Appellant1 appeals from the
Examiner’s decision to reject claims 1–7, 9, 17, 18, and 28–31.2,3 Claims 8,

1 We use the word “Appellant” to refer to “applicant” as defined in
37 C.F.R. § 1.42(a). Appellant identifies the real party in interest as
Facebook, Inc. Appeal Br. 2.
2 We note Appellant failed to include page numbing in both the Appeal Brief
and Reply Brief. As such, our citations to these documents reference them
as though the pages were numbered with the first page as page 1.
3 We further note Appellant filed the wrong claims with the Appeal Brief
and made arguments directed to those wrongly filed claims. The claims
Appeal 2019-003274
Application 13/297,117

2
10–16, and 19–27 are cancelled. We have jurisdiction under
35 U.S.C. § 6(b).
We affirm.
CLAIMED SUBJECT MATTER
The claims are directed to generating clusters of similar users for
advertisement targeting. Claim 1, reproduced below, is illustrative of the
claimed subject matter:
1. A method comprising:
maintaining a plurality of ad objects in a social networking
system corresponding to a plurality of advertisements for display
to users of the social networking system;
maintaining a plurality of user profile objects in a social
networking system, the plurality of user profile objects
associated with a plurality of users of the social networking
system;
receiving a graphical interface displaying a selection of
targeting criteria for an advertisement from an advertiser,
wherein the targeting criteria of the advertisement indicate one
or more actions performed by users in the social networking
system;
transmitting the advertisement for presentation to viewing
users in a targeting cluster of users that interact with the social
networking system, the targeting cluster of users selected by:
for each user of a first subset of users of the social
networking system,

included with the Appeal Brief included non-entered amendments from
Appellant’s after-final response. Our references to the claims herein are
made to the claims that are currently of record—the claims as amended in
the Office Action response captioned AMENDMENT C, filed by Appellant
on December 18, 2017 (“OA Response”).
Appeal 2019-003274
Application 13/297,117

3
assigning the user to a bin of a set of bins according
on a past engagement history of the user with advertisers;
comparing the features defined by a user model with
features for the user;
computing a confidence score based on a number of
features for the user that match the features defined by the
user model, the confidence score proportional to the
number of matching features; and
selecting from each bin of the set of bins a
predetermined number of users having highest confidence
scores for the targeting cluster of users;
wherein the user model is generated by:
determining, by a processor, a training cluster of users for
the advertisement based on the selection of the targeting criteria,
comprising,
selecting as the training cluster of users those users
of a second subset of users of the social networking system
that have user profile objects indicating a threshold
number of actions performed that match the actions
indicated by the targeting criteria, the second subset of
users of the social networking system distinct from the
first subset of users of the social networking system;
generating the user model from the training cluster
of users, the user model representative of the training
cluster of users and including features of the users of the
training cluster of users that are common to the users of
the training cluster of users and which are extracted from
the user profile objects of the users of the training cluster
the generation of the user model comprising:
selecting features for the user model to
include selected social graph features, each social
graph feature indicating one or more connections to
users in the training set of users, the connections
stored in a social graph of the social networking
system, such that users that match a social graph
feature have connections to the users in the training
Appeal 2019-003274
Application 13/297,117

4
set of users matching the connections indicated in
the social graph feature;
training an optimizer model with a training
set of data, the optimizer model using a machine
learning algorithm, the input features of the training
set of data including the features for historical
viewing users previously presented with the
advertisement, the output labels of the training set
of data including a conversion rate of the historical
viewing users, the optimizer model generating
weightings for each of the input features based on
an effect of each input feature on the conversion
rate; and
removing features from the user model
matching the input features of the optimizer model
that have corresponding weightings generated by
the optimizer model that are below a threshold
value.
OA Response 2–5.
REFERENCES
The prior art relied upon by the Examiner is:
Name Reference Date
Bentolila US 8,046,797 B2 Oct. 25, 2011
Duan US 2006/0224532 A1 Oct. 5, 2006
Kendall US 2009/0119167 A1 May 7, 2009
Elvekrog US 2011/0153423 A1 June 23, 2011
Bagherjeiran US 2012/0054040 A1 Mar. 1, 2012
REJECTIONS
Claims 1–7, 9, 17, 18, and 28–31 stand rejected under 35 U.S.C.
§ 112, second paragraph, as being indefinite. Final Act. 2–4.
Appeal 2019-003274
Application 13/297,117

5
Claims 1–7, 9, 17, 18, and 28–31 stand rejected under 35 U.S.C. § 101
as being directed to ineligible subject matter. Final Act. 4–12.
Claims 1–7, 9, 17, 18, and 28 stand rejected under 35 U.S.C. § 103(a)
as being unpatentable over Kendall, Bentolila, Bagherjeiran, and Duan.
Final Act. 12–28.
Claims 29–31 stand rejected under 35 U.S.C. § 103(a) as being
unpatentable over Kendall, Bentolila, Bagherjeiran, Duan, and Elvekrog.
Final Act. 28–31.
REJECTION UNDER 35 U.S.C. § 112
The Examiner rejects the claims under 35 U.S.C. § 112, second
paragraph, as being indefinite. Final Act. 2–4. Prior to filing the Appeal
Brief, Appellant submitted proposed amendments seeking to address certain
aspects of the indefiniteness rejection. The amendments were not entered by
the Examiner. See Aug. 10, 2018 Advisory Act. 1–2. Despite the non-entry
of the amendments, Appellant filed the amended claims together with the
Appeal Brief. Appeal Br. 19–25.
Appellant argues against the indefiniteness rejection with reference to
the claims as they appear in the erroneous claims appendix, and not to the
claims as they actually stand in this record. Compare Appeal Br. 19–25
(claims appendix) with OA Response 2–7 (Listing of Claims). Because
Appellant’s arguments do not address the actual language of the claims of
record, they do not substantively address the rejection made. See Ans. 3–7.
Accordingly, we summarily sustain the indefiniteness rejection of claims 1–
7, 9, 17, 18, and 28–31 under 35 U.S.C. § 112, second paragraph.
Appeal 2019-003274
Application 13/297,117

6
REJECTION UNDER 35 U.S.C. § 101
Standard for Patent Eligibility
In issues involving subject matter eligibility, our inquiry focuses on
whether the claims satisfy the two-step test set forth by the Supreme Court in
Alice Corp. v. CLS Bank Int’l, 573 U.S. 208 (2014). The Court instructs us
to “first determine whether the claims at issue are directed to a patent-
ineligible concept,” id. at 218, and, in this case, the inquiry centers on
whether the claims are directed to an abstract idea. If the initial threshold is
met, we then move to the second step, in which we “consider the elements of
each claim both individually and ‘as an ordered combination’ to determine
whether the additional elements ‘transform the nature of the claim’ into a
patent-eligible application.” Id. at 217–18 (quoting Mayo Collaborative
Servs. v. Prometheus Labs., Inc., 566 U.S. 66, 79, 78 (2012)). The Court
describes the second step as a search for “an ‘“inventive concept”’—i.e., an
element or combination of elements that is ‘sufficient to ensure that the
patent in practice amounts to significantly more than a patent upon the
[ineligible concept] itself.’” Id. (quoting Mayo, 566 U.S. at 72–73).
The USPTO has published revised guidance on the application of
§ 101 consistent with Alice and subsequent Federal Circuit decisions.
USPTO, 2019 Revised Patent Subject Matter Eligibility Guidance, 84 Fed.
Reg. 50 (Jan. 7, 2019) (“Guidance”), updated by USPTO, October 2019
Update: Subject Matter Eligibility (available at https://www.uspto.gov/sites/
default/files/documents/peg_oct_2019_update.pdf) (“October 2019
Guidance Update”).
Appeal 2019-003274
Application 13/297,117

7
Under the Guidance, we first look to whether the claim recites:
(1) any judicial exceptions, including certain groupings of
abstract ideas (i.e., mathematical concepts, certain methods of
organizing human activity such as a fundamental economic
practice, or mental processes) (referred to Step 2A, prong 1 in
the Guidance); and
(2) additional elements that integrate the judicial exception into
a practical application (see MPEP § 2106.05(a)–(c), (e)–(h))
(referred to Step 2A, prong 2 in the Guidance).
See Guidance, 84 Fed. Reg. at 52–55. Only if a claim (1) recites a judicial
exception and (2) does not integrate that exception into a practical
application, do we then move to Step 2B of the Guidance. There, we look to
whether the claim:
(3) adds a specific limitation beyond the judicial exception that
is not “well-understood, routine, conventional” in the field (see
MPEP § 2106.05(d)); or
(4) simply appends well-understood, routine, conventional
activities previously known to the industry, specified at a high
level of generality, to the judicial exception.
See Guidance, 84 Fed. Reg. at 56.
Examiner’s Findings and Conclusions4
The Examiner rejects claim 1 as being directed to a judicial exception
without significantly more under Alice. Final Act. 4–12; Ans. 7–16. Under
the first step of the Alice inquiry, the Examiner determines claim 1 recites a

4 The Final Office Action was mailed prior to the Guidance. The Final
Office Action applied the case-law-based approach from previous eligibility
guidance in rejecting the claims under § 101. Some of the documents in this
appeal, including the Examiner’s Answer and Reply Brief, were filed
subsequent to the issuance of the Guidance. The Examiner’s Answer
updated the rejection to incorporate the procedure set forth in the Guidance,
and we refer primarily to the arguments as updated by those findings and
conclusions in the Answer.
Appeal 2019-003274
Application 13/297,117

8
judicial exception. Ans. 4–5. Specifically, the Examiner determines that
“the claims at issue are directed . . . towards the idea of targeted advertising,
which is a basic economic practice,” and the Examiner identifies the
limitations reciting the judicial exception. Ans. 9. The Examiner
determines the bulk of the remaining limitations in the claim relate to the use
of machine learning to implement the abstract idea. Ans. 9–10.
The Examiner determines, “as is evident from Appellant’s original
disclosure, no new or novel machine learning techniques are the subject of
Appellant’s invention.” Ans. 10 (citing Spec. ¶¶ 15, 38, 41). The Examiner
also concludes,
[These] steps appear to be directed simply towards applying
well-known machine learning techniques in order to build and
train a model with data from a training cluster of uses (the model
called by the Appellant an “optimizer model”) which may be
trained from historical data of users in the training cluster of
users who have already seen and interacted with the
advertisement, which is typical in machine learning.
Ans. 10.
The Examiner further supports the rejection by characterizing claim 1
as a process for collecting data, analyzing data, and displaying the results of
that analysis—similar to claims found ineligible in prior cases. Ans. 11–12.
Under Alice step 2, the Examiner determines that the claims do not
amount to significantly more than the abstract idea because:
[T]here is no indication in the record that any specialized
computer hardware or other “inventive” computer components
are required. In fact, the Specification explicitly discloses that
the claimed invention is implemented using conventional
computer system and standard communication technologies and
protocols (see, e.g., Spec[. ]¶0021). Furthermore, claim 1 merely
employs generic computer components (i.e. a “processor” and a
“social networking system”) to perform generic computer
Appeal 2019-003274
Application 13/297,117

9
functions (i.e., maintaining, receiving, determining, selecting),
which is not enough to transform an abstract idea into a patent
eligible invention.
Ans. 12–13.
The Examiner addresses the machine learning limitations, stating that
they “are generally directed towards the implementation of machine learning
(which is not Appellant’s invention as noted supra) for the purpose of
collection and analysis of data used to select the users to whom the ad will
be targeted which is not significantly more than the abstract idea of targeted
advertising itself.” Ans. 13. The Examiner further cites TLI
Communications LLC v. A.V. Automotive, LLC, 823 F.3d 607 (Fed. Cir.
2016) as evidence in support of the Alice step 2 determination. Ans. 14.
Appellant’s Contentions
Appellant presents several arguments for eligibility. Because the
Guidance was issued after the Appeal Brief but before the Reply Brief, the
arguments made in the Reply Brief focus on the application of the Guidance
to the claims—and we primarily address those arguments herein. Appellant
first argues the Examiner fails to identify the grouping under the Guidance
to which the abstract idea belongs. Reply Br. 1. Appellant further argues
the Examiner’s analysis of “machine learning” is flawed because “this
abstract idea does not meet Prong One because ‘applying machine learning’
does not recite any of the judicial exceptions enumerated in the 2019 PEG.”
Reply Br. 3.
Appellant further argues the Examiner errs under Step 2A, prong 2,
because “[t]he claimed invention integrates ‘targeting advertisements’ into a
practical application, and as such, the claims are not directed to this alleged
abstract idea.” Reply Br. 4. Specifically, Appellant argues,
Appeal 2019-003274
Application 13/297,117

10
When considered as a whole, the claims are directed to a very
specific computer implemented process of receiving targeting
criteria of an advertisement from an advertiser and presenting the
advertisement to a specific targeting cluster of users, where each
user in the targeting cluster is selected based on a comparison of
features associated with the user and features defined by a user
model for the specified targeting criteria. Further, the user model
is trained to select the features for defining the targeting cluster
using a training cluster of users selected based on actions
performed by the training cluster of users that match the
advertiser specified targeting criteria.
Reply Br. 4.
Appellant also asserts the invention, as claimed, “is a practical
application because the claims describe an improved technical solution for
delivering an advertisement to a target group that is likely to be interested in
the advertisement.” Reply Br. 5. Appellant also argues, “the claims use
machine learning in specific ways recited through the various steps to
improve a different technical field, namely online advertising and targeting
content to users.” Id.
Under Step 2B, Appellant contends the Examiner failed to address of
the additional limitations by “gloss[ing] over all of the remaining specific
limitations of the claim and only mentions the hardware elements in the
claim, concluding it to be generic.” Appeal Br. 9. Appellant also argues the
Examiner failed to meet the fact-finding requirements outlined in the
Berkheimer Memorandum. Appeal Br. 10–11; see USPTO, Changes in
Examination Procedure Pertaining to Subject Matter Eligibility, Recent
Subject Matter Eligibility Decision (Berkheimer v. HP, Inc.), at 3–4,
Apr. 19, 2018..

Appeal 2019-003274
Application 13/297,117

11
Revised Guidance, Step 2A, Prong One5
The Judicial Exception
Applying the Guidance, we are not persuaded the Examiner has erred
in determining that the claims recite a judicial exception to patent eligible
subject matter. The Guidance identifies three judicially-excepted groupings:
(1) mathematical concepts, (2) certain methods of organizing human activity
such as fundamental economic practices and commercial interactions
(including . . . advertising, marketing or sales activities or behaviors;
business relations), and (3) mental processes. We focus our analysis on the
second and third groupings—certain methods of organizing human activity
and mental processes.6
We conclude the limitations of claim 1 recite a process of delivering
advertisement to selected recipients who are targeted according to a
behavioral-based model, which amounts to a combination of abstract ideas
under the Guidance.7 For example, claim 1 recites (1) “receiving . . . a

5 Throughout this opinion, we give the claim limitations the broadest
reasonable interpretation consistent with the Specification. See In re Morris,
127 F.3d 1048, 1054 (Fed. Cir. 1997).
6 Appellant’s arguments against the § 101 rejection are made to the claims
generally. We treat claim 1 as representative. 37 C.F.R. § 41.37(c)(1)(iv)
(2018) (“When multiple claims subject to the same ground of rejection are
argued as a group or subgroup by Appellant, the Board may select a single
claim from the group or subgroup and may decide the appeal as to the
ground of rejection with respect to the group or subgroup on the basis of the
selected claim alone.”).
7 RecogniCorp, LLC v. Nintendo Co., 855 F.3d 1322, 1327 (Fed. Cir. 2017)
(“Adding one abstract idea . . . to another abstract idea . . . does not render
the claim non-abstract.”); see also FairWarning IP, LLC v. Iatric Sys., Inc.,
839 F.3d 1089, 1093–94 (Fed. Cir. 2016) (patent-ineligible claims were
directed to a combination of abstract ideas).
Appeal 2019-003274
Application 13/297,117

12
selection of targeting criteria for an advertisement from an advertiser,
wherein the targeting criteria of the advertisement indicate one or more
actions performed by users of the social networking system,” and
(2) “transmitting the advertisement for presentation to viewing users in a
targeting cluster of users that interact with the social networking system.”
These limitations recite the familiar commercial practice of receiving
advertisements from advertisers and selecting the intended audience for
receiving the advertisement, and delivering the advertisement to the intended
audience.
Claim 1 also recites (3) “the targeting cluster of users selected by: for
each user of a first subset of users of the social networking system,”
(4) “assigning the user to a bin of a set of bins according on a past
engagement history of the user with advertisers,” (5) “comparing the
features defined by a user model with features for the user,” (6) “computing
a confidence score based on a number of features for the user that match the
features defined by the user model, the confidence score proportional to the
number of matching features,” and (7) “selecting from each bin of the set of
bins a predetermined number of users having highest confidence scores for
the targeting cluster of users.” OA Response 2–5. These limitations recite
the familiar practice of target selection in advertising, not meaningfully
different from choosing customer to receive a mailed advertisement based
on prior purchases made by the users, or selecting a particular television
program on which to run ads because of the demographic characteristics of
its typical audience. These limitation can also reasonably be characterized
as a mental process because they could be performed by a person in their
mind with the aid of pen and paper. See October 2019 Guidance Update at 9
Appeal 2019-003274
Application 13/297,117

13
(“A claim that encompasses a human performing the step(s) mentally with
the aid of a pen and paper recites a mental process”) (emphasis omitted).
Under the Guidance, these limitations recite both a commercial
interaction of advertising (a certain method of organizing human activity)
and a mental process for identifying consumers to target with advertising.
Accordingly, we conclude the claimed process of delivering advertisement
to selected recipients who are targeted according to a behavioral-based
model, as set forth in claim 1, recites judicial exceptions of both a mental
process and of a commercial interaction, which is a certain method of
organizing human activity under the Guidance.8
Revised Guidance, Step 2A, Prong Two
Integration of the Judicial Exception into a Practical Application
Having determined that claim 1 recites a judicial exception, our
analysis under the Guidance turns now to determining whether claim 1
recites any additional elements that integrate the judicial exception into a
practical application. See Guidance, 84 Fed. Reg. at 54–55 (citing MPEP
§ 2106.05(a)–(c), (e)–(h)).

8 As noted above, Appellant argues the Examiner fails to identify in which
grouping the abstract idea belongs. Reply Br. 3. This argument is not
persuasive, as the Examiner identified the limitations that recite the abstract
idea. Ans. 9 (“the claims at issue are directed . . . towards the idea of
targeted advertising, which is a basic economic practice”). Appellant further
argues the Examiner’s analysis of “machine learning” is flawed because
“this abstract idea does not meet Prong One because ‘applying machine
learning’ does not recite any of the judicial exceptions enumerated in the
2019 PEG.” Reply Br. 2. This argument is not persuasive because the
“machine learning” aspects of the claim are addressed under Prong 2 below.

Appeal 2019-003274
Application 13/297,117

14
Under the Guidance, limitations that are indicative of “integration into
a practical application” include:
1. Improvements to the functioning of a computer, or to any other
technology or technical field — see MPEP § 2106.05(a);
2. Applying the judicial exception with, or by use of, a particular
machine — see MPEP § 2106.05(b);
3. Effecting a transformation or reduction of a particular article to
a different state or thing — see MPEP § 2106.05(c); and
4. Applying or using the judicial exception in some other
meaningful way beyond generally linking the use of the judicial
exception to a particular technological environment, such that
the claim as a whole is more than a drafting effort designed to
monopolize the exception — see MPEP § 2106.05(e).
In contrast, limitations that are not indicative of “integration into a
practical application” include:
1. Adding the words “apply it” (or an equivalent) with the judicial
exception, or merely include instructions to implement an
abstract idea on a computer, or merely uses a computer as a tool
to perform an abstract idea — see MPEP § 2106.05(f);
2. Adding insignificant extra-solution activity to the judicial
exception — see MPEP § 2106.05(g); and
3. Generally linking the use of the judicial exception to a
particular technological environment or field of use — see
MPEP § 2106.05(h).
See Guidance, 84 Fed. Reg. at 54–55 (“Prong Two”).
Appeal 2019-003274
Application 13/297,117

15
As shown above, most of the claim limitations in claim 1 recite
abstract ideas. Additional to those abstract limitations, claim 1 recites that
(a) various steps are performed in “a social networking system,”
(b) targeting criteria is received in “a graphical user interface,” and
(c) additional steps are performed “by a processor.” OA Response 2–5. We
conclude that these limitations are insufficient to integrate the recited
judicial exception into a practical application.
Each of these limitations merely recites the use of conventional
computer technology to implement the otherwise abstract process on a
computer and to display the output. It is well-established, however, that the
use of generic technology to implement an abstract idea is insufficient to
integrate it into a practical application. See MPEP 2106.05(f) (explaining
that it is not indicative of integration into a practical application where the
claims “merely include instructions to implement an abstract idea on a
computer, or merely uses a computer as a tool to perform an abstract idea”).
Claim 1 additionally recites (d) “maintaining a plurality of ad objects
in a social networking system corresponding to a plurality of advertisements
for display to users of the social networking system,” and (e) “maintaining a
plurality of user profile objects in a social networking system, the plurality
of user profile objects associated with a plurality of users of the social
networking system.” OA Response 2. These limitations do not integrate the
abstract idea into a practical application because they are extra-solution
activity. Specifically, they recite data gathering and storage operations
which are incidental to the targeted advertising in the claim. MPEP
§ 2106.05(h) (“An example of pre-solution activity is a step of gathering
data for use in a claimed process.”); see also Content Extraction &
Appeal 2019-003274
Application 13/297,117

16
Transmission LLC v. Wells Fargo Bank, Nat'l Ass’n, 776 F.3d 1343, 1347
(Fed. Cir. 2014) (holding ineligible claims “drawn to the abstract idea of
1) collecting data, 2) recognizing certain data within the collected data set,
and 3) storing that recognized data in a memory”).
Claim 1 also recites (f) “wherein the user model is generated by:
determining, by a processor, a training cluster of users for the advertisement
based on the selection of the targeting criteria, comprising,” (g) “selecting as
the training cluster of users those users of a second subset of users of the
social networking system that have user profile objects indicating a
threshold number of actions performed that match the actions indicated by
the targeting criteria, the second subset of users of the social networking
system distinct from the first subset of users of the social networking
system,” (h) “generating the user model from the training cluster of users,
the user model representative of the training cluster of users and including
features of the users of the training cluster of users that are common to the
users of the training cluster of users and which are extracted from the user
profile objects of the users of the training cluster the generation of the user
model comprising,” (i) “selecting features for the user model to include
selected social graph features, each social graph feature indicating one or
more connections to users in the training set of users, the connections stored
in a social graph of the social networking system, such that users that match
a social graph feature have connections to the users in the training set of
users matching the connections indicated in the social graph feature,”
(j) “training an optimizer model with a training set of data, the optimizer
model using a machine learning algorithm, the input features of the training
set of data including the features for historical viewing users previously
Appeal 2019-003274
Application 13/297,117

17
presented with the advertisement, the output labels of the training set of data
including a conversion rate of the historical viewing users, the optimizer
model generating weightings for each of the input features based on an
effect of each input feature on the conversion rate,” and (k) “removing
features from the user model matching the input features of the optimizer
model that have corresponding weightings generated by the optimizer model
that are below a threshold value.” OA Response 3–5.
These limitations generally recite the use of machine learning to
create the user model that serves as the basis for selecting the targets for the
advertisement. We agree with the Examiner that these limitations are
insufficient to integrate the abstract idea into a practical application. As
explained by the Examiner, the machine learning concepts applied in
Appellant’s claims are not novel, and can be reasonably seen as the
conventional application of well-known machine learning concepts to build
and train a model. Ans. 10.
Appellant’s Specification describes the use of machine learning in
general terms, without any specifics about the algorithms employed in
connection with the machine learning. See, e.g., Spec. ¶¶ 15, 38, 41
(describing results of applying machine learning algorithms without
specifics of the algorithms applied). Because of the high-level and general
description, we do not discern in these limitations any improvements to the
functioning of a computer, or to any other technology or technical field.
MPEP § 2106.05(a). Rather, these additional limitations are more akin to
examples “of limitations that the courts have described as merely indicating
a field of use or technological environment in which to apply a judicial
exception.” MPEP § 2106.05(h).
Appeal 2019-003274
Application 13/297,117

18
One example provided in MPEP § 2106.05(h) of merely indicating a
field of use or technological environment is a step of “[i]dentifying the
participants in a process for hedging risk as commodity providers and
commodity consumers. MPEP § 2106.05(h)(ii) (citing Bilski v. Kapps, 561
U.S. 593, 595 (2010)). As the MPEP explains, “limiting the use of the
process to these participants did no more than describe how the abstract idea
of hedging risk could be used in the commodities and energy markets.” Id.
Similarly here, the machine learning limitations are used merely to identify
the targets for the abstract advertising process, and simply recite how targets
are selected for the targeted advertising.9
Limitations (f) and (g), for example, recite the initial steps of
generating a user model: determining an initial group of users to consider
and selecting a more specific group of users to include in the user model
according to keywords relevant to the advertising targeting criteria—such as
user interests. Spec. ¶¶ 30–32. The Specification includes no description
indicating that the selection is anything meaningfully more than applying
routine data processing concepts to further the abstract advertising process
recited in the claim. See MPEP § 2106.05(f) (not indicative of practical
application where computer is used merely as a tool to perform an abstract
idea). Limitations (h), (i), (j), and (k) recite using commonalities among the

9 We further note that although the Examiner did not do so, the machine
learning limitations are recited at a high level and also could be properly
characterized as a mental process because they recite steps that could be
performed by a human in their mind via observation, exercising judgment,
and with the aid of pen and paper. The “learning” aspect of a machine
learning, is quintessentially a process carried out by a human in their mind—
learning. See Guidance, 84 Fed. Reg. at 52.
Appeal 2019-003274
Application 13/297,117

19
selected cluster to generate and train the user model from which advertising
targets are selected. These limitations also utilize a computer as a tool to
generate the user model, and are described at a high level in the
Specification. Spec. ¶¶ 32–38. In particular, the modules described in the
Specification that perform the recited functions are described exclusively in
functional terms, with the description revealing only what the modules do,
but not how they do it. See, e.g., Spec. 34–38 (describing various modules).
As we noted above, Appellant argues the claimed solution “is a
practical application because the claims describe an improved technical
solution for delivering an advertisement to a target group that is likely to be
interested in the advertisement” and also because “the claims use machine
learning in specific ways recited through the various steps to improve a
different technical field, namely online advertising and targeting content to
users.” Reply Br. 5. We disagree.
Appellant’s invention is focused on improving the commercial
practice of targeted advertising using machine learning. Thus, purported
improvement identified by Appellant is not to machine learning, but instead
to targeted advertising and, therefore, is not an improvement to technology.
The improvement provided by these so-called machine learning process
steps improves the abstract idea itself. It is well-established, however, that
improvements in the abstract idea are insufficient to confer eligibility on an
otherwise ineligible claim. SAP Am. Inc. v. InvestPic, LLC, 898 F.3d 1161,
1168 (Fed. Cir. 2018). We conclude claim 1 is directed to a judicial
exception under step 2A, prong 2, of the Guidance.
Appeal 2019-003274
Application 13/297,117

20
The Inventive Concept – Step 2B
Having determined the claim is directed to a judicial exception, we
proceed to evaluating whether claim 1 adds a specific limitation beyond the
judicial exception that is not “well-understood, routine, conventional” in the
field (see MPEP § 2106.05(d)) or simply appends well-understood, routine,
conventional activities previously known to the industry, specified at a high
level of generality, to the judicial exception. See Guidance, 84 Fed. Reg.
at 56.
Our review of the Examiner’s rejection under Step 2B is guided by the
Berkheimer Memorandum, which sets forth what fact finding requirements
are applicable to rejections under § 101. Consistent with the Berkheimer
Memorandum, we agree with the Examiner that claim 1 does not add
specific limitations beyond what is well-understood, routine, and
conventional.
Our analysis focuses largely on the same limitations addressed in
Step 2A, prong 2, above. We agree with the Examiner that the additional
limitations do not supply an inventive concept under Step 2B because the
machine learning algorithms used in the claim are described at a high level
without detail in the Specification. Spec. ¶¶ 15, 38, 41. This lack of detailed
description evidences their well-understood, routine, and conventional
nature.
Under Step 2B, Appellant contends the Examiner failed to address the
additional limitations by “gloss[ing] over all of the remaining specific
limitations of the claim and only mentions the hardware elements in the
claim, concluding it to be generic.” Appeal Br. 9. We disagree. The
Examiner addressed the remaining limitations in detail in the Answer. See,
Appeal 2019-003274
Application 13/297,117

21
e.g., Ans. 12–14 (addressing the remaining limitations and finding them to
not recite significantly more than the abstract idea itself).
Appellant also argues the Examiner failed to meet the fact-finding
requirements outlined in the Berkheimer Memorandum. Appeal Br. 10–11.
However, consistent with the Berkheimer Memorandum, the Examiner cited
a case from MPEP § 2106.05(d)—TLI Comminications LLC v. A.V.
Automotive, LLC, as evidence of the well-understood, routine, and
conventional nature of the additional limitations.
Because the Examiner correctly concluded claim 1 is directed to a
judicial exception, and because Appellant does not identify any error in the
Examiner’s determination under step 2B of the Guidance, we sustain the
rejection of representative claim 1 under 35 U.S.C. § 101, as well as of the
remaining claims.
REJECTIONS UNDER 35 U.S.C. § 103
We reverse the prior art rejection. In rejecting claim 1 as obvious, the
Examiner relies on Kendall as generally teaching targeted advertising in a
social networking system. Final Act. 13–14. The Examiner acknowledges
that Kendall does not teach the specific way of selecting targeted users that
is recited in Appellant’s claim 1, and instead relies on Bentolila to cure the
deficiency. Final Act. 14–15. Appellant argues Bentolila is deficient in
several respects, but we need only address one specific argument to resolve
the § 103 issues.
Claim 1 recites, in relevant part,
selecting features for the user model to include selected social
graph features, each social graph feature indicating one or more
connections to users in the training set of users, the connections
stored in a social graph of the social networking system, such that
Appeal 2019-003274
Application 13/297,117

22
users that match a social graph feature have connections to the
users in the training set of users matching the connections
indicated in the social graph feature
OA Response 4. The Examiner cites Bentolila as teaching this limitation.
Final Act. 17 (citing Bentolila col. 8, ll. 3–9). Specifically, the Examiner
finds Bentolila’s aggregation of user profiles into groups teaches this
limitation.
Appellant argues the Examiner has failed to address the “social graph”
aspect of the limitation and that Bentolila does not teach any “social graph
feature indicating one or more connections to users.” Appeal Br. 16–17;
Reply Br. 9–10. We agree.
Although Bentolila describes aggregating user profiles into groups,
we discern nothing in Bentolila that can be reasonably understood to
suggesting the use of social graphs or any “social graph feature indicating
one or more connections to users in the training set of users.” As such, we
are persuaded the Examiner erred in finding Bentolila teaches or suggests
the recited “social graph” limitations. We, therefore, do not sustain the
rejection of claim 1, or of dependent claims 2–7, 9, 17, 18, and 28–31, under
35 U.S.C. § 103.
CONCLUSION
Because we have affirmed at least one ground of rejection for each
claim on appeal, we affirm the Examiner’s decision to reject the claims.
37 C.F.R. § 41.50(a)(1).
More specifically:
We affirm the rejection of claims 1–7, 9, 17, 18, and 28–31 under
35 U.S.C. § 112, second paragraph as being indefinite.
Appeal 2019-003274
Application 13/297,117

23
We affirm the rejection of claims 1–7, 9, 17, 18, and 28–31 under
35 U.S.C. § 101 as being directed to ineligible subject matter.
We reverse the rejection of claims 1–7, 9, 17, 18, and 28 under
35 U.S.C. § 103 as being unpatentable over Kendall, Bentolila,
Bagherjeiran, and Duan.
We reverse the rejection of claims 29–31 under 35 U.S.C. § 103 as
being unpatentable over Kendall, Bentolila, Bagherjeiran, Duan, and
Elvekrog.

DECISION SUMMARY
Claims
Rejected
35 U.S.C.
§
Reference(s)/Basis Affirmed Reversed
1–7, 9, 17,
18, 28–31
112,
second
paragraph
Indefiniteness 1–7, 9, 17,
18, 28–31

1–7, 9, 17,
18, 28–31
101 Eligibility 1–7, 9, 17,
18, 28–31

1–7, 9, 17,
18, 28
103(a) Kendall, Bentolila,
Bagherjeiran, Duan
1–7, 9, 17,
18, 28
29–31 103(a) Kendall, Bentolila,
Bagherjeiran,
Duan, Elvekrog
29–31
Overall
Outcome
1–7, 9, 17,
18, 28–31

TIME PERIOD FOR RESPONSE
No time period for taking any subsequent action in connection with
this appeal may be extended under 37 C.F.R. § 1.136(a). See 37 C.F.R.
§ 1.136(a)(1)(iv).
AFFIRMED