Ex Parte Proux

Patent Trial and Appeal Board

Jan 26, 2018

13974515 (P.T.A.B. Jan. 26, 2018)

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.
13/974,515 08/23/2013 Denys Proux 20130948USNP-CNDT3039US01 1031
144578 7590 01/26/2018
FAY SHARPE LLP / CONDUENT
1228 Euclid Avenue, 5th Floor
The Halle Building
Cleveland, OH 44115
EXAMINER
WOZNIAK, JAMES S
ART UNIT PAPER NUMBER
2657
MAIL DATE DELIVERY MODE
01/26/2018 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.
PTOL-90A (Rev. 04/07)
UNITED STATES PATENT AND TRADEMARK OFFICE
BEFORE THE PATENT TRIAL AND APPEAL BOARD
Ex parte DENYS PROUX1
Appeal 2017-002867
Application 13/974,515
Technology Center 2600
Before CAROLYN D. THOMAS, AARON W. MOORE, and
DAVID J. CUTITTAII, Administrative Patent Judges.
THOMAS, Administrative Patent Judge.
DECISION ON APPEAL
Appellant seeks our review under 35 U.S.C. § 134(a) of the
Examiner’s Final Rejection of claims 1—20, all the pending claims in the
present application. See Claims Appendix. We have jurisdiction over the
appeal under 35 U.S.C. § 6(b).
We AFFIRM.
The present invention relates generally to speech to text transcription
using a knowledge base containing solution descriptions. See Abstract.
1 Appellant names Xerox Corporation as the real party in interest (App. Br.
1).
Appeal 2017-002867
Application 13/974,515
Claim 1 is illustrative:
1. A method for speech to text transcription comprising:
providing access to a knowledge base containing solution
descriptions, each solution description including a textual description
of a solution to a respective problem;
with a speech to text decoder, generating a preliminary
transcription of at least an agent’s part of an audio recording of a
dialogue between the agent and a user in which the agent had access
to the knowledge base, the generating comprising:
identifying a sequence of phonemes based on the
agent’s part of the audio recording, and
based on the identified sequence of phonemes,
generating the preliminary transcription, the preliminary
transcription including a sequence of words recognized as
corresponding to phonemes in the sequence of phonemes and
unrecognized phonemes from the phoneme sequence that are
not recognized as corresponding to one of the recognized
words; and
revising the preliminary transcription, the revising
comprising replacement of unrecognized phonemes with at
least one word from one of the solution descriptions in the
knowledge base, the solution description including words
which match words of the sequence of recognized words,
wherein at least one of the generating of the preliminary
transcription and the revising of the preliminary transcription is
performed with a processor.
Appellant appeals the following rejections:2
Rl. Claims 1, 2, 5, 7—9, 11, 12, and 15—20 are rejected under
35 U.S.C. § 103(a) as being unpatentable over Chen (US 2013/0080177 Al,
Mar. 28, 2013), Jeong (US 2007/0185713 Al, Aug. 9, 2007), and Byrd (US
2010/0274618 Al, Oct. 28, 2010) (id. at 7-13);
R2. Claims 3 and 6 are rejected under 35 U.S.C. § 103(a) as being
2 The Examiner withdrew the rejection of claim 20 under 35 U.S.C. § 101
(see Ans. 4).
2
Appeal 2017-002867
Application 13/974,515
unpatentable over Chen, Jeong, Byrd, and Stewart (US 7,437,291 Bl, Oct.
14, 2008) (id. at 13-14);
R3. Claim 4 is rejected under 35 U.S.C. § 103(a) as being
unpatentable over Chen, Jeong, Byrd, Stewart, and Kuspa (US
2013/0124984 Al, May 16, 2013) (id. at 14-15);
R4. Claim 10 is rejected under 35 U.S.C. § 103(a) as being
unpatentable over Chen, Jeong, Byrd, and Duke (US 2008/0082336 Al, Apr.
3, 2008) (id. at 15—16); and
R5. Claims 13 and 14 are rejected under 35 U.S.C. § 103(a) as being
unpatentable over Chen, Jeong, Byrd, and Park (US 2010/0104086 Al, Apr.
29, 2010) (id. at 16).
We review the appealed rejections for error based upon the issues
identified by Appellant, and in light of the arguments and evidence produced
thereon. Ex parte Frye, 94 USPQ2d 1072, 1075 (BPAI 2010) (precedential).
ANALYSIS
Rejection under § 103(a)
Claims 1, 11—14, and 16—20
Issue: Did the Examiner err in finding that Chen, Jeong, and Byrd
collectively teaches or suggests based on the identified sequence of
phonemes, generating the preliminary transcription, as set forth in claim 1?
Appellant contends that “[tjhere is no suggestion in Chen of
generating the text 203 from an identified sequence of phonemes” (App.
Br. 12), “[tjhere is no suggestion of generating a preliminary transcription”
in Jeong (id.), and Byrd “does not mention phonemes” (id.). Appellant
further contends that “[ejven assuming that the preliminary transcription of
3
Appeal 2017-002867
Application 13/974,515
Chen were to be generated by phoneme recognition, as the Examiner asserts,
there is no suggestion that the preliminary transcription of Chen would also
include unrecognized phonemes” (App. Br. 16).
In response, the Examiner finds that “Jeong was never relied upon to
provide a preliminary transcription. . . . Jeong was described as ‘merely
add[ing] the well-known technique ... for generating word candidates using
phoneme recognition’” (Ans. 14) and that “[t]he primary reference Chen . . .
already shows the generation of a preliminary or ‘initial transcription’ by the
use of a speech recognition system which is to be repaired . . . [and] the
argued limitation is taught by the combination” {id. at 14—15). The
Examiner further finds that “Chen does teach unrecognized phonemes
because Chen teaches unrecognized words” {id. at 15). We agree with the
Examiner.
We refer to, rely on, and adopt the Examiner’s findings and
conclusions set forth in the Answer. Our discussions here will be limited to
the following points of emphasis.
As an initial matter, we note that Appellant’s arguments do not take
into account what the collective teachings of the prior art would have
suggested to one of ordinary skill in the art and are therefore ineffective to
rebut the Examiner’s prima facie case of obviousness. See In re Keller, 642
F.2d 413, 425 (CCPA 1981) (“The test for obviousness is not whether the
features of a secondary reference may be bodily incorporated into the
stmcture of the primary reference; nor is it that the claimed invention must
be expressly suggested in any one or all of the references. Rather, the test is
what the combined teachings of the references would have suggested to
those of ordinary skill in the art.”) (citations omitted). This reasoning is
4
Appeal 2017-002867
Application 13/974,515
applicable here because Appellant primarily argues that features are missing
from one reference when the Examiner is relying on another reference for
such features, e.g., Appellant argues: (1) there is no suggestion of generating
a preliminary transcription in Jeong when the Examiner is relying on Chen;
and (2) Byrd does not mention phonemes when the Examiner is relying on
Jeong.
Furthermore, regarding the Examiner’s finding that “some sequences
of speech do not map to words that are recognized/knowri’'' (Final Act. 8)
and “Chen does teach unrecognized phonemes because Chen teaches
unrecognized words” (Ans. 15), Appellant merely contends that “Chen
makes no suggestion of there being unrecognized phonemes” (see App.
Br. 16) without specifically rebutting the Examiner’s aforementioned
findings about “unrecognized words.” In addition, we agree with the
Examiner that Appellant’s arguments fail to take into account what the
collective teachings of Chen and Jeong would have suggested to one of
ordinary skill in the art (see Ans. 16). Thus, we find Appellant’s arguments
unpersuasive.
Appellant further contends that “[combining Jeong with Chen would
serve merely to provide the preliminary transcript, and would not aid in
matching contacts in a database. Thus, there would have been no reason to
combine Jeong with Chen” (App. Br. 18). In response, the Examiner finds,
and we agree, that the “‘secondary reference Jeong provides the well-known
technique of phoneme-based speech recognition which is easily substituted
into the teachings of Chen . . . which yields a predictable result of providing
the phonemes needed for repair in the speech recognition process’”
(Ans. 19). “The combination of familiar elements according to known
5
Appeal 2017-002867
Application 13/974,515
methods is likely to be obvious when it does no more than yield predictable
results,” KSRInt’l Co. v. Teleflex, Inc., 550 U.S. 398, 416 (2007), especially
if the combination would not be “uniquely challenging or difficult for one of
ordinary skill in the art,” Leapfrog Enters., Inc. v. Fisher-Price, Inc., 485
F.3d 1157, 1162 (Fed. Cir. 2007) (citing KSR, 550 U.S. at 418).
The Examiner has found actual teachings in the prior art and has
provided a rationale for the combination, i.e., in order to substitute similar
elements for achieving predictable results of detecting valid words although
also providing a manner in which to acquire the phonemes needed for the
phonetic matching of unknown words (see Final Act. 9). Further, we agree
with the Examiner that the teachings suggest that the combination involves
the predictable use of prior art elements according to their established
functions. Accordingly, we find that the Examiner has provided sufficient
motivation for modifying Chen with the teachings of Jeong.
Finally, Appellant contends that “in Byrd, the knowledge base is not
used to transcribe the customer-agent spoken interaction” (App. Br. 19) and
“it would not have been obvious to use a database such as Byrd’s to repair
the transcript of Chen” (id. at 19—20). In response, the Examiner finds that
“Byrd teaches the same type of solution database as is claimed which
provides solution descriptions for different problems in customer-agent
transcriptions [and] [cjombining Byrd and Chen and Jeong only involves a
database substitution and the algorithm used for repair would remain intact”
(Ans. 20). We agree with the Examiner that the combined teachings of
Chen, Jeong, and Byrd teaches and/or suggests the argued limitation and
Appellant’s aforementioned arguments fail to consider the combined
teachings.
6
Appeal 2017-002867
Application 13/974,515
Accordingly, we sustain the Examiner’s rejection of claim 1.
Appellant’s arguments regarding the Examiner’s rejection of independent
claims 18 and 20 rely on the same arguments as for claim 1, and Appellant
does not argue separate patentability for the dependent claims, except as
noted below. We, therefore, also sustain the Examiner’s rejection of claims
11-14, and 16-20.
Claims 2 and 3
Appellant contends “Chen does not suggest new words comprising
different phonemes . . . and [tjhere is no suggestion that the phonetic
matching algorithm of Chen be a fuzzy matching algorithm” (App. Br. 20—
21) and Byrd “does not suggest identifying candidate solution descriptions”
(id. at 21). In response, the Examiner finds, and we agree, that
“[Ajppellant’s arguments [largely] rely on features which are not claimed”
(Ans. 20).
For example, claim 2 fails to recite “new words comprising different
phonemes” or “fuzzy matching.” Instead, claim 2 recites, inter alia,
“identify candidate solution descriptions . . . which are determined to match
at least some of the identified words in the preliminary transcription” (see
claim 2). For this feature, the Examiner relies upon Chen’s teaching of
“matching phoneme transcriptions of an unrecognized initial transcription
(e.g., ‘Marc Dick son’). . . where at least some words are in common”
(Ans. 21). Appellant’s contentions fail to persuasively rebut such teachings.
Accordingly, we sustain the Examiner’s rejection of claims 2 and 3.
7
Appeal 2017-002867
Application 13/974,515
Claim 4
Appellant contends “Kuspa also does not suggest that a gap between
words be limited to a threshold size. Rather, Kuspa uses the Levenshtein
Word Edit Distance algorithm, which is the minimum number of single­
character edits” (App. Br. 21—22). In response, the Examiner finds that
Kuspa teaches that “a generated transcript from speech must be within a
certain number of edits of a database transcription. . . . Chen also teaches a
matching based algorithm .... Thus, words would not be considered to
match if too many missing additions were required or a gap is too large”
(Ans. 21—22). The Examiner further finds, and we agree, that “the claims do
not specify how the so called ‘threshold size’ is actually measured (e.g., in
words, phonemes, etc.) [and] under the broadest reasonable interpretation”
claim 4 reads on the combined teachings of Kuspa and Chen {id. at 22).
Claim 4 merely recites “wherein the gap between the matching pair of
words in the solution description is permitted to be no more than a threshold
size” {see claim 4). We agree with the Examiner that the claimed “gap” and
“threshold size” are not limited in the claim to any particular metric, so the
metric could be a distance (Ans. 22). Appellant fails to demonstrate that
“gap” and “threshold size” have been described in Appellant’s Specification
in a way that is inconsistent with the Examiner’s interpretation. Thus, we
find unpersuasive Appellant’s contention that claim 4 does not read on the
combined Kuspa’s word edit distance and Chen’s matching algorithm.
Accordingly, we sustain the Examiner’s rejection of claim 4.
8
Appeal 2017-002867
Application 13/974,515
Claim 5
Appellant contends “there is no suggestion of using a matching pair of
words in a solution database to identify an unrecognized phoneme. Rather,
Chen’s algorithm looks for words with matching phonemes” (App. Br. 22).
In response, the Examiner directs our attention to “the specific
example provided by Chen in which a database entry ‘Marc Dickinson’ is
spaced by at least one unrecognized phoneme ‘in’ . . . (e.g., ‘Marc Dick
son’)” (Ans. 22; see also Chen | 66). We agree with the Examiner that
Chen would require some ‘threshold degree of phonetic
matching/similarity’” otherwise “the entire database contents would match
the phonemes of the initial transcription” (id.). Further, Chen teaches that
“[ijnstead of telling the user that the assistant cannot complete the operation
because it cannot find ‘Mark Dick son’ in the database, speech repair can . . .
use a fuzzy-matching algorithm to generate a more plausible alternative
transcription” (| 66). Thus, Chen clearly considers phonetic matching of
unrecognized words. The Examiner further finds Jeong is used to teach
phoneme-based speech recognition which is combinable with the teachings
of Chen for generating word candidates using phoneme recognition ((Ans.
19).
Accordingly, we sustain the Examiner’s rejection of claim 5.
Claim 6
Appellant contends “Chen does not suggest relying on an inherent
threshold to differentiate matches from non-matches between phonemes. . . .
Chen uses fuzzy matching for text, but not for phonemes” (App. Br. 23). In
response, the Examiner finds, and we agree, that “these arguments are []
focused on the teachings of Chen rather than the language actually featured
9
Appeal 2017-002867
Application 13/974,515
in [claim 6] . . . [and] Chen uses phonetic matching to determine a database
match for an unrecognized word” as discussed regarding claim 5 (Ans. 23).
Accordingly, we sustain the Examiner’s rejection of claim 6.
Claim 7
Appellant contends “Chen does not suggest associating text sequences
of solution descriptions in a knowledge base with respective sequences of
phonemes” (App. Br. 24). In response, the Examiner correctly points out
that Appellant is again arguing the references separately and fails to consider
the combined teachings, i.e., “piecemeal analysis” (see Ans. 23).
Accordingly, we sustain the Examiner’s rejection of claim 7.
Claims 8—10
Appellant contends “the Examiner fails to particularly point out where
in Chen, Jeong, and/or Byrd there is any suggestion of retrieving an
identifier . . . [and] [t]he cited paragraphs make no suggestion of retrieving
phoneme sequences for a cluster of words associated in memory with the
solution identifier” (App. Br. 25). In response, the Examiner finds, and we
agree, that Appellant’s “argument amounts to a general allegation of
patentability since it does not explain how/why the combined prior art of
record fails to teach the argued limitations . . . Byrd teaches a cluster . . .
[and] [t]hese stored text clusters may be identified with a number of
different identifiers . . . Chen provides the remaining processing directed to
phoneme matching” (Ans. 24).
Here, Appellant merely argues that neither reference teaches or
suggests the above-noted limitations without providing any meaningful
analysis that explains why the Examiner erred (App. Br. 25), A statement
10
Appeal 2017-002867
Application 13/974,515
which merely points out what a claim recites will not he considered an
argument for separate patentability of the claim. See 37 C.F.R.
§ 41.37(c)(l)(vii). We note that arguments which Appellant could have
made but chose not to make in the Briefs have not been considered and are
deemed to be waived. Therefore, Appellant’s arguments are unpersuasive.
Accordingly, we sustain the Examiner’s rejection of claims 8—10.
Claim 15
Appellant contends that in Byrd “[tjhere is no reference here to
phonemes being drawn from a finite alphabet” (App. Br. 27). In response,
the Examiner finds that “[i]t is a basic fact that human languages are known
to comprise a certain finite set of sounds with some languages possessing
certain sounds while other do not. ... It is noted that there is no natural
language having an infinite alphabet. . . [because there is an] inherently
finite set [of] sound sets” (Ans. 25). In other words, the Examiner finds, and
we agree, that each language has a finite alphabet pertaining thereto,
therefore, the combined teachings of Chen, Jeong, and Byrd would teach the
limitations of claim 15.
Accordingly, we sustain the Examiner’s rejection of claim 15.
DECISION
We affirm the Examiner’s § 103(a) rejections Rl—R5.
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