Ex Parte 7685276 et al

Patent Trial and Appeal Board

Oct 28, 2014

95001699 (P.T.A.B. Oct. 28, 2014)

MOD PTOL-90A (Rev.06/08)

APPLICATION NO./
CONTROL NO.
FILING DATE FIRST NAMED INVENTOR /
PATENT IN REEXAMINATION
ATTORNEY DOCKET NO.
95/001,699 08/03/2011 7685276

EXAMINER
DENTONS US LLP
P.O. BOX 061080
CHICAGO, IL 60606-1080
DESAI, RACHNA SINGH
ART UNIT PAPER NUMBER
3992
MAIL DATE DELIVERY MODE

10/31/2014 PAPER

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.

UNITED STATES DEPARTMENT OF COMMERCE
U.S. Patent and Trademark Office
Address : COMMISSIONER FOR PATENTS
P.O. Box 1450
Alexandria, Virginia 22313-1450
www.uspto.gov
UNITED STATES PATENT AND TRADEMARK OFFICE
_____________________________________________________________________________________
UNITED STATES PATENT AND TRADEMARK OFFICE
____________

BEFORE THE PATENT TRIAL AND APPEAL BOARD
____________

PERSONALIZED USER MODEL LLP
Patent Owner and Appellant

v.

GOOGLE INC.
Requester and Respondent
____________

Appeal 2014-007047
Reexamination Control 95/001,699
Patent 7,685,276 B2
Technology Center 3900
____________

Before BRADLEY W. BAUMEISTER, JENNIFER L. McKEOWN, and
IRVIN E. BRANCH, Administrative Patent Judges.

BRANCH, Administrative Patent Judge.

Appeal 2014-007047
Reexamination Control 95/001,699
Patent 7,685,276 B2

2
DECISION ON APPEAL
Patent Owner, Personalized User Model LLP (“Patent Owner”),

appeals under 35 U.S.C. §§ 134 and 315 the Examiner’s rejecting claims 1,
3, 5–7, 14, and 21–24, all reexamined claims. PO App. Br. 1.
1

We have jurisdiction under 35 U.S.C. §§ 134 and 315. Oral
arguments were heard on September 24, 2014. We affirm.

STATEMENT OF THE CASE
Google Inc. (“Requester”) filed a Request for Inter Partes
Reexamination on July 29, 2011, of claims 1, 3, 5–7, 14, and 21–24 of
United States Patent 7,685,276 B2 (hereinafter the “’276 patent”).2
According to the Patent Owner, the ’276 patent describes
a method of personalizing user interactions across one or more
modes of interaction and applications, including searching,
browsing, email reading and navigation.[] Learning about user
interests in one mode benefits all other modes.[] Users are
modeled independent of any specific representation or data
structure, allowing use of a variety of knowledge sources and
the addition of new knowledge sources into the user model.[]
The user model estimates user interest in unseen documents by
generalizing from the information learned about the user.

1
Throughout this opinion, we refer to (1) the Right of Appeal Notice
mailed August 17, 2012 (“RAN”); (2) the Examiner’s Answer mailed
January 10, 2014 (“Ans.”) (incorporating the RAN by reference); (3) Patent
Owner’s Appeal Brief filed November 19, 2012 (“PO App. Br.”); (4)
Requester’s Corrected Respondent Brief filed October 23, 2013 (“TPR Resp.
Br.”); and (5) Patent Owner’s Rebuttal Brief filed February 10, 2014 (“PO
Reb. Br.”).
2
The ’276 patent issued to Konig, et al., on March 23, 2010, from
Application 12/008,148, filed January 8, 2008.
Appeal 2014-007047
Reexamination Control 95/001,699
Patent 7,685,276 B2

3
PO App. Br. 6–7 (citation to various passages of the ’276 patent
omitted).
Claims 1 and 23, reproduced below, are illustrative of the claimed
subject matter:
1. A computer-implemented method for providing
personalized information services to a user, the method
comprising:
transparently monitoring user interactions with data
while the user is engaged in normal use of a browser program
running on the computer;
analyzing the monitored data to determine documents of
interest to the user;
estimating parameters of a user-specific learning machine
based at least in part on the documents of interest to the user;
receiving a search query from the user;
retrieving a plurality of documents based on the search
query;
for each retrieved document of said plurality of retrieved
documents:
identifying properties of the retrieved document,
and applying the identified properties of the retrieved document
to the user-specific learning machine to estimate a probability
that the retrieved document is of interest to the user; and
using the estimated probabilities for the respective
plurality of retrieved documents to present at least a portion of
the retrieved documents to the user.

23. A computer-implemented method for providing
personalized information services to a user, the method
comprising:
transparently monitoring user interactions with data
while the user is engaged in normal use of a browser program
running on the computer;
analyzing the monitored data to determine documents of
interest to the user;
estimating parameters of a user-specific learning machine
based at least in part on the documents of interest to the user;
Appeal 2014-007047
Reexamination Control 95/001,699
Patent 7,685,276 B2

4
collecting a plurality of documents of interest to a user;
for each of said plurality of collected documents:
identifying properties of the collected document,
and applying the identified properties of the collected document
to the user-specific learning machine to estimate a probability
that the collected document is of interest to the user;
using the estimated probabilities for the respective
plurality of collected documents to select at least a portion of
the collected documents;
presenting said selected collected documents to
said user.

RELATED PROCEEDINGS
We are informed that the ’276 patent and its parent, United States
Patent 6,981,040 B1 (“the ’040 patent”), are the subject of litigation in
Personalized User Model LLP v. Google, Inc., Case No. 1:09-CV-525-LPS,
USDC D. Del.
This panel, in reexamination control No. 95/001,569, affirmed the
Examiner’s rejection of all reexamined claims of the ’040 patent. See
Decision on Appeal 2014-002460 mailed May 30, 2014.

THE APPEALED REJECTIONS
Patent Owner appeals the Examiner’s rejecting the claims as follows:
1. Claims 1, 3, 5, 6, 14, 21, and 22 over Wasfi3 and Mladenic4 under
35 U.S.C. § 103(a);

3
Ahmad M. Ahmad Wasfi, Collecting User Access Patterns for Building
User Profiles and Collaborative Filtering, School of Computer Sciences
(1999).
Appeal 2014-007047
Reexamination Control 95/001,699
Patent 7,685,276 B2

5
2. Claims 23 and 24 over Wasfi under 35 U.S.C. § 102(a);
3. Claim 7 over Wasfi, Mladenic, and Refuah5 under 35 U.S.C. § 103(a);
4. Claims 1, 3, 5, 6, 7, 14, and 21–24 over Refuah under 35 U.S.C.
§ 102(a);
5. Claims 1, 3, 6, 7, 21, 22, and 23 over Culliss6 under 35 U.S.C.
§ 102(e);
6. Claim 5 over Culliss and Mladenic under 35 U.S.C. § 103(a);
7. Claim 24 over Culliss and Montebello7 under 35 U.S.C. § 103(a);
8. Claims 1, 3, 6, 7, and 21–24 over Montebello under 35 U.S.C.
§§ 102(a) and 102(b);
9. Claim 5 over Montebello and Mladenic under 35 U.S.C. § 103(a);
10. Claims 1, 3, 5, 6, 21, and 22 over Barrett8 and Mladenic under
35 U.S.C. § 103(a);
11. Claim 7 over Barrett, Mladenic, and Refuah under 35 U.S.C. § 103(a);
12. Claims 23 and 24 over Barrett under 35 U.S.C. §§ 102(a) and 102(b);
13. Claims 23 and 24 over Asnicar9 under 35 U.S.C. §§ 102(a) and
102(b);

4
Dunja Mladenić (hereinafter “Mladenic”), Personal Web Watcher: design
and implementation, Dpt. for Intelligent Systems, J. Stefan Institute, School
of Computer Science, Carnegie Mellon University (1996).
5
Refuah et al., US 7,631,032 B1, issued December 8, 2009.
6
Gary Culliss, US 6,182,068 B1, issued January 30, 2001.
7
M. Montebello, et al., A Personal Evolvable Advisor for WWW Knowledge-
Based Systems, Computer Science Department, University of Wales,
Cardiff. (1998).
8
Rob Barrett, et al., How to Personalize the Web, IBM Almaden Research
Center (1997).
Appeal 2014-007047
Reexamination Control 95/001,699
Patent 7,685,276 B2

6
14. Claims 1, 5, 6, 14, 21, and 22 are patentable over Asnicar in view of
Mladenic under 35 U.S.C. § 103(a);
15. Claims 3 and 7 are patentable over Asnicar, Mladenic, and Culliss
under 35 U.S.C. § 103(a);
16. Claims 23 and 24 over Stefani10 under 35 U.S.C. §§ 102(a) and
102(b);
17. Claims 1, 5, 6, and 21 over Stefani and Mladenic under 35 U.S.C.
§ 103(a);
18. Claims 3 and 7 over Stefani, Mladenic, and Culliss under 35 U.S.C.
§ 103(a); and
19. Claim 22 over Stefani, Mladenic, and Refuah under 35 U.S.C.
§ 103(a).

THE REJECTION OF CLAIMS 23 AND 24 OVER WASFI
The Examiner rejected claims 23 and 24 as anticipated by Wasfi.
RAN 6 (adopting Third Party Requester’s proposed rejection); see Request
for Inter Partes Reexamination (“Request” filed July 29, 2011) 29–33 and
Claim Chart CC-B; see also RAN 23–28.

9
Fabio A. Asnicar, et al., ifWeb: A Prototype of User Model-Based
Intelligent Agent for Document Filtering and Navigation in the World Wide
Web, Proceedings of the workshop “Adaptive Systems and User Modeling
on the World Wide Web”, Sixth International Conference on User
Modeling, Chia Laguna, Sardinia (1997).
10
Anna Stefani, et al., Personalizing Access to Web Sites: The SiteIF
Project, “Proceedings of the 2
nd
Workshop on Adaptive Hypertext and
Hypermedia HYPERTEXT’98”, Pittsburgh, USA (1998).
Appeal 2014-007047
Reexamination Control 95/001,699
Patent 7,685,276 B2

7
Wasfi discloses “ProfBuilder,” a recommendation system for
individual pages found within a single web site. Wasfi p. 57. See also PO
App. Br. 7. Wasfi creates two recommendation lists for pages within a
single web site a user is currently visiting. Wasfi Fig. 2 and pp. 60–63. One
list is created by a content-based filter, and the other is created by a
collaborative filter. Id.
Wasfi’s content-based filter “translate[s] pages of [a] web site to a
representation used for a user profile and determines similarities between the
pages and the profile.” PO App. Br. 7 (citing Wasfi p. 61). “Similarity is
computed as the scalar product of weighted coefficients (specifically,
keywords) of the page vector and the user profile vector.” Id. “Those pages
deemed most similar to ones the user has already visited are ranked higher
than others.” Id.
Patent Owner argues Wasfi “fails to teach any application of
identified properties of documents to a user-specific learning machine to
estimate a probability that a document is of interest to the user.” PO App.
Br. 14. Patent Owner further argues that neither Wasfi’s “similarity
measures” nor Wasfi’s “probabilities” is “equivalent to the claimed
probability of a user’s interest in a document.” Id. Patent Owner points out
that Wasfi’s similarity measures are not probabilities because they can have
a value between 0 and infinity and infinity cannot be normalized to a
probability. Id. at 13. Patent Owner reasons that “[i]nasmuch as no
probabilities regarding whether the document is of interest to a user are
estimated, no such probabilities are available to use in making any
selections.” Id. at 14.
Appeal 2014-007047
Reexamination Control 95/001,699
Patent 7,685,276 B2

8
Requester argues “Wasfi’s vector-space measurements are simply a
means for estimating the probability that a document is of interest to the user
— a smaller distance between document and user vectors means a greater
probability that the document is interesting to the user, while a larger
distance means a lesser probability that the document is interesting to the
user.” TPR Resp. Br. 3. Requester also argues that “even if Patent Owner
were correct that Wasfi’s methods would not yield a true probability in the
extreme case of infinitely dissimilar vectors, Wasfi would still yield a
probability in every other case.” Id. at 5. We agree.
Wasfi describes recommending pages to a user based on 1) the user’s
continuously updated profile determined by pages visited by the specific
user; 2) the similarity among pages as determined by a text analysis of the
pages; and 3) the patterns of other users based on their navigation among the
pages. See, e.g., 61. Patent Owner’s argument that “the interestingness of
any page at the subject web site is dependent upon (and determined solely
by) the visiting patterns of other users and cannot be deemed a user-specific
learning machine, as claimed” (PO Reb. Br. 3) is unpersuasive because it
fails to consider that Wasfi takes into consideration the instant user’s actions
with respect to pages visited. Whether Wasfi is unable to compute
interestingness absent knowledge of others’ accesses to web pages that the
subject user might visit next (id.) is unpersuasive because Wasfi describes
using both a user’s specific actions and page similarities based on keywords
(i.e., properties) as claimed. Moreover, the fact that Wasfi operates within a
specific web site (PO App. Br. 8) does not negate that Wasfi’s learning
module operates to recommend other pages (i.e., documents) within that site.
Appeal 2014-007047
Reexamination Control 95/001,699
Patent 7,685,276 B2

9
Id. In other words, one skilled in the art would understand Wasfi describes
“identifying properties of the collected document, and applying the
identified properties of the collected document to the user-specific learning
machine to estimate a probability that the collected document is of interest to
the user.” Accordingly, we are unpersuaded of error in the rejection of claim
23.
Patent owner does not argue separately the rejection claim 24 over
Wasfi.

THE REJECTION OF CLAIMS 1, 3, 5, 6, 14, 21, AND 22 OVER WASFI AND
MLADENIC
The Examiner rejected claims 1, 3, 5, 6, 14, 21, and 22 as obvious
over Wasfi and Mladenic. RAN 6 (adopting Third Party Requester’s
proposed rejection); see Request 33–39 and Claim Chart CC-B; see also
RAN 28.
Patent Owner argues Wasfi and Mladenic fail to teach or suggest
claim 1’s “receiving a search query” and “retrieving a plurality of documents
based on the search query.” PO App. Br. 11–12; PO Reb. Br. 1–2. Patent
Owner cites the Examiner’s statement, while discussing whether Mladenic
anticipates certain claims (RAN 12–22), that Mladenic does not “disclose”
the limitations relating to search. PO App. Br. 11 (citing RAN 14). Patent
Owner also argues that, when considered in its entirety, Mladenic would
have led one away from “a scheme based on or requiring search.” Id. See
also PO Reb. Br. 1–2.
Appeal 2014-007047
Reexamination Control 95/001,699
Patent 7,685,276 B2

10
Requester points out that, in addition to describing “Personal Web
Watcher,” which “eschews the use of search,” Mladenic also discloses “Web
Watcher,” which does use search. TPR Resp. Br. 1–2 (citing Mladenic at 2).
We agree.
In Mladenic’s Web Watcher “the user provides a few keywords
describing a search goal and WebWatcher [] suggest[s] pages related to the
current page.” Mladenic at 2. Hence, we are persuaded that Mladenic’s
Web Watcher teaches or suggests “receiving a search query” and “retrieving
documents based on the search query.”
Moreover, we are not persuaded Mladenic teaches away from query-
based search. It is true that “[a] reference may be said to teach away when a
person of ordinary skill, upon reading the reference, would be discouraged
from following the path set out in the reference, or would be led in a
direction divergent from the path that was taken by the applicant.” In re
Gurley, 27 F.3d 551, 553 (Fed. Cir. 1994). However, “the prior art’s mere
disclosure of more than one alternative does not constitute a teaching away
from any of these alternative because such disclosure does not criticize,
discredit, or otherwise discourage the solution claimed . . .” In re Fulton,
391 F.3d 1195, 1201 (Fed. Cir. 2004). Hence, while Mladenic’s Personal
Web Watcher provides an alternative to Web Watcher, Patent Owner does
not argue persuasively that Mladenic criticizes, discredits, or discourages
query-based search.
Patent Owner also argues error in the Examiner’s rejection of claims
1, 3, 5, 6, 14, 21, and 22 over Wasfi and Mladenic because, according to
Patent owner, Wasfi does not “identify[] properties of [a] retrieved
Appeal 2014-007047
Reexamination Control 95/001,699
Patent 7,685,276 B2

11
document,” apply the properties to a “user-specific learning machine,” or
estimate probabilities as recited in claim 1. PO App. Br. 12–14. See also
PO Reb. Br. 3. Claim 1 is identical to claim 23 in this regard, so we are
unpersuaded by Patent Owner’s arguments for the reasons discussed supra.
Patent Owner does not argue separately the rejection of claims 3, 5, 6,
14, 21, and 22 over Wasfi and Mladenic.

THE REJECTION OF CLAIMS 1, 3, 5–7, AND 21–24 OVER REFUAH
The Examiner rejected claims 1, 3, 5–7, and 21–24 as anticipated by
Refuah. RAN 7 (adopting Third Party Requester’s proposed rejection); see
Request 40–50 and Claim Chart CC-C; see also RAN 29–37.
Refuah discloses tracking a user’s interactions with the Internet and
adopting future interactions based on the tracked interactions, including
future interactions related to site content. Refuah Abstract.
Patent Owner argues Refuah does not disclose “applying the
identified properties of the retrieved document to the user-specific learning
machine to estimate a probability that the retrieved document is of interest to
the user” and “estimating parameters of a user-specific learning machine.”
PO App. Br. 15–19 (emphasis omitted). Patent Owner asserts Refuah does
not discuss the concept of probabilities because information-dependent
results are “incongruent” with estimating probability. Id. at 16. Patent
Owner also argues that, when “applying the identified properties of the
retrieved document to the user-specific learning machine to estimate a
probability that the retrieved document is of interest to the user” is properly
construed, “[t]here is no learning machine in Refuah in or to which to apply
Appeal 2014-007047
Reexamination Control 95/001,699
Patent 7,685,276 B2

12
the identified properties of the retrieved document.” Id. at 16–18. Patent
Owner also argues that Refuah does not teach receiving a search query and,
therefore, does not teach the limitations that flow therefrom. Id. at 18.
We are unpersuaded of error at least for the reasons stated by the
Examiner. See RAN 29–37. We specifically note the factual underpinnings
embodied in Requester’s claim chart CC-C. See Request Appendix CC-C.
We further note the following for emphasis.
Patent Owner’s argument that “on the face of the Refuah patent,
nowhere are the terms ‘learning,’ ‘learning machine,’ ‘probability,’ or
‘estimating a probability’ described” (PO App. Br. 15) is unpersuasive
because there is no ipsissimis verbis test for determining whether a reference
discloses a claim element, i.e., identity of terminology is not required. In re
Bond, 910 F.2d 831, 832 (Fed. Cir. 1990). Patent Owner’s arguments
relating to Refuah’s lack of teaching relating to probabilities is belied by, for
example, Refuah’s teaching that “[i]n a preferred embodiment of the
invention, depending on a persona [“user-specific learning machine”],
several characteristics [“properties”] of a site [web site having pages or
“documents”] may be defined, which may be used in filtering out [not
recommending] and/or prioritizing [sorting based on a numeric predictor]
such a site.” Refuah 7:53–58. We are persuaded that one skilled in the art
would understand Refuah’s filtering to be a measure of the probability that
the web site is of interest to the user and Refuah’s prioritizing to be
reflective of the actual probability.
Patent Owner’s argument that Refuah operates on web sites and
therefore does not describe “documents of interest to a user” (PO App. Br.
Appeal 2014-007047
Reexamination Control 95/001,699
Patent 7,685,276 B2

13
18) is unavailing. Patent Owner does not provide argument or evidence
sufficient to persuade us that the Examiner was unreasonable in construing
“document” such that it includes web sites or html web pages within a web
site, as Refuah describes. Moreover, the ’276 patent states that “document”
“includes not just text, but any type of media, including, but not limited to,
hypertext, database, spreadsheet, image, sound, and video.” The ’276 patent
9:15–18. We find a web site and its pages are encompassed by “any type of
media.”
We are persuaded that Refuah describes receiving a search query
because Refuah states “a persona is used to personalize information retrieval
. . . search engines return matches for a particular query, while personality
and mood are designed to affect the results of substantially any query.”
Refuah 17:21–30. See also id. at 34–43. Accordingly, we are persuaded
that Refuah describes “applying the identified properties of the retrieved
document to the user-specific learning machine to estimate a probability that
the retrieved document is of interest to the user.”
We are unpersuaded by Patent Owner’s arguments (PO App. Br. 18–
19) that Refuah does not describe “estimating parameters of a user-specific
learning machine.” Refuah describes that “parameters” of a persona “may
define the traits [that] should be evaluated when determining a suitability of
a site to a persona” and may define “how to modify the persona and/or a
mood based on user activities.” Refuah 6:49–64.
Patent Owner argues further in the rebuttal brief that Requester’s and
Patent Owner’s experts in the related litigation refute that Refuah discloses a
learning machine. PO Reb. Br. 4–5. These arguments are unavailing. We
Appeal 2014-007047
Reexamination Control 95/001,699
Patent 7,685,276 B2

14
are unpersuaded that Refuah does not describe a learning machine for the
reasons discussed supra. We are not bound by the experts’ statements, at
least because they are based on a construction of learning machine that we
are not bound to adopt.
For at least the foregoing reasons, we are unpersuaded by Patent
Owner’s arguments and find Refuah describes “estimating parameters of a
user-specific learning machine based at least in part on the documents of
interest to the user” and “applying the identified properties of the retrieved
document to the user-specific learning machine to estimate a probability that
the retrieved document is of interest to the user.” Accordingly, we are
unpersuaded of error in the rejection of claims 1, 3, 5–7, and 21–24 as
anticipated by Refuah.

THE REJECTION OF CLAIM 7 OVER WASFI, MLADENIC, AND
REFUAH
The Examiner rejected claim 7 as obvious over Wasfi, Mladenic, and
Refuah. RAN 6–7 (adopting Third Party Requester’s proposed rejection);
see Request 39–40 and Claim Chart CC-B; see also RAN 28–29.
Patent owner argues claim 7 is patentable over Wasfi, even when
considered in combination with Mladenic and Refuah. PO App. Br. 15.
Patent Owner’s arguments, however, are premised on shortcomings in the
rejections of claim 1 over Wasfi or Refuah (id.), which we do not find. See
supra. Accordingly, we are unpersuaded of error in the rejection of claim 7
over Wasfi, Mladenic, and Refuah.

Appeal 2014-007047
Reexamination Control 95/001,699
Patent 7,685,276 B2

15
CONCLUSIONS
We sustain the Examiner’s rejections of claims 23 and 24 as
anticipated by Wasfi; claims 1, 3, 5, 6, 14, 21, and 22 as obvious over Wasfi
and Mladenic; claims 1, 3, 5–7, and 21–24 as anticipated by Refuah; and
claim 7 as obvious over Wasfi, Mladenic, and Refuah.
Affirmance of these rejections renders it unnecessary to reach the
remaining rejections, as all claims subject to reexamination have been
addressed and found unpatentable. Cf. In re Gleave, 560 F.3d 1331, 1338
(Fed. Cir. 2009) (not reaching additional obviousness rejections).
DECISION
We affirm the Examiner’s decision to reject claims 1, 3, 5–7, 14, and
21–24.
Requests for extensions of time in this inter partes reexamination
proceeding are governed by 37 C.F.R. § 1.956. See 37 C.F.R. § 41.79.
Pursuant to 37 C.F.R. § 41.79(d), this decision is final for the purpose
of judicial review. A party seeking judicial review must timely serve notice
on the Director of the United States Patent and Trademark Office. See
37 C.F.R. §§ 90.1 and 1.983.

AFFIRMED

alw

Appeal 2014-007047
Reexamination Control 95/001,699
Patent 7,685,276 B2

16
Patent Owner:

DENTONS US LLP
P.O. BOX 061080
CHICAGO, IL 60606-1080

Third-Party Requester:

ROBERT WILSON
QUINN, EMANUEL, URQUHART & SULLIVAN, LLP
51 MADISON AVENUE
22
ND
FLOOR
NEW YORK, NY 10010