Adobe Inc.

Patent Trials and Appeals Board

Jan 12, 2021

2019003892 (P.T.A.B. Jan. 12, 2021)

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.
14/031,256 09/19/2013 Sayaji Hande 3463US01 1085
108982 7590 01/12/2021
FIG. 1 Patents
116 W. Pacific Avenue
Suite 200
Spokane, WA 99201
EXAMINER
STARKS, WILBERT L
ART UNIT PAPER NUMBER
2122
NOTIFICATION DATE DELIVERY MODE
01/12/2021 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):
Fig1Docket@fig1patents.com
PTOL-90A (Rev. 04/07)
UNITED STATES PATENT AND TRADEMARK OFFICE
____________________

BEFORE THE PATENT TRIAL AND APPEAL BOARD
____________________

Ex parte SAYAJI HANDE, VINEET GUPTA, and
SANDEEP ZECHARIAH GEORGE KOLLANNUR
____________________

Appeal 2019-003892
Application 14/031,256
Technology Center 2100
____________________

Before JOHN A. EVANS, JUSTIN BUSCH, and JOHN P. PINKERTON,
Administrative Patent Judges.

PINKERTON, Administrative Patent Judge.

DECISION ON APPEAL
Appellant1 appeals under 35 U.S.C. § 134(a) from the Examiner’s
Final Rejection of claims 1–7, 9–14, and 16–23, which are all of the claims
pending in the application. Claims 8 and 15 are canceled. We have
jurisdiction under 35 U.S.C. § 6(b).
We reverse.

1 We use the word “Appellant” to refer to “applicant” as defined in
37 C.F.R. § 1.42. Appellant identifies the real party in interest as Adobe
Systems Incorporated. Appeal Br. 3.
Appeal 2019-003892
Application 14/031,256

2
STATEMENT OF THE CASE
Appellant’s Invention
Appellant generally describes the disclosed and claimed invention as
relating to “techniques for predicting spread of content across social
networks.” Spec. ¶ 3.
Illustrative Claim
Claims 1, 10, and 17 are independent. Independent claim 1 is
illustrative of the subject matter on appeal and provides as follows:
1. A computer-implemented method comprising:
receiving at least one parameter related to a page on a
social network;
accessing user interactions with content posted to the page
on the social network during an initial duration of time after the
content is posted to the page on the social network, the content
comprising a short-text post, a picture, or a video, and the user
interactions with the content comprising at least one of a view of
the content or a share of the content; and
predicting future user interactions with the content during
at least one subsequent duration of time by applying the received
parameters and the accessed user interactions to a prediction
model that is configured to predict the future user interactions
based at least in part on a total number of the user interactions
with the content during the initial duration of time, the future user
interactions comprising at least one of viewing the content,
clicking on the content, responding or commenting on the
content, replying to responses to the content, liking the content,
disliking the content, loving the content, sharing the content, or
retweeting the content.
Appeal Br. 40 (Claims App.).
Appeal 2019-003892
Application 14/031,256

3
Reference and Rejection on Appeal
Claims 1–7, 9–14, and 16–23 stand rejected under 35 U.S.C.
§ 102(a)(1) as being anticipated by Lars Backstrom and Jure Leskovec,
Supervised Random Walks: Predicting and Recommending Links in Social
Networks, in WSDM’11: PROC. OF THE FOURTH ACM INT’L CONF. ON WEB
SEARCH AND DATA MINING 635–44 (2011) (“Backstrom”).2 Ans. 4–44; see
also Final Act. 4–29.
ANALYSIS
In rejecting claim 1, the Examiner finds Backstrom discloses, among
other limitations, “accessing user interactions with content posted to the
page on the social network . . . and predicting future user interactions with
the content . . . by applying . . . the accessed user interactions to a prediction
model.” Ans. 5–8 (citing Backstrom 635–637, 644); see also Final Act. 3–5.
In particular, the Examiner finds Backstrom discloses predicting what new
edges (friendships) a user node s will create over a time interval based in
part on interaction attributes such as how many messages edge user nodes u
and v exchanged, or how many photos they appeared in together. Ans. 5–7
(citing Backstrom 635, 637); see also id. at 46–54. The Examiner explains
that Backstrom’s interaction attributes anticipate claim 1’s “user
interactions” because Backstrom discloses content including messages and
photos that are exchanged. Id. at 46, 48, 51, 52. The Examiner explains
further that Backstrom “predicts new edges and edges are metadata of
interactions,” which “must be accessed, at least to the level of counting them

2 The copy of Backstrom provided in the record does not include page
numbers. We identify the pages of this paper according to its original
pagination, starting with page 635 and ending with page 644.
Appeal 2019-003892
Application 14/031,256

4
in order to make the edges.” Id. at 56, 58. The Examiner finds Backstrom
further discloses “predictions for like recommendations” because its
“Supervised Random Walks” algorithm can be applied to “like
recommendations.” Id. at 7–8 (quoting Backstrom 644 (emphasis added)).
The Examiner finds Backstrom also discloses “future user interactions
comprising . . . clicking on the content” because Backstrom discloses its
algorithm can be used for “link recommendation (positive nodes are those
which user clicks on).” Id. at 8–9 (quoting Backstrom 636 (emphasis
added)). The Examiner explains that in Backstrom, “the user ‘clicks on’
links and the prior art predicts links.” Id. at 59.
Appellant contends the Examiner erred in finding that Backstrom
discloses “accessing user interactions with content posted to the page on the
social network . . . and predicting future user interactions with the content
. . . by applying . . . the accessed user interactions to a prediction model” as
recited in claim 1. Appeal Br. 15–22; Reply Br. 2–7. Appellant asserts that
the Examiner’s rejection maps the term “content” in claim 1’s “accessing”
step to Backstrom’s “messages” exchanged between nodes u and v, but then
does not map “the content” of claim 1’s “predicting” step to this feature of
Backstrom. Reply Br. 3–4 (citing Ans. 45–46, 54, 58). Appellant argues
that, to the extent the Examiner asserts Backstrom’s messages disclose the
“content” of claim 1, in order to properly reject claim 1 under § 102,
Backstrom must also disclose
predicting future user interactions with [the messages]. . . the
future user interactions comprising at least one of viewing the
[messages], clicking on the [messages], responding or
commenting on the [messages], replying to responses to the
[messages], liking the [messages], disliking the [messages],
Appeal 2019-003892
Application 14/031,256

5
loving the [messages], sharing the [messages], or retweeting the
[messages].
Id. at 4–5. According to Appellant, however, “the Examiner does not assert
Backstrom as disclosing this subject matter of claim 1, nor does Backstrom
disclose this subject matter of claim 1.” Id. at 4–5. Rather, Appellant
asserts that “Backstrom describes predicting for each user in the social
network ‘what new edges (friendships) that user will create between t and
some future time t’” including “a model that treats users as nodes and is
useable to suggest to an individual user ‘a list of people that the user is likely
to create new connections to.’” Id. at 5 (citing Backstrom 635, 637).
Appellant therefore argues that “Backstrom merely describes predicting a
number of future friendships that will form between users in a social
network,” which “is not the same as ‘predicting future user interactions with
the content.’” Id.
Appellant also disputes the Examiner’s assertion that “the user ‘clicks
on’ links and the prior art predicts links” as a mischaracterization of
Backstrom. Id. at 6 (citing Ans. 58–59). To the contrary, Appellant asserts
that Backstrom does not “include[] a description that suggests that a user
clicks on links,” but “[r]ather, . . . merely describes that nodes a user clicks
on can be used as positive training examples for use in link prediction (e.g.,
predicting future friendships).” Id.
Anticipation requires a single prior art reference to disclose each and
every claim element, in complete detail, and as arranged in the claim. Net
MoneyIN, Inc. v. VeriSign, Inc., 545 F.3d 1359, 1369 (Fed. Cir. 2008);
Richardson v. Suzuki Motor Co., 868 F.2d 1226, 1236 (Fed. Cir. 1989). For
the reasons discussed below, we are persuaded of error because the
Appeal 2019-003892
Application 14/031,256

6
Examiner fails to show that Backstrom discloses each and every element of
claim 1.
Backstrom describes predicting new edges (friendships) that a user
node will create within a social network. E.g., Backstrom 635. But the
Examiner has not shown, nor do we find, that Backstrom discloses
“accessing user interactions with content posted to the page on the social
network . . . and predicting future user interactions with the content” as
recited in claim 1. Although Backstrom describes an algorithm that predicts
new edges or links for node s using messages exchanged between edge
nodes u and v (and photos including edge nodes u and v), a new edge or link
is not the same as content. Nor has the Examiner shown that Backstrom
predicts future user interactions with the messages or photos involving nodes
u and v. And even though Backstrom may suggest predicting future user
interactions with content (for example, that user node s will add user node x
as a friend and, thereafter, communicate with user node x), that content
relates to future user interactions, not past user interactions that were
accessed and used in Backstrom’s prediction algorithm.
Nor does Backstrom’s disclosure of “link recommendation (positive
nodes are those which user clicks on)” (Backstrom 636) fill the gaps in the
Examiner’s rejection. As best understood, this disclosure describes a first
user node clicking on a second user node, and using this user interaction to
predict a future link or friendship between the first and second nodes. See
Reply Br. 6 (arguing this passage “merely describes that nodes a user clicks
on can be used as positive training examples for use in link prediction (e.g.,
predicting future friendships)”). While this may suggest accessing user
interactions with content and using them to predict future user links or
Appeal 2019-003892
Application 14/031,256

7
friendships between nodes, the Examiner has not shown, nor do we find, that
this also discloses “predicting future user interactions with the content”
itself. For example, the Examiner has not shown that Backstrom discloses a
prediction that the first user node will click on the second user node again.
In addition, the Examiner’s finding that “predictions for like
recommendations is taught by Backstrom” incorrectly interprets the cited
disclosure of Backstrom. The cited disclosure of Backstrom states that
“Supervised Random Walks are not limited to link prediction, and can be
applied to many other problems that require learning to rank nodes in a
graph, like recommendations, anomaly detection, missing link, and expertise
search and ranking.” Backstrom 644. Contrary to the Examiner’s finding,
however, the term “like” is not being used here as an adjective to modify the
word “recommendations,” but rather to introduce the following examples of
problems that require learning to rank nodes in a graph—“recommendations,
anomaly detection, missing link, and expertise search and ranking.” Id. We
therefore disagree with the Examiner’s finding that Backstrom discloses
“like recommendations.” Nor has the Examiner otherwise shown that the
cited disclosure of Backstrom satisfies the disputed claim limitations.
Accordingly, the Examiner has not adequately shown that Backstrom
discloses each and every claim element, in complete detail, and as arranged
in the claim. We decline to resort to speculation to fill the gaps in the
Examiner’s rejection. See Ex parte Braeken, 54 USPQ2d 1110, 1112 (BPAI
1999). We therefore determine the Examiner erred in rejecting claim 1
under 35 U.S.C. § 102(a)(1) as anticipated by Backstrom.
Accordingly, constrained by this record, we do not sustain the
Examiner’s § 102(a)(1) rejection of claim 1. For similar reasons, we do not
Appeal 2019-003892
Application 14/031,256

8
sustain the Examiner’s § 102(a)(1) rejection of claims 2–7, 9–14, and 16–23,
each of which includes similar limitations. For these claims, the Examiner
fails to provide a finding that cures the defects discussed above. See Ans. 9–
44; Final Act. 12–29; cf. In re Fritch, 972 F.2d 1260, 1266 (Fed. Cir. 1992)
(“dependent claims are nonobvious if the independent claims from which
they depend are nonobvious”).
CONCLUSION
We do not sustain the Examiner’s decision to reject claims 1–7, 9–14,
and 16–23 under 35 U.S.C. § 102(a)(1).

DECISION SUMMARY
Claims
Rejected 35 U.S.C. §
Reference(s)/
Basis Affirmed Reversed
1–7, 9–14,
16–23 102(a)(1) Backstrom
1–7, 9–14,
16–23

REVERSED