Ex Parte Udupa et al

Patent Trial and Appeal Board

Apr 8, 2013

10890496 (P.T.A.B. Apr. 8, 2013)

UNITED STATES PATENT AND TRADEMARK OFFICE
____________

BEFORE THE PATENT TRIAL AND APPEAL BOARD
____________

Ex parte RAGHAVENDRA U. UDUPA and TANVEER A. FARUQUIE
____________

Appeal 2010-007997
Application 10/890,496
Technology Center 2600
____________

Before ALLEN R. MacDONALD, DEBRA K. STEPHENS, and
LYNNE E. PETTIGREW, Administrative Patent Judges.

PETTIGREW, Administrative Patent Judge.

DECISION ON APPEAL
This is an appeal under 35 U.S.C. § 134(a) from a final rejection of
claims 1-3, 5, 8-12, 14, 17-21, 23, 26, and 27.1 We have jurisdiction under
35 U.S.C. § 6(b).
We affirm-in-part and enter a new ground of rejection pursuant to our
authority under 37 C.F.R. § 41.50(b).

1 Claims 4, 6, 7, 13, 15, 16, 22, 24, 25, and 28 have been cancelled.
Appeal 2010-007997
Application 10/890,496

2
STATEMENT OF THE CASE
Introduction
According to Appellants, their invention relates to machine translation
of text from a source language to a target language. Spec. 1, ll. 5-6.
Claims 1, 9, 10, 18, 19, and 27 are independent. Claims 1 and 9 are
illustrative of the invention (disputed limitations in italics):
1. A computer-implemented method for decoding
source text in a first language to target text in a second
language, said method comprising:
decoding, by a computer, said source text having a word
length, M, based on a specified alignment, from all possible
alignments, between words in said source text and said target
text having a word length, L,
wherein said decoding provides a probability of an
optimal translation in said target text from said source text;
subsequently, specifying, by said computer, another
specified alignment, from said all possible alignments, and
decoding said source text, based on said another specified
alignment;
determining, by said computer, said probability of an
optimal translation for each of said all possible alignments; and
outputting, by said computer, said target text, in said
second language, associated with a highest probability value of
probabilities of an optimal translation of said all possible
alignments.
Appeal 2010-007997
Application 10/890,496

3

9. A computer-implemented method for decoding
source text in a first language to target text in a second
language, said method comprising:
inputting, by a computer, said source text of word length,
M;
initializing, by said computer, a value of a most probable
translation to a target language as null, and initializing, by said
computer, a most probable alignment between said source text
and said target text as null;
initializing, by said computer, a first processing loop
variable, L, for each value of L from M/2 to 2M,
wherein L corresponds to a word length of said
target text;
initializing, by said computer, a second processing loop
variable, A, for each alignment between said source text and
said target text;
from said initial value and for each successive value of L
of said first processing loop, determining, by said computer, a
probability of said most probable translation of said target text
for each value of A of said second processing loop, based on
decoding said corresponding source text; and
outputting, by said computer, said target text, in said
second language, associated with a highest probability value of
probabilities of an optimal translation of said all possible word
lengths and alignments.
Appeal 2010-007997
Application 10/890,496

4
Rejections on Appeal2
The Examiner has rejected claims 9, 18, and 27 under 35 U.S.C.
§ 112, second paragraph, as being indefinite for failing to particularly point
out and distinctly claim the subject matter which Appellants regard as their
invention. Ans. 3-4.
The Examiner has rejected claims 1-3, 10-12, and 19-21 under 35
U.S.C. § 102(b) as being anticipated by Marcu (US 2002/0188439 A1, Dec.
12, 2002). Ans. 4-6.
The Examiner has rejected claims 5, 8, 14, 17, 23, and 26 under 35
U.S.C. § 103(a) as being unpatentable over Marcu and Tillmann (Tillmann
and Ney, Word Re-ordering and DP-based Search in Statistical Machine
Translation, International Conference on Computational Linguistics,
Proceedings of the 18th Conference on Computational Linguistics, vol. 2,
2000). Ans. 6-8.
Issues on Appeal
Based on Appellants’ arguments, the issues on appeal are:
(1) Did the Examiner err in rejecting claims 9, 18, and 27 under 35
U.S.C. § 112, second paragraph, as being indefinite for failure to particularly
point out and distinctly claim the subject matter which Appellants regard as
the invention (App. Br. 15-18)?

2 The Examiner has objected to claims 14, 17, and 26 as depending
from cancelled claims 13, 16, and 25, respectively. Ans. 3. For purposes of
this appeal, the Examiner treats claim 14 as depending from claim 10, claim
17 as depending from claim 14, and claim 26 as depending from claim 23.
Id. We adopt the Examiner’s treatment of these claims for purposes of this
appeal. We also note Appellants’ statement that the Examiner did not enter
proposed claim amendments (Amendment, Sept. 21, 2009) that would have
corrected the dependencies. App. Br. 22.
Appeal 2010-007997
Application 10/890,496

5
(2) Did the Examiner err in rejecting claims 1, 10, and 19 under 35
U.S.C. § 102(b) because Marcu fails to disclose “determining, by said
computer, said probability of an optimal translation for each of said all
possible alignments,” as recited in claim 1 and similarly recited in claims 10
and 19 (App. Br. 18-21)?
(3) Did the Examiner err in rejecting claims 5, 8, 14, 17, 23, and 26
under 35 U.S.C. § 103(a) because the combination of Marcu and Tillmann
fails to teach or suggest the recited limitations (App. Br. 21-22)?
ANALYSIS
Rejection under 35 U.S.C. § 112, Second Paragraph:
Claims 9, 18, and 27
Claim 9 recites “initializing, by said computer, a first processing loop
variable, L, for each value of L from M/2 to 2M,” and “from said initial
value and for each successive value of L of said first processing loop,
determining . . . a probability . . . for each value of A of said second
processing loop.” The Examiner concludes the claim is indefinite because
“it is unclear how and why a variable [L] is initialized for each value of itself
from a range,” and “it is unclear to which initial value the [second]
limitation refers since there seems to be a plurality of initialized values
recited prior to this limitation.” Ans. 4.
Appellants contend that claim 9 clearly recites two nested processing
loops—a first processing loop for L (word length of the target text) and a
second processing loop for A (alignment between target text and source
text). App. Br. 16-18. According to Appellants, the invention “initializes
the first processing loop, L, to M/2 . . . and then determines the probability
of a most probable translation of the target text for each possible alignment,
Appeal 2010-007997
Application 10/890,496

6
A.” App. Br. 16. After processing this first loop with L set to M/2,
Appellants explain that “the first processing loop L then iterates to another
loop corresponding to the next incremental target text word length, M/2 +
1,” and “[t]he first processing loop, L, is incremented until it reaches the
value 2M.” App. Br. 17; see also Reply Br. 4.
We agree with the Examiner that the claim language is not clear
enough to particularly point out and distinctly claim the subject matter
which Appellants regard as the invention as explained in their arguments.
Ans. 8. Appellants argue the claim recites initializing L to M/2 and then
incrementing L by 1 until it reaches 2M, as described by Appellants;
however, the Examiner interprets claim 9 as reciting initializing L to M/2
and then re-initializing L “for each value of L from M/2 to 2M.”
Because claim 9 has at least two plausible claim constructions, it is
indefinite for that reason alone. See Ex parte Miyazaki, 89 USPQ2d
1207, 1211 (BPAI 2008) (precedential). Appellant must therefore more
precisely define the metes and bounds of the claimed invention to clarify
this ambiguity. Additionally, because the Examiner’s interpretation
results in multiple initial values of L, it is unclear which of the initial
values provides antecedent basis for the later recitation of “said initial
value.”
Claims 18 and 27 include similar limitations and therefore suffer from
the same lack of clarity. For these reasons, we sustain the Examiner’s
rejection of claims 9, 18, and 27 under 35 U.S.C. § 112, second paragraph,
for failure to particularly point out and distinctly claim the subject matter
which Appellants regard as the invention.
Appeal 2010-007997
Application 10/890,496

7
Rejection under 35 U.S.C. § 102(b):
Claims 1, 10, and 19
Appellants contend that Marcu does not disclose “determining, by
said computer, said probability of an optimal translation for each of said all
possible alignments,” as recited in independent claim 1 and similarly recited
in claims 10 and 19. App. Br. 18-21; Reply Br. 5-7. Rather than determine
a probability of a most probable translation for each of all possible
alignments, Appellants argue that Marcu’s “decoder tries to find an
alignment (and implicitly, a translation) of higher probability by applying
one or more sentence modification operators.” Marcu, ¶ [0027]; App. Br.
20. Appellants further contend that Marcu’s sentence modification operators
do not “necessarily find all possible alignments as does the present
invention.” Reply Br. 6 (emphasis in original).
We agree with Appellants. Marcu discloses an iterative process in
which the decoder tries to improve the current translation by applying
sentence modification operations “until the sentence modification operations
cease (as determined in block 435) to produce translation solutions having
higher probabilities of correctness, at which point, the decoding process
halts and the current translation is output as the final decoding solution
(block 445).” Marcu, ¶ [0035]; see also Marcu, Fig. 4. Alternatively,
Marcu teaches that “the decoder 115 could cease after a predetermined
number of iterations chosen, for example, either by a human end-user or by
an application program.” Marcu, ¶ [0035]. Thus, whether Marcu’s process
terminates when it produces an alignment that does not have a higher
probability of being correct than the current translation or terminates after a
predetermined number of iterations, Marcu’s process does not necessarily
Appeal 2010-007997
Application 10/890,496

8
determine the probability of an optimal translation for all possible
alignments.
The Examiner points to Marcu’s disclosure that, in each iteration, the
decoder “may use a process loop (blocks 425-440) to iterate exhaustively
over all alignments that are one [sentence modification] operation away
from the alignment under consideration.” Ans. 9 (quoting Marcu, ¶ [0036]);
see also Marcu, Fig. 4. The Examiner also notes that one of Marcu’s
sentence modification operations is swapSegments, which “creates a new
alignment from the old one by swapping non-overlapping English word
segments” and “iterates over all possible non-overlapping span pairs that can
be built on a sequence of length |e|.” Marcu, ¶ ¶ [0032], [0037]; see also
Ans. 9.
We disagree with the Examiner’s finding that these passages disclose
determining the probability of an optimal translation for each of all possible
alignments, as recited in claim 1. Rather, we find these passages describe
the new alignments obtained during one iteration of Marcu’s process by
applying sentence modification operators to the current alignment. As
discussed, this iterative process does not necessarily determine the
probability of an optimal translation for all possible alignments, as recited in
claim 1.
For these reasons, we do not sustain the Examiner’s rejection of
independent claims 1, 10, and 19 under 35 U.S.C. § 102(b) as anticipated by
Marcu, nor do we sustain the § 102(b) rejection of dependent claims 2, 3, 11,
12, 20, and 21. See App. Br. 20.
Appeal 2010-007997
Application 10/890,496

9
Rejection under 35 U.S.C. § 103(a):
Claims 5, 8, 14, 17, 23, and 26
As set forth above, we do not sustain the Examiner’s anticipation
rejection based on Marcu of independent claims 1, 10, and 19, from which
claims 5 and 8, claims 14 and 17, and claims 23 and 26 depend, respectively.
Because the Examiner has not shown that Tillmann cures the deficiencies of
Marcu, as applied by the Examiner, we do not sustain the Examiner’s
rejection of dependent claims 5, 8, 14, 17, 23, and 26 under 35 U.S.C.
§ 103(a) as being unpatentable over Marcu and Tillmann.
New Ground of Rejection under 35 U.S.C. § 103(a):
Claims 1, 10, and 19
Although we find that Marcu does not anticipate the invention as
recited in claims 1, 10, and 19, we conclude that, in view of Marcu, it would
have been obvious to a person having ordinary skill in the art to determine
the probability of an optimal translation for each of all possible alignments.
Marcu teaches an iterative process that tries to improve the current
translation by applying sentence modification operations to find an
alignment having a higher probability of correctness. Marcu, ¶¶ [0027],
[0035]. While Marcu describes terminating the process once the probability
of correctness can no longer be improved, the reference also suggests
continuing the process until the occurrence of some other condition. Marcu,
¶ [0035], claim 8 (“repeating said applying, said determining and said
setting until occurrence of a termination condition”). In view of that
suggestion, we conclude that it would have been obvious to a person of
ordinary skill in the art to continue Marcu’s iterative process until the
probabilities of correctness of all possible alignments between source text
Appeal 2010-007997
Application 10/890,496

10
and target text have been determined, rather than terminate the process at
some earlier point. Moreover, determining the probability of an optimal
translation for each of said all possible alignments, as recited in claim 1 and
similarly recited in claims 10 and 19, would have been within the skills of an
ordinary artisan and would have yielded predictable results. See KSR Int’l
Co. v. Teleflex Inc., 500 U.S. 398, 417 (2007).
We designate our analysis in combination with the Examiner’s
analysis as to the remaining limitations in claims 1, 10, and 19 to be a new
ground of rejection of claims 1, 10, and 19 under 35 U.S.C. § 103(a) over
Marcu.
Although we have rejected independent claims 1, 10, and 19 under 35
U.S.C. § 103(a) pursuant to our discretionary authority under 37 C.F.R.
§ 41.50(b), we have not reviewed the remaining claims to the extent
necessary to determine whether these claims are unpatentable under 35
U.S.C. § 103(a). We leave it to the Examiner to evaluate the patentability of
the remaining claims and determine the appropriateness of any further
rejections. See MPEP 1213.02.
DECISION
The Examiner’s rejection of claims 9, 18, and 27 under 35 U.S.C.
§ 112, second paragraph, is affirmed.
The Examiner’s rejection of claims 1-3, 10-12, and 19-21 under 35
U.S.C. § 102(b) as being anticipated by Marcu is reversed.
The Examiner’s rejection of claims 5, 8, 14, 17, 23, and 26 under 35
U.S.C. § 103(a) as being unpatentable over Marcu and Tillmann is reversed.
Appeal 2010-007997
Application 10/890,496

11
Pursuant to 37 C.F.R. § 41.50(b), we enter a new ground of rejection
of claims 1, 10, and 19 under 35 U.S.C. § 103(a) as being unpatentable over
Marcu.
Rule 37 C.F.R. § 41.50(b) states that “[a] new ground of rejection
pursuant to this paragraph shall not be considered final for judicial review.”
37 C.F.R. § 41.50(b) also provides that the Appellant, WITHIN TWO
MONTHS FROM THE DATE OF THE DECISION, must exercise one of
the following two options with respect to the new ground of rejection to
avoid termination of the appeal as to the rejected claims:
(1) Reopen prosecution. Submit an appropriate
amendment of the claims so rejected or new evidence relating
to the claims so rejected, or both, and have the matter
reconsidered by the examiner, in which event the proceeding
will be remanded to the examiner . . . .
(2) Request rehearing. Request that the proceeding be
reheard under § 41.52 by the Board upon the same record . . . .
No time period for taking any subsequent action in connection with
this appeal may be extended under 37 C.F.R. § 1.136(a)(1)(iv).
AFFIRMED-IN-PART
37 C.F.R. § 41.50(b)

msc