Ex Parte Liu et al

Patent Trial and Appeal Board

Aug 22, 2018

12764983 (P.T.A.B. Aug. 22, 2018)

UNITED STA TES p A TENT AND TRADEMARK OFFICE
APPLICATION NO. FILING DATE
12/764,983 04/22/2010
69316 7590 08/24/2018
MICROSOFT CORPORATION
ONE MICROSOFT WAY
REDMOND, WA 98052
FIRST NAMED INVENTOR
Chao Liu
UNITED STATES DEPARTMENT OF COMMERCE
United States Patent and Trademark Office
Address: COMMISSIONER FOR PATENTS
P.O. Box 1450
Alexandria, Virginia 22313-1450
www .uspto.gov
ATTORNEY DOCKET NO. CONFIRMATION NO.
329405.01 8447
EXAMINER
BRANDENBURG, WILLIAM A
ART UNIT PAPER NUMBER
3681
NOTIFICATION DATE DELIVERY MODE
08/24/2018 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):
usdocket@microsoft.com
chriochs@microsoft.com
PTOL-90A (Rev. 04/07)
UNITED STATES PATENT AND TRADEMARK OFFICE
BEFORE THE PATENT TRIAL AND APPEAL BOARD
Exparte CHAO LIU and YI-MIN WANG
Appeal2017-004814
Application 12/764,983
Technology Center 3600
Before JEAN R. HOMERE, JOSEPH P. LENTIVECH, and
MICHAEL J. ENGLE, Administrative Patent Judges.
LENTIVECH, Administrative Patent Judge.
DECISION ON APPEAL
Pursuant to 35 U.S.C. § 134(a), Appellants 1 appeal from the
Examiner's decision to reject claims 1, 3, 5-11, and 13-23. Claims 2, 4, and
12 have been canceled. See App. Br. 21-27 (Claims App'x). We have
jurisdiction over the pending claims under 35 U.S.C. § 6(b ).
We reverse.
1 According to Appellants, the real party in interest is Microsoft Technology
Licensing, LLC. App. Br. 2.
Appeal2017-004814
Application 12/764,983
STATEMENT OF THE CASE
Appellants 'Invention
Appellants' invention generally relates to "learning a domain-specific
ranker that can be configured to rank entities that belong to a domain."
Spec. ,r 5. "[A]n entity can be a person, object, place, etc. and an entity may
belong to a certain domain that may be used to classify the entity. Spec.
,r 18. For example, a domain may be "actors" and particular actors may be
entities within the domain. Id. Claim 1, which is illustrative of the claimed
invention, reads as follows:
1. A method executed by at least one processor of at least one
computing device, the method comprising:
receiving, at a computer, a plurality of computer-readable
entities that each belong to a domain, wherein each computer-
readable entity in the plurality of computer-readable entities
represents an entity referenced in a computer-readable
document, each computer-readable entity comprises:
an identifier for the computer-readable entity; and
a feature vector that describes the entity represented
by the computer-readable entity, the feature vector has a
plurality of values therein, the plurality of values
respectively corresponds to a plurality of attributes of the
entity;
applying a preference rule over pairs of computer-readable
entities from amongst the plurality of computer-readable entities
to generate preference pairs, each preference pair identifies
which computer-readable entity in a pair of computer-readable
entities in the preference pair is preferable over the other
computer-readable entity in the pair of computer-readable
entities, the preference rule comprising a defined preference
between a first value of a first attribute in the plurality of
attributes and a second value of the first attribute in the plurality
of attributes; and
2
Appeal2017-004814
Application 12/764,983
learning a computer-implemented ranker that is specific to
the domain based upon the preference pairs, wherein the
computer-implemented ranker, subsequent to being learned, is
configured to output a ranked list of entities in the domain in
response to receipt of a query, wherein the acts of receiving,
applying, and learning are performed by the at least one
processor.
References
The Examiner relies on the following prior art in rejecting the claims:
Liao et al.
Chapelle et al.
US 2009/0006360 Al
US 2011/0016065 Al
Rejections
Jan. 1, 2009
Jan. 20, 2011
Claims 1, 3, 5-11, and 13-23 stand rejected under 35 U.S.C. § 101 as
directed to judicially-excepted subject matter under 35 U.S.C. § 101. Final
Act. 2--4.
Claims 1, 3, 5-7, 9-11, and 13-23 stand rejected under 35 U.S.C.
§ I02(b) as being anticipated by Liao. Final Act. 5-12.
Claim 8 stands rejected under 35 U.S.C. § I03(a) as being
unpatentable over the combination of Liao and Chapelle. Final Act. 13-15.
ANALYSIS
Rejection under 35 U.S.C. § 101
Under 35 U.S.C. § 101, a patent may be obtained for "any new and
useful process, machine, manufacture, or composition of matter, or any new
and useful improvement thereof." The Supreme Court has "long held that
this provision contains an important implicit exception: Laws of nature,
natural phenomena, and abstract ideas are not patentable." Alice Corp. v.
3
Appeal2017-004814
Application 12/764,983
CLS Bankint'l, 134 S. Ct. 2347, 2354 (2014) (quotingAss'nfor Molecular
Pathology v. Myriad Genetics, Inc., 569 U.S. 576, 589 (2013)). The
Supreme Court in Alice reiterated the two-step framework previously set
forth in Mayo Collaborative Services v. Prometheus Laboratories, Inc., 566
U.S. 66, 82-84 (2012), "for distinguishing patents that claim laws of nature,
natural phenomena, and abstract ideas from those that claim patent-eligible
applications of these concepts." Alice, 134 S. Ct. at 2355. The first step in
that analysis is to "determine whether the claims at issue are directed to one
of those patent-ineligible concepts," such as an abstract idea. The inquiry
often is whether the claims are directed to "a specific means or method" for
improving technology or whether they are simply directed to an abstract
end-result. McRO, Inc. v. Bandai Namco Games Am. Inc., 837 F.3d 1311,
1314 (Fed. Circ. 2016). If the claims are not directed to a patent-ineligible
concept, the inquiry ends. Otherwise, the inquiry proceeds to the second
step, where the elements of the claims are considered "individually and 'as
an ordered combination' to determine whether the additional elements
'transform the nature of the claim' into a patent-eligible application." Alice,
134 S. Ct. at 2355 (quoting Mayo, 566 U.S. at 79-80). We, therefore, look
to whether the claims focus on a specific means or method that improves the
relevant technology or are instead directed to a result or effect that itself is
the abstract idea and merely invoke generic processes and machinery. See
Enfzsh, LLC v. Microsoft Corp., 822 F.3d 1327, 1336 (Fed. Cir. 2016).
The Examiner determines the claims are directed to "outputting a
ranked list of entities according to a defined preference rule." Final Act. 3.
The Examiner determines the claims "compare[] new entity and feature
vector information with preference rules to determine and output a ranked
4
Appeal2017-004814
Application 12/764,983
list of entities" and, therefore, are directed to an abstract idea. Id. ( citing
SmartGene, Inc. v. Advanced Biological Labs., SA, 555 F. App'x 950 (Fed.
Cir. 2014) (unpublished)); see also Ans. 8 (citing Cybersource Corp. v.
Retail Decisions, Inc., 654 F.3d 1366 (Fed. Cir. 2011)).
Appellants argue the claims are not directed to an abstract idea but,
instead, are directed to "learning a domain-specific computer-implemented
ranker." App. Br. 7. Appellants argue learning a domain-specific,
computer-implemented ranker is not abstract because:
1) learning a computer-implemented ranker that is specific to
the domain necessarily occurs in the context of computers; 2) the
claims are directed to solving a problem that exists in a
computing environment; and 3) learning a computer-
implemented ranker that is specific to the domain cannot be
fairly characterized as corresponding to any of the judicial
descriptors of abstract ideas.
App. Br. 8; see also App. Br. 9-10.
For Mayo/Alice step two, Appellants argue
[The] claims recite features in addition to "outputting a ranked
list of entities according to a predefined preference rule."
Specifically, independent claim 1 (for example) recites applying
a preference rule over pairs of computer-readable entities from
amongst the plurality of computer-readable entities to generate
preference pairs ... ; and learning a computer-implemented
ranker that is specific to the domain based upon the preference
pairs. Therefore, even if the claims were deemed to be directed
towards outputting a ranked list of entities according to a
predefined preference rule, as alleged by the [Examiner], it is
clear that these claims recite additional features pertaining to
acquiring domain-specific training data and learning a domain-
specific ranker based upon such training data.
App. Br. 12. Appellants argue that these additional features "provide a
solution to the problem of conventional rankers lacking robustness and
5
Appeal2017-004814
Application 12/764,983
accuracy when ranking entities belonging to a specific domain" (App.
Br. 13) and "are directed to a new and useful application embodied through
the specific combination of novel, non-generic (i.e., not well known, routine,
or conventional) computer functions" (App. Br. 14).
We find Appellants' arguments persuasive. Turning to the
Examiner's rejection, we note that, apart from listing multiple concepts, the
Examiner does not identify specific claim limitations that the Examiner
believes may be an abstract idea. Final Act. 3--4; Ans. 2-8. In particular,
the Examiner determines:
Such as in Cybersource (Cybersource Corp. v. Retail
Decisions, Inc., 654 F.3d 1366 (Fed. Cir. 2011).), which obtains
and compares intangible data, the current invention obtains
vector-based entity data and compares multiple entities of that
obtained data together by applying a preference rule to learn a
ranker. Therefore an abstract idea is present in the claims.
Ans. 8. The Examiner does not persuasively explain why or how "learning a
computer-implemented ranker that is specific to the domain based upon the
preference pairs" (the "learning limitation"), as recited in independent claim
1, is a step of "obtaining and comparing intangible data." Next, we consider
whether the learning limitation amounts to significantly more under step two
of the analysis.
According to Appellants, the learning limitation contributes to the
invention's alleged improvement over the prior art. App. Br. 13-14; Reply
Br. 7. For example, Appellants' Specification provides "today's search
engines ... are not configured to rank entities belonging to a certain
domain" and that entity ranking is difficult because "entities in different
domains have quite different attributes and the weights [ assigned to the
attributes when performing ranking] are dependent on the attributes when
6
Appeal2017-004814
Application 12/764,983
entity ranking is undertaken." Spec. ,r 3. Appellants explain that to address
this difficulty, the claims "recite a specific approach for acquiring domain-
specific training data (applying a preference rule over pairs of computer-
readable entities that belong to a same domain), and further recite use of that
training data to solve a technological problem (the inability of conventional
rankers to act as robust domain-specific rankers)." Reply Br. 7 (citing
Ultramercial, Inc. v. Hulu, LLC, 772 F.3d 709 (Fed. Cir. 2014)).
"Whether something is well-understood, routine, and conventional to
a skilled artisan at the time of the patent is a factual determination."
Berkheimer v. HP Inc., 881 F.3d 1360, 1369 (Fed. Cir. 2018). Based on the
record before us, we agree with Appellants that the Examiner has not
adequately demonstrated that the learning limitation was a well-understood,
routine, conventional activity already engaged by the relevant community.
See Final Act. 2-3; Ans. 8-20. Because the Examiner has not adequately
demonstrated that fact, the Examiner has not established patent ineligibility.
Hence, we do not sustain the § 101 rejection of independent claims 1.
The remaining claims were rejected under the same rationale. Final
Act. 2-3; Ans. 2-20. Because we do not sustain the Examiner's rejection of
claim 1 as being directed to ineligible subject matter, we also do not sustain
the rejection of independent claims 13 and 20, which recite similar
limitations, and dependent claims 3, 5-11, 14--19, and 21-23 for the same
reasons. 2
2 Appellants raise additional arguments. See App. Br. 6-15. Because the
identified issue is dispositive of the rejection of claims 1, 3, 5-11, and 13-23
under 35 U.S.C. § 101, we do not reach the additional arguments.
7
Appeal2017-004814
Application 12/764,983
Rejection under 35 U.S.C. § 102(b)
Appellants contend Liao fails to disclose
applying a preference rule over pairs of computer-readable
entities from amongst the plurality of computer-readable entities
to generate preference pairs, each preference pair identifies
which computer-readable entity in a pair of computer-readable
entities in the preference pair is preferable over the other
computer-readable entity in the pair of computer-readable
entities,
as recited in claim 1. App. Br. 16-19, Reply Br. 12-15.
Liao relates to providing "query relaxative ranking for support vector
machines in order to provide more accurate and useful search results for a
user, client, or application." Liao ,r 16. To provide the search results, Liao
discloses obtaining a set of ranked query item pairs. Id. ,r 17. Each query
item pair corresponds to a query and a hit document. Id. Liao discloses that
each query item pair also corresponds to a set of feature vectors and that
each feature vector includes a set of features and corresponding feature
weights. Id.
The Examiner finds the features included in Liao' s feature vectors
disclose the claimed "entities." Ans. 23 (citing Liao ,r 35). The Examiner
finds Liao discloses "for the same query, thus for the same particular
'domain' as claimed, feature items of documents (i.e., 'entities' as claimed)
are evaluated and ranked relative to one another." Ans. 23 (citing Liao
,r 35). The Examiner finds Liao discloses analyzing the pair of feature items
to produce a relevancy score and applying weighted factors to learn a
ranking function via a machine learning application and training vectors and,
therefore, Liao discloses "'applying a preference rule' as claimed." Ans. 24
(citing Liao ,r,r 38-39). We disagree.
8
Appeal2017-004814
Application 12/764,983
Claim 1 recites that each computer-readable entity comprises "a
feature vector that describes the entity represented by the computer-readable
entity" and "has a plurality of values therein, the plurality of values
respectively correspond[ing] to a plurality of attributes of the entity."
We agree with Appellants (Reply Br. 13-14 ), that Liao teaches that the
features, which the Examiner relies upon for disclosing the claimed
"plurality of computer-readable entities" (Ans. 23), are included in a feature
vector that represents each item (Liao ,r 3 5) but not that the features
comprise a feature vector, as required by claim 1. Further, the Examiner's
findings are insufficient to show that Liao' s features have a plurality of
values respectively corresponding to a plurality of attributes of the entity, as
also required by claim 1.
Accordingly, we do not sustain the Examiner's rejection under 35
U.S.C. § 102(b) of claim 1, independent claims 13 and 20, which recite
corresponding limitations, and claims 3, 5-7, 9-11, 13-19, and 21-23,
which depend from claims 1 and 13. 3
Rejection under 35 U.S.C. § 103(a)
Claim 8 depends from claim 1. The Examiner does not find Chapelle
cures the deficiencies in the teachings of Liao discussed with respect to
claim 1. See Final Act. 13-14. Accordingly, we do not sustain the
Examiner's rejection of claim 8 under 35 U.S.C. § 103(a) for the reasons
discussed supra.
3 Appellants raise additional arguments. See App. Br. 15-19. Because the
identified issue is dispositive of the rejection of claims 1, 3, 5-7, 9-11, and
13-23 under 35 U.S.C. § 102(b), we do not reach the additional arguments.
9
Appeal2017-004814
Application 12/764,983
DECISION
We reverse the Examiner's rejection of claims 1, 3, 5-11, and 13-23
under 35 U.S.C. § 101.
We reverse the Examiner's rejection of claims 1, 3, 5-7, 9-11, and
13-23 under 35 U.S.C. § 102(b).
We reverse the Examiner's rejection of claim 8 under 35 U.S.C.
§ 103(a).
REVERSED
10