Ex Parte Kirshenbaum

Patent Trial and Appeal Board

Feb 28, 2017

13286024 (P.T.A.B. Feb. 28, 2017)

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.
13/286,024 10/31/2011 Evan R. Kirshenbaum 82859659 4111
56436 7590
Hewlett Packard Enterprise
3404 E. Harmony Road
Mail Stop 79
Fort Collins, CO 80528
EXAMINER
ROSTAMI, MOHAMMAD S
ART UNIT PAPER NUMBER
2154
NOTIFICATION DATE DELIVERY MODE
03/02/2017 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):
hpe.ip.mail@hpe.com
chris. mania @ hpe. com
PTOL-90A (Rev. 04/07)
UNITED STATES PATENT AND TRADEMARK OFFICE
BEFORE THE PATENT TRIAL AND APPEAL BOARD
Ex parte EVAN R. KIRSHENBAUM
Appeal 2016-005528
Application 13/286,024
Technology Center 2100
Before ALLEN R. MacDONALD, JOHN P. PINKERTON, and
GARTH D. BAER, Administrative Patent Judges.
MacDONALD, Administrative Patent Judge.
DECISION ON APPEAL
STATEMENT OF CASE
Appellant appeals under 35 U.S.C. § 134(a) from a Final Rejection of
claims 1—16. We have jurisdiction under 35 U.S.C. § 6(b).
We affirm.
Exemplary Claim
Exemplary claim 1 under appeal reads as follows (emphasis added):
1. A method for constructing an analysis of a document,
comprising:
determining a plurality of features based on the document,
wherein each of the plurality of features is associated with a
subset of a set of concepts;
Appeal 2016-005528
Application 13/286,024
constructing a set of concept candidates based on the
plurality of features, each concept candidate associated with at
least one concept in the set of concepts;
choosing a subset of the set of concept candidates as
winning concept candidates, including:
iteratively removing worst concept candidates
from the set of concept candidates; and
adding removed concept candidates to the winning
concept candidates based on votes for concept candidates from
within the set of concept candidates from each of a subset of
features from within the plurality of features,
wherein each feature within the subset of
features has a particular number of votes based on a weight
associated with the feature in a feature count map; and
constructing an analysis that includes at least one concept
in the set of concepts associated with at least one of the winning
concept candidates.
Rejections on Appeal
1. The Examiner rejected claims 1—12 and 14—16 under
35 U.S.C. § 103(a) as being unpatentable over the combination of Willse
(US 2008/0109454 Al; pub. May 8, 2008), Houghton
(US 2010/0293195; Al; pub. Nov. 18, 2010), andNakano (US
2003/0084022 Al; pub. May 1, 2003).1
1 The patentability of claims 2—9, 11—12, and 15—16 is not separately argued
from that of claims 1,10, and 14. See App. Br. 12. In Appellant’s argument
for patentability of claim 13, Appellant references his argument for
patentability of claim 10, and additionally argues there is no motivation to
combine Willse, Houghton, and Nakano with Thibault. See App. Br. 12—13.
We agree with the Examiner that the cited references of Willse, Houghton
and Nakano indicate a need for an efficient way to locate a document and
provide it to individuals (see, e.g., Willse 140), and thus, the cited
references indicate a motivation to combine Willse, Houghton, and Nakano
2
Appeal 2016-005528
Application 13/286,024
2. The Examiner rejected claim 13 under 35 U.S.C. § 103(a) as
being unpatentable over the combination of Willse, Houghton, Nakano, and
Thibault (US 2010/0070930 Al; pub. Mar. 18, 2010).
Appellant’s Contentions
1. Appellant contends that the Examiner erred in rejecting claims
1, 10, and 14 under 35 U.S.C. § 103(a) because:
Appellant’s concept candidates are associated with concepts
related to features. These features are extracted from documents
and include words, phrases, word sequences, characters, etc.
and/or information relating to document relationships. In
contrast, Willse’s concept representation is a multi-level acyclic
graph organization, with nodes of the graph corresponding to
concepts. Appellant respectfully submits that the two are not
analogous. Rather, Willse’s concept representation includes
one layer or level at a time in a hierarchical fashion from the
lowest to highest level concepts. Further, in contrast to
Appellant’s independent claims Willse’s concept
representation does not include a chosen winner. For instance,
Willse does not describe a winning concept candidate or the
process of choosing one from a set of candidates in an election.
Appellant’s winning concept candidate can include a candidate
associated with a concept most likely to be referred to by a
feature as one led to by text. In contrast, Willse describes
identifying a number of first level and second level concepts of
text documents, but these concepts are not described as being
chosen (as a winner) from a set of candidates in an election, as
provided in the independent claims; rather they are placed into
a concept representation after being identified as first or second
level concepts.
Similarly . . . the Examiner points to subsets of concepts in
Willse, but as noted, the Examiner has equated Willse’s concept
with Thibault. See Final Act. 26—27. Thus, except for our ultimate decision,
claims 2—9, 11—13, and 15—16 are not discussed further herein.
3
Appeal 2016-005528
Application 13/286,024
representations to Appellant’s concept candidates. As such,
subsets of concepts in Willse are not analogous to “choosing a
subset of the set of concept candidates” as provided in
Appellant’s independent claims.
In addition . . . the Examiner points to hierarchical subsets of a
concept representation in Willse as teaching a subset of the set of
concept candidates. Appellant respectfully disagrees. Even if the
elements were analogous (which Appellant does not admit), the
method of choosing the subsets is not analogous. The concept
representation subsets of Willse are determined based on finding
those that minimally disrupt the ‘goodness of fit’ as measured by
a likelihood function of the representation. This is not
analogous to “choose a subset of the set of concept candidates
as winning concept candidates using a feature weight and a
concept probability,” or “based on a conditional probability
between the first concept candidate and the second concept
candidate,” as provided in independent claims 10 and 14, for
examples.
Further, regarding independent claims 10 and 14 ... the
Examiner suggests that independent claims 10 and 14, “clearly
show a device for performing the methods of claim 1,” and
therefore does not provide citations in Willse for each limitation
in independent claims 10 and 14. Appellant respectfully submits
that the limitations are not the same, and hence the rejection is
improper. For example, Appellant respectfully submits that
Willse does not describe, “choose a subset of the set of concept
candidates as winning concept candidates using a feature weight
and a concept probability, wherein the feature weight indicates a
distribution of a feature in the document and the concept
probability includes a likelihood that a first concept candidate is
in the subset if a second concept candidate is in the subset,” as
provided in independent 10. Further, Willse does not describe,
“choose the first concept candidate as a winning concept
candidate based on a conditional probability between the first
concept candidate and the second concept candidate,” as
provided in independent claim 14.
4
Appeal 2016-005528
Application 13/286,024
App. Br. 7—9, Appellant’s emphasis and citations omitted, panel’s emphasis
added.
2. In the Reply Brief, further as to above contention 1, Appellant also
contends that the Examiner erred in rejecting claims 1,10, and 14 under
35 U.S.C. § 103(a) because:
[T]he Examiner . . . states, “Examiner hereby specifies that, the
highest weight for the concept are [sic] considered as the winning
candidate.” This is in contrast to Appellant’s independent
claims, which describe choosing a subset of the set of concept
candidates as winning concepts based on multiple factors
including iteratively removing worst concept candidates from
the set of concept candidates and adding removed concept
candidates based on votes for concept candidates. The “highest
weight” as provided by the Examiner is further in contrast to
Appellant’s independent claim 10 which describes choosing a
subset of the set of concept candidates as winning concept
candidates using a feature weight and a concept probability.
[T]he Examiner appears to be analogizing Appellant’s
features/votes to Willse’s features. However, as noted by the
Examiner, [Appellant’s] features are voters (not votes), in that
they cast votes for concept candidates; therefore even if
Appellant’s features were analogous to Willse’s features
(which the Appellant does not admit), Willse’s features would
not be analogous to Appellant’s votes, rather they would be
analogous to voters (again, which the Appellant does not
admit).
Reply Br. 3, Appellant’s emphasis and citations omitted, panel’s emphasis
added.
5
Appeal 2016-005528
Application 13/286,024
3. Appellant also contends that the Examiner erred in rejecting
claims 1,10, and 14 under 35 U.S.C. § 103(a) because:
Houghton describes that candidate entities (equated to
Appellant’s “features”) that have been rejected or discarded as
being unlikely to refer to any known entity may be added to
reference chains associated with known entities if those reference
chains have been constructed based on evidence from other
candidate entities that were not rejected or discarded (e.g., they
were previously discarded as false positives, but may be true
entity candidates).
In contrast, Appellant’s features (equated to Houghton’s
“candidate entities”) vote for concept candidates (equated to
Houghton’s “entities”) and have weights associated with them.
Further, Appellant’s worst concept candidate is removed,
whereas in Houghton, candidate entities (equated to
Appellant’s “features” or “votes”) are removed.
App. Br. 10, Appellant’s emphasis and citations omitted, panel’s emphasis
added.
4. In the Reply Brief, further as to above contention 3, Appellant
also contends that the Examiner erred in rejecting claims 1,10, and 14 under
35 U.S.C. § 103(a) because:
In the spirit of furthering prosecution, it is noted that as defined
[in] Houghton, a candidate entity is a reference (e.g., a word or
phrase) in a content item that meets a threshold probability of
corresponding to a known entity.
In contrast, Appellant’s concept candidates are associated with
concepts related to features. These features are extracted from
documents and include words, phrases, word sequences,
characters, etc. and/or information relating to document
relationships. Therefore, Houghton’s candidate entities are
not analogous to Appellant’s concept candidates.
Reply Br. 4 (Appellant’s citations omitted, panel’s emphasis added).
6
Appeal 2016-005528
Application 13/286,024
5. Appellant also contends that the Examiner erred in rejecting
claims 1, 10, and 14 under 35 U.S.C. § 103(a) because:
Nakano describes, “a weighted occurrence count of the keywords
corresponding to each evaluation section for the three (priority,
regular and auxiliary) correspondence tables.” As such, in
Nakano, the occurrence count (which the Examiner equates to
Appellant’s vote) is weighted. This is in contrast Appellant’s
independent claims, which provide that the votes are based on
a weight associated with a feature in a feature count map, as
provided in the independent claims.
App. Br. 11—12, Appellant’s emphasis and citations omitted, panel’s
emphasis added.
6. In the Reply Brief, further as to above contention 5, Appellant
also contends that the Examiner erred in rejecting claims 1,10, and 14 under
35 U.S.C. § 103(a) because:
The Examiner states, “Examiner has interpreted the number of
occurrences [in Nakano] as votes,” and also states, “Examiner
hereby specifies that the features associated with the concepts are
interpreted as votes per concept.” . . . However, for the same
reasons as noted with respect to votes versus voters, Appellant
respectfully disagrees with the feature-vote analogy.
Reply Br. 4, Appellant’s emphasis and citations omitted, panel’s emphasis
added.
Issue on Appeal
Did the Examiner err in rejecting claims 1,10, and 14 as being obvious?
7
Appeal 2016-005528
Application 13/286,024
ANALYSIS
We have reviewed the Examiner’s rejections in light of Appellant’s
arguments that the Examiner has erred. We disagree with Appellant’s
conclusions. Except as noted herein, we adopt as our own: (1) the findings
and reasons set forth by the Examiner in the action from which the appeal is
taken (Final Act. 2—28); and (2) the reasons set forth by the Examiner in the
Examiner’s Answer (Ans. 2—22) in response to the Appellant’s Appeal
Brief. We concur with the conclusions reached by the Examiner. We
highlight the following.
As to Appellant’s above contentions 1 and 2, we are not persuaded the
Examiner erred. Regarding Appellant’s argument that Willse’s concept
representation does not teach the claimed “concept candidate,” we disagree
with Appellant, as we agree with the Examiner that Willse teaches a concept
representation that is a representation of a set of concepts associated with a
set of text documents as a function of terminological features contained
within the text documents, and that the set of concepts contained within the
concept representation teaches the claimed “set of concept candidates based
on the plurality of features, each concept candidate associated with at least
one concept in the set of concepts.” See Ans. 2—\ (citing Willse 12, 14,
39, 77—79). Regarding Appellant’s argument that Willse does not describe a
winning concept candidate or process of choosing one from a set of
candidates in an election, Appellant’s argument is not commensurate with
the scope of the claims, as claims 1,10, and 14 each fail to recite the feature
that Appellant argues distinguishes the claims from Willse (i.e., choosing a
winning concept candidate from a set of candidates in an election).
8
Appeal 2016-005528
Application 13/286,024
Further, regarding Appellant’s argument that Willse’s method of
choosing concept representation subsets is based on finding those that
minimally disrupt a “goodness of fit” measured by a likelihood function of
representation and does not teach or suggest “choosing a subset of the set of
concept candidates,” as recited in claims 1,10, and 14, we do find this
argument persuasive. Appellant’s argument fails to persuasively distinguish
the claimed “[choosing] a subset of the set of concept candidates as winning
concept candidates using a feature weight and a concept probability,” as
recited in claim 10, and the claimed “[choosing] the first concept candidate
as a winning concept candidate based on a conditional probability between
the first concept candidate and the second concept candidate,” as recited in
claim 14, from Willse’s choosing hierarchical groupings (i.e., subsets) of
concepts within the concept representation based on a likelihood function of
the representation. See Ans. 4—6 (citing Willse 1 80). Regarding
Appellant’s argument that Willse’s features fails to teach “votes for concept
candidates,” as recited in the independent claims, we agree with the
Examiner that Willse’s teaching of an association of a feature with a concept
teaches the claimed “votes for concept candidates,” as Appellant’s argument
fails to persuasively distinguish the claimed “votes for concept candidates”
from Willse’s indications of associations between features and concepts.
See Ans. 12—13 (citing Willse 1 77).
Additionally, regarding Appellant’s argument that the rejection of
claims 10 and 14 in the Final Office Action is not proper because the
rejection relies upon the rejection of claim 1, yet claims 10 and 14 recite
limitations not found in claim 1, we do not find this argument persuasive.
While we agree with Appellant that claims 10 and 14 recite certain
9
Appeal 2016-005528
Application 13/286,024
limitations not recited in claim 1, and that the Final Office Action should
have included separate rejections for claims 10 and 14, rather than merely
referencing the rejection of claim 1, we further find that, in the Examiner’s
Answer, the Examiner correctly found Willse teaches “[choosing] a subset
of the set of concept candidates as winning concept candidates using a
feature weight and a concept probability,” as recited in claim 10, and the
claimed “[choosing] the first concept candidate as a winning concept
candidate based on a conditional probability between the first concept
candidate and the second concept candidate,” as recited in claim 14, and that
Appellant has not persuasively rebutted these findings. See Ans. 4—6.
As to Appellant’s above contentions 3 and 4, we are not persuaded the
Examiner erred. Appellant’s contentions are based on an underlying
argument that Houghton’s candidate entity does not teach the claimed
“concept candidate.” However, the Examiner relied upon Willse to teach the
claimed “concept candidate.” See Ans. 2-4. Thus, Appellant’s argument
attacks Houghton individually, rather than the Examiner’s combination of
references. It is well established that one cannot show non-obviousness by
attacking references individually where the rejection is based upon the
teachings of a combination of references. See In re Merck & Co., 800 F.2d
1091, 1097 (Fed. Cir. 1986); see also In re Keller 642 F.2d 413, 425 (CCPA
1981). As Appellant’s argument does not address the actual reasoning of the
Examiner’s rejection, we do not find it persuasive.
As to Appellant’s above contentions 5 and 6, we are not persuaded the
Examiner erred. Appellant’s argument fails to persuasively distinguish the
claimed “particular number of votes based on a weight associated with the
feature,” as recited in claim 1, from Nakano’s keyword occurrence count,
10
Appeal 2016-005528
Application 13/286,024
which is composed of a multiplication of a count of each extracted keyword
with a corresponding weighting. See Ans. 14—15 (citing Nakano 161).
CONCLUSIONS
(1) The Examiner has not erred in rejecting claims 1—16 as being
unpatentable under 35 U.S.C. § 103(a).
(2) Claims 1—16 are not patentable.
DECISION
We affirm the Examiner’s rejections of claims 1—16 as being
unpatentable under 35 U.S.C. § 103(a).
No time period for taking any subsequent action in connection with
this appeal may be extended under 37 C.F.R. § 1.136(a)(l)(iv).
AFFIRMED
11