Ex Parte Kim et al

Patent Trial and Appeal Board

Mar 30, 2017

13766019 (P.T.A.B. Mar. 30, 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.
13/766,019 02/13/2013 Hyun Duk Kim 83106283 4873
56436 7590
Hewlett Packard Enterprise
3404 E. Harmony Road
Mail Stop 79
Fort Collins, CO 80528
EXAMINER
MISHRA, RICHA
ART UNIT PAPER NUMBER
2673
NOTIFICATION DATE DELIVERY MODE
04/03/2017 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 HYUN DUK KIM, MARIA G. CASTELLANOS,
MEICHUN HSU, and CHENG XIANG ZHAI
Appeal 2016-005698
Application lS/766,0191
Technology Center 2600
Before JASON V. MORGAN, NABEEL U. KHAN, and
KAMRAN JIVANI, Administrative Patent Judges.
KHAN, Administrative Patent Judge.
DECISION ON APPEAL
Appellants seek our review, under 35 U.S.C. § 134(a), of the
Examiner’s final decision rejecting claims 1—20. App. Br., Claims
App’x. We have jurisdiction under 35 U.S.C. § 6(b).
We AFFIRM.
1 Appellants identify Hewlett-Packard Development Company, LP, as
real party in interest. App. Br. 1.
Appeal 2016-005698
Application 13/766,019
THE INVENTION
The present invention relates to “[a] technique [that] may include
generating a plurality of segments from sentences in a data set. . . [and]
further include determining the explanatoriness of each segment.” Abst.
Independent claims 1,12, and 14 are reproduced below.
1. A method, comprising:
generating, using at least one processor, a plurality of segments
from sentences in a first data set related to an opinion, the
plurality of segments including at least some segments that are
smaller than a sentence from which it was generated;
determining, using the at least one processor, an explanatoriness
score of each segment, wherein determining the explanatoriness
of each segment includes at least evaluating the
discriminativeness of features of the respective segment by
comparing the features to a second data set, wherein the
explanatoriness score of each segment indicates a likelihood that
the segment describes a reason for the opinion; and
ranking, using the at least one processor, the plurality of
segments according to their explanatoriness scores.
12. A system, comprising:
a segment generator executed by at least one processor to
generate a parse tree for each sentence in a first data set and
generate a plurality of segments from the parse trees, wherein the
first data set is associated with an opinion of a product or service;
an explanatoriness scorer executed by the at least one processor
to generate an explanatoriness score of each segment based on
an explanatoriness evaluation, the explanatoriness evaluation
including comparing words in each segment to words in a second
data set, wherein the explanatoriness score of each segment
indicates a likelihood that the segment describes a reason for the
opinion; and
2
Appeal 2016-005698
Application 13/766,019
a summary generator executed by the at least one processor to
generate a summary of the first data set based on the
explanatoriness scores, the summary including a subset of the
plurality of segments.
14. A non-transitory computer readable storage medium
storing instructions that, when executed by a processor, cause a
computer to:
generate a parse tree for each sentence in a data set, the data set
related to an opinion;
generate a plurality of segments from the parse trees, wherein at
least some of the segments are shorter than a sentence from
which they were generated;
determine an explanatoriness score for each segment, the
explanatoriness score indicating a likelihood that the respective
segment describes a reason for the opinion; and
rank the plurality of segments according to their explanatoriness
scores.
THE REJECTIONS
Claims 1—20 stand rejected, under 35 U.S.C. § 101, as being directed
to patent-ineligible subject matter. Final Act. 3^4 (May 12, 2015).
Claims 1—9 stand rejected, under 35 U.S.C. § 103(a), as being obvious
over Sharp et al. (US 2011/0093467 Al; Apr. 21, 2011) (“Sharp”) and
further in view of Reisman et al. (US 2009/0265307 Al; Oct. 22, 2009)
(“Reisman”). Final Act. 4—6.
Claim 10 stands rejected, under 35 U.S.C. § 103(a), as being obvious
over Sharp, Reisman, and Huang et al. (US 2008/0215571 Al;
Sept. 4, 2008) (“Huang”). Final Act. 6.
3
Appeal 2016-005698
Application 13/766,019
Claim 11 stands rejected, under 35 U.S.C. § 103(a), as being obvious
over Sharp, Reisman, Huang, and Reis et al. (US 2009/0193328 Al; July
30, 2009) (“Reis”). Final Act. 7.
Claims 12 and 13 stand rejected, under 35 U.S.C. § 103(a), as being
obvious over Reis, Knoll et al. (US 2003/0216904; Nov. 20, 2003)
(“Knoll”), and Reisman. Final Act. 7—9.
Claim 19 stands rejected, under 35 U.S.C. § 103(a), as being obvious
over Reis, Knoll, Reisman, and Huang. Final Act. 9.
Claim 20 stands rejected, under 35 U.S.C. § 103(a), as being obvious
over Reis, Knoll, Reisman, and Sharp. Final Act. 9—10.
Claims 14 and 18 stand rejected, under 35 U.S.C. § 103(a), as being
obvious over Huang and Knoll. Final Act. 10-11.
Claim 15 stands rejected, under 35 U.S.C. § 103(a), as being obvious
over Huang, Knoll, and Reis. Final Act. 11.
Claims 16 and 17 stand rejected, under 35 U.S.C. § 103(a), as being
obvious over Huang, Knoll, and Sharp. Final Act. 11—12.
35 U.S.C. § 101
Claims 1—20 stand rejected as drawn to patent ineligible subject
matter. Appellants address the claims collectively. We select claim 14 as
representative. 37 C.F.R. § 41.37(c)(l)(iv) (representative claims). For the
reasons below, Appellants fail to show error in the rejection of claim 14.
The Examiner analyzed the claims under the Supreme Court’s two-
step framework for determining whether claimed subject matter is judicially-
excepted from patent eligibility under § 101. Ans. 23; Final Act. 3; see also
Alice Corp. Pty. Ltd. v. CLS Bank Inti, 134 S. Ct. 2347, 2355 (2014)
(discussing the framework per its introduction by Mayo Collaborative Servs.
4
Appeal 2016-005698
Application 13/766,019
v. Prometheus Labs., Inc., 132 S. Ct. 1289 (2012)). Performing the first
step, the Examiner finds the claims recite an abstract idea of determining a
sentence segment’s likelihood of describing a reason for an opinion and
accordant ranking of the segment. Id. Performing the second step, the
Examiner finds the additionally claimed elements “amount to no more than
recitation of generic software code to perform generic functions” for such
determining and ranking. Id.
Appellants argue the Examiner fails to support the above
findings. App. Br. 18—20; Reply Br. 11—12. The arguments, below, are also
directed to the two steps of the analysis under Alice.
As to the first step, Appellants contend the Examiner fails to establish
the claims are directed to an abstract idea because the findings have “failed
to establish that the asserted abstract idea ... is not similar to any abstract
idea previously identified by the courts.” App. Br. 18; see also Reply
Br. 11. In support, Appellants further contend “the asserted abstract idea
clearly fails to conform to the U.S. Supreme Court’s descriptions of abstract
ideas as ‘the basic tools of scientific and technological work’ and the
‘building blocks of human ingenuity.’” App. Br. 18 (quoting Alice, 134 S.
Ct. at 2354); see also Reply Br. 12.
As to the second step, Appellants contend the Examiner erred in
finding the claims lack an inventive concept because the granularity of the
claimed segments (“smaller than a sentence”) and measurement of their
claimed score (“likelihood . . . segment describes a reason for the opinion”)
constitute significantly more than the alleged abstract idea and generic
software. App. Br. 19. In support, Appellants present two further
contentions. Id. at 19—20. First, Appellants contend the measurement is
5
Appeal 2016-005698
Application 13/766,019
undisclosed by the applied art and, thus, not shown to be generic. Id. at
19. Second, Appellants contend the granularity prevents the claims from
“tying up” use of the measurement and, thus, prevents preemption of the
alleged abstract idea. Id. at 19—20 (emphasis omitted).
We address the above contentions in seriatim, below.
Contention 1: Examiner has not shown
the alleged abstract idea is similar to an abstract idea
previously identified by the courts.
Claim 1 recites three steps: generating segments from sentences;
determining the explanatoriness score of each segment; and ranking the
segments via the scores. All three steps generate information sets—namely
segments, respective scores, and accordant rankings—from a prior
information set. Such generating of information, from prior generated
information, is plainly an abstract idea category of judicially excepted
subject matter. See, e.g., Bascom Glob. Internet Servs., Inc. v. AT&T
Mobility LLC, 827 F.3d 1341, 1350 (Fed. Cir. 2016) (“filtering content”);
Enfish, LLC v. Microsoft Corp., 822 F.3d 1327, 1337—38 (Fed. Cir. 2016)
(“organizing information using tabular formats”); Digitech Image Techs.,
LLCv. Elecs. for Imaging, Inc., 758 F.3d 1344, 1350 (Fed. Cir. 2014)
(“organizing information through mathematical correlations”); Content
Extraction & Transmission LLC v. Wells Fargo Bank, Nat. Ass ’n, 116 F.3d
1343, 1347 (Fed. Cir. 2014) (“1) collecting data, 2) recognizing certain data
within the collected data set, and 3) storing that recognized data in a
memory”).
Moreover, the Specification elaborates that text “[sjegments may be
evaluated for explanatoriness in a variety of ways” {id. 123) including,
“based on the conditional probability” equation of Bayes Rule {id. H 27—
6
Appeal 2016-005698
Application 13/766,019
28), steps of “using a unigram language model and taking logarithm of both
[Bayes Rule equation] sides to obtain [an] explanatoriness scoring function”
{id. 130). As noted above, “organizing information through mathematical
correlations” constitutes an abstract idea. Digitech, 758 F.3d at 1350. Thus,
“[w]ithout additional limitations, a process that employs mathematical
algorithms to manipulate existing information to generate additional
information is not patent eligible.” Id. at 1351. And, “‘[i]f a claim is
directed essentially to a method of calculating, using a mathematical
formula, even if the solution is for a specific purpose, the claimed method is
nonstatutory.’” Id. The claimed invention plainly falls within the above
categories of abstract ideas.
Contention 2: Examiner has not shown the alleged abstract
idea is a basic tool of scientific/technological work
or building block of human ingenuity.
As discussed above, claim 1 recites generating information from prior
generated information, and performing calculations on that information,
which is a fundamental building block of research and human ingenuity. See
supra.
Contention 3: Claimed measurement is not generic and thus
adds “significantly more ” to the alleged abstract idea.
As reflected supra, “there may be close calls about how to
characterize what the claims are directed to.” Bascom, 827 F.3d at 1348
(quoting Enfish, 822 F.3d at 1339-40). We thus consider whether the
recited granularity and measurement add “specific improvements in the
recited computer technology [that] go beyond ‘well-understood, routine,
7
Appeal 2016-005698
Application 13/766,019
conventional activities] ’ and render the invention patent-eligible.”
Bascom, 827 F.3d at 1348. That is, because “the claims and their specific
limitations do not readily lend themselves to a step-one finding that they are
directed to a nonabstract idea[, we] therefore defer our consideration of the
specific claim limitations’ narrowing effect for step two” of the Alice two
step framework. Id. at 1349.
There is no evidence, before us, the recited granularity and
measurement provide a “specific technical solution” in computer technology
{id. at 1352) much less provide an unconventional, technical solution (id. at
1348). For example, Appellants present no evidence that parsing of text into
sentences and sentence segments (e.g., separating semi-colon phrases)
improved computer technology or was unconventional itself or in
combination with the recited measurement.
The Specification also lacks such evidence, merely asserting “[t]he
inventors have discovered that using a parse tree to identify segment
boundaries may be beneficial because explanatory phrase boundary is likely
to match with syntax boundary.” Spec. 12; see also id. 119 (“[A] single
sentence may have both relevant and irrelevant information.”). Such a mere
allegation of “discovery” cannot alone persuade us the matched combination
of granularity (explanatory phrase) and measurement (explanatory
likelihood) constitutes an unconventional, technical solution. See
Bascom, 827 F.3d at 1348; see also In re Lindner, 457 F.2d 506, 508 (CCPA
1972) (“The affidavit and specification do contain allegations that
8
Appeal 2016-005698
Application 13/766,019
synergistic results are obtained with all the claimed compositions, but those
statements are not supported by any factual evidence.”).2
Further, the parsing of sentences into segments and subsequent
determination of explanatoriness score of those segments and ranking the
segments according to the explanatoriness scores are forms of mental steps
or calculations. Thus, these steps fail to transform the claim into something
more than an abstract idea.
Contention 4: Claimed granularity prevents
preemption of the alleged abstract idea.
“While preemption may signal patent ineligible subject matter, the
absence of complete preemption does not demonstrate patent eligibility.”
Ariosa Diagnostics, Inc. v. Sequenom, Inc., 788 F.3d 1371, 1379
(Fed. Cir. 2015); see also OIP Techs., Inc. v. Amazon.com, Inc., 788 F.3d
1359, 1362-63 (Fed. Cir. 2015), cert, denied, 136 S. Ct. 701, 193 (2015)
(“[T]hat the claims do not preempt all price optimization or may be limited
to price optimization in the e-commerce setting do not make them any less
abstract.”). And, “[wjhere a patent’s claims are deemed only to disclose
patent ineligible subject matter under the Mayo framework, as they are in
this case, preemption concerns are fully addressed and made moot.” Ariosa,
788 F.3d at 1379.
Conclusion
Accordingly, for the foregoing reasons, we sustain the rejection of
claims 1—20 under 35 U.S.C. § 101.
2 The assertion of such a “discovery” is also dubious because artisans, and
even laypersons, would know syntax boundaries (e.g., semi colon) signal
changes in idea, tone, opinion, etc.
9
Appeal 2016-005698
Application 13/766,019
35 U.S.C. § 103(a)
Claims 1—11
Independent claim 1 and depending claims 2—11 stand rejected as
obvious over Sharp and Reisman (and further references for claims 10 and
11). Sharp’s cited disclosure (see infra) generally teaches parsing of
sentences into segments and associating of some segments with respective
objects (e.g., with “cup”). See, e.g., Sharp || 104—106. Reisman’s cited
disclosure (see infra) generally teaches gathering of opinions on a topic
(e.g., a retail product) and generating of both a numerical score and textual
summary of sentiment for the topic. See, e.g., Reisman || 3, 38, 81.
The Examiner finds Sharp teaches parsing and scoring of text
segments smaller than a sentence. Final Act. 4 (citing Sharp 88, 104—
105, 156, 161). The Examiner finds Reisman teaches scoring of a text
segment’s likelihood of describing a reason for an opinion. Id. at 5 (citing
Reisman || 3, 28, 48, 81). Combining these alleged teachings, the Examiner
finds it would have been obvious to add Reisman’s scoring to Sharp’s
invention so as to additionally “generate [Reisman’s] fluent textual summary
which takes multiple feature into account.” Id. (citing Reisman || 3, 28);
see also Ans. 14 (reiterating the rationale without elaboration or revision).
Appellants argue:
[I]t is abundantly clear that the cited portions of Reisman say
nothing whatsoever about any score of each
segment. Further, the cited portions of Reisman say nothing
whatsoever about a score that “indicates a likelihood that the
segment describes a reason for the opinion.” Rather, the cited
material describes something substantially different, namely
generating a “summary” that summarizes multiple opinions
related to the same topic. Note that, because the “summary” is
10
Appeal 2016-005698
Application 13/766,019
generated from multiple opinions, it is clear that the “summary”
is not a “explanatoriness score of each segment,” and clearly
does not indicate “a likelihood that the segment describes a
reason for the opinion.”
App. Br. 8 (emphasis omitted); see also Reply Br. 2-4.
We are unpersuaded by Appellants’ arguments. The Examiner relies
upon Sharp as teaching or suggesting generating sentence segments and
determining an explanatoriness score for the segments. Final Act. 4. The
Examiner relies upon Reisman as teaching that the explanatoriness score
indicates a likelihood that the segment describes a reason for the opinion.
Id. at 5. Appellants do not dispute the Examiner’s findings with respect to
Sharp. See App. Br. 7—9. Instead, Appellants argue that Reisman is directed
at summarizing multiple opinions related to a topic rather than determining
an explanatoriness score of each segment, where the explanatoriness score
indicates a likelihood that the segment describes a reason for the opinion.
Such arguments attack Reisman individually and fail to account for the
Examiner’s findings as a whole.
As state above, the Examiner relies upon Sharp for teaching
generating sentence segments and determining an explanatoriness score for
the segments. Final Act. 4. Reisman teaches that textual opinions can be
analyzed to determine “what people think about topic X; how much people
liked or disliked X; why they liked or disliked about X. . . .” Reisman 128
(emphasis added). We agree with the Examiner that by teaching a
determination of why people liked or disliked a topic, Reisman teaches
indicating a likelihood that the text describes a reason for the opinion (i.e., a
reason why the topic was liked or disliked). When combined with the
Examiner’s findings regarding Sharp, we agree that the combination teaches
11
Appeal 2016-005698
Application 13/766,019
“determining ... an explanatoriness score of each segment,. . . wherein the
explanatoriness score of each segment indicates a likelihood that the
segment describes a reason for the opinion,” as recited in claim 1.
Characterizing the Examiner’s rationale as asserting “that it would
have been obvious to modify Sharp to include the missing subject matter
(i.e. the ‘summary’ of Reisman) because this would provide the missing
subject matter (i.e. ‘to generate fluent textual summary’),” Appellants
further argue that the Examiner’s rationale for combining Sharp with
Reisman “is merely circular logic.” App. Br. 9 (emphasis omitted). The
Examiner finds “[i]t would have been obvious having the concept of Sharp
to further include the concept of Reisman to generate a fluent texual [sic]
summary which takes multiple feature[s] into account.” Final Act. 5
(emphasis omitted) (citing Reisman || 3, 28). We disagree that this is
circular logic. Instead, the Examiner merely states that it would have been
obvious to add a feature from Reisman (the textual summaries) to Sharp to
take multiple features of the opinions into account. Rather than circular
logic, this is a simple addition of a feature from one reference to another.
Accordingly, for the foregoing reasons, we sustain the rejection of
claims 1—11 under 35 U.S.C. § 103(a).
Claims 12, 13, 19, and 20
Independent claim 12 and depending claims 13, 19, and 20 stand
rejected as obvious over Reis, Knoll, and Reisman (and further references
for claims 19 and 20). Appellants present patentability arguments for only
claim 12 and reference those arguments for depending claims 13, 19, and 20
(App. Br. 13—15).
12
Appeal 2016-005698
Application 13/766,019
Appellants again argue, now for claim 12, “Reisman says nothing
whatsoever about an ‘explanatoriness score of each segment[.]’
. . . Rather, at best, [Reisman] describes ... a ‘sentiment score’ that
corresponds to each attribute of a topic.” App. Br. 13 (emphasis omitted).
Appellants also argue, for claim 1 and now for claim 12, Reisman’s scoring
does not indicate the likelihood that a segment describes a reason for an
opinion. App. Br. 13—14 (claim 12); see also id. at 8—9 (claim 1).
We disagree with Appellants’ arguments for the same reasons stated
with respect to claim 1. In particular, Appellants again attack Reisman
individually rather than address the Examiner’s findings as a whole. The
Examiner finds Reis and Knoll teach or suggest generating segments from
sentences and Reis as teaching or suggesting an explanatoriness scorer to
generate an explanatoriness score for each segment. Final Act. 7. Thus,
Appellants argument that “Reisman says nothing whatsoever about an
‘explanatoriness score of each segment’” is unpersuasive because such a
finding is addressed by Reis. As to Appellants’ argument that Reisman does
not indicate the likelihood that a segment describes a reason for an opinion,
we disagree as explained above, namely, Reisman’s disclosure of analyzing
an opinion to determine why a person liked or disliked a topic teaches
indicating a likelihood that a segment describes a reason for an opinion.
Accordingly, for the foregoing reasons, we sustain the rejection of
claims 12, 13, 19, and 20 under 35 U.S.C. § 103(a).
Claims 14—18
Independent claim 14 and depending claims 15—18 stand rejected as
obvious over Huang and Knoll (and further references for claims 15—
13
Appeal 2016-005698
Application 13/766,019
17). Appellants present patentability arguments for only claim 14 and
reference those arguments for depending claims 15—18 (App. Br. 17—18).
The Examiner finds Huang teaches all but the claimed generating of a
parse tree for each sentence; relying on Knoll therefor. Final Act. 10.
Appellants argue:
Huang says nothing whatsoever about the cited “affinity rank”
being a score for each segment, or being a “score indicating a
likelihood that the respective segment describes a reason for the
opinion.” Rather, at best, Huang describes something
different, namely that the “affinity rank” refers to measures of
the “richness” and “diversity” of topics included in a set of search
results.
App. Br. 16 (emphasis omitted). Addressing the Examiner’s focus on a
“richness” aspect of Huang’s affinity rank (Ans. 20-21), Appellants add in
the Reply Brief:
[A]s best understood, the Answer apparently argued that the
“richness in opinion” discussed in Huang discloses a “likelihood
of opinion,” and thus somehow discloses the aforementioned
subject matter of claim 14.
First, it is noted that claim 14 actually recites “the
explanatoriness score indicating a likelihood that the respective
segment describes a reason for the opinion.” However, the
asserted phrase “likelihood of opinion” in the Answer clearly
misinterprets this claim limitation. . . .
Second, the conclusory assertion that “richness in
opinion” discloses a “likelihood of opinion” is not supported
. . . [and] appears to be erroneous. . . . [T]he term “richness”
appears to refer to the number of opinions or topics included in a
single document.... However, a person of ordinary skill in the
art will readily appreciate that the number of opinions in a single
document does not disclose or suggest the asserted “likelihood
14
Appeal 2016-005698
Application 13/766,019
of opinion” [or claimed] . . . “likelihood that the respective
segment describes a reason for the opinion[.]”
Third, . . . [the] number of opinions included in an entire
document ... is not a “score for each segment,” and does not
indicate “a likelihood that the respective segment describes a
reason for the opinion.”
Reply Br. 9—10 (emphasis omitted).
We are not persuaded by Appellants’ arguments. Huang’s affinity
ranking and included richness measure are part of several operations for
presenting opinion snippets. See Huang || 40-69. First, the system filters
subjective statements (i.e., removing objective statements) from returned
“documents” such as Internet product reviews. Huang || 6, 41. Second, the
system ranks, weights, etc., each document based in part upon “how many
different topics [the] single document contains[.]” Id. 1 51; see also id. 148.
Third, the system weights the snippets of each document. Id. 1 59.
Fourth, the system displays the documents by rank and accompanied by the
respective best snippets. Id. 171. These steps are expressly disclosed as
yielding snippets more “helpful... to understand the actual reviews or
ratings of the target product” {id. 1 5) and better “directed towards the
product reviewf.]” Id. 1 6. Thus, Huang affinity ranking (i.e., score)
identifies snippets with a higher likelihood of indicating an opinion and
reason therefor.
Accordingly, for the foregoing reasons, we sustain the rejection of
claims 14—18 under 35 U.S.C. § 103(a).
15
Appeal 2016-005698
Application 13/766,019
DECISION
The Examiner’s rejections of claims 1—20 under 35 U.S.C. § 101 are
affirmed.
The Examiner’s rejections of claims 1—20 under 35 U.S.C. § 103(a)
are affirmed.
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
16