Murat Akbacak et al.

Patent Trials and Appeals Board

Dec 23, 2019

14308174 - (D) (P.T.A.B. Dec. 23, 2019)

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.
14/308,174 06/18/2014 Murat Akbacak 340625-US-NP/199893 6253
45809 7590 12/23/2019
SHOOK, HARDY & BACON L.L.P.
(MICROSOFT TECHNOLOGY LICENSING, LLC)
INTELLECTUAL PROPERTY DEPARTMENT
2555 GRAND BOULEVARD
KANSAS CITY, MO 64108-2613
EXAMINER
SHAH, BHARATKUMAR S
ART UNIT PAPER NUMBER
2677
NOTIFICATION DATE DELIVERY MODE
12/23/2019 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):
IPRCDKT@SHB.COM
docketing.shb@clarivate.com
usdocket@microsoft.com
PTOL-90A (Rev. 04/07)
UNITED STATES PATENT AND TRADEMARK OFFICE
BEFORE THE PATENT TRIAL AND APPEAL BOARD
Ex parte MURAT AKBACAK, DILEK Z. HAKKANI-TUR,
GOKHAN TUR, and LARRY P. HECK
Appeal 2018-007780
Application 14/308,174
Technology Center 2600

Before JOHNNY A. KUMAR, LINZY T. McCARTNEY, and MATTHEW
J. McNEILL, Administrative Patent Judges.

KUMAR, Administrative Patent Judge.

DECISION ON APPEAL

Pursuant to 35 U.S.C. § 134(a), Appellant1 appeals from the
Examiner’s decision to reject claims 1–7 and 9–21. Appeal Br. 3. Claim 8
has been cancelled. Id. We have jurisdiction under 35 U.S.C. § 6(b).
We affirm.

1 We use the word Appellant to refer to “applicant” as defined in 37 C.F.R.
§ 1.42(a). Appellant identifies the real party in interest as MICROSOFT
TECHNOLOGY LICENSING, LLC. Appeal Br. 3.
Appeal 2018-007780
Application 14/308,174

2
STATEMENT OF THE CASE
Introduction
Appellant’s invention relates to “systems and methods for improving
language models and spoken language understanding (SLU) models for
speech recognition by adapting knowledge sources utilized by the models to
session contexts.” Spec. ¶ 5.
Representative Claim
Representative claim 1 under appeal reads as follows:
1. A computer-implemented method comprising:
receiving usage history information comprising one or
more sequences of user-interaction events, each sequence being
a series of turns of a session and each event in the sequence
corresponding to a respective user tum of the series of turns;
for each event in the one or more sequences, determining
a likely user intent corresponding to the event;
for each event, determining a set of intent transition
probabilities, each intent transition probability representing a
probability of a transition occurring from a first user intent to a
second user intent and corresponding to a frequency of the
transition being detected in the one or more sequences using the
likely user intents determined for each event;
utilizing the set of intent transition probabilities with a
language model to provide a session context model for
computer-performed speech recognition or contextual
understanding;
determining a user intent of a user based on the set of
intent transition probabilities using the session context model;
and
providing user content for the computer-performed
speech recognition or contextual understanding based on the
determined user intent.
Appeal Br. 14 (Claims Appendix).

Appeal 2018-007780
Application 14/308,174

3
REFERENCES AND REJECTIONS2
I) Claims 1–7, 9–11, 13‒16, and 18‒21 are rejected under 35 U.S.C.
§ 103(a) as being unpatentable over Lloyd (US 8,352,246 B1; iss. Jan. 8,
2013) in view of Arellano et al. (US 2004/0128624 A1; publ. July 1, 2004)
and in further view of Nitz et al. (US 2014/0310002 A1; publ. Oct. 16,
2014).3
II) Claims 12 and 17 are rejected under 35 U.S.C. § 103(a) as being
unpatentable over Lloyd in view of Arellano and in further view of Nitz and
in further view of Prous Blancafort et al. (US 2006/0195319 A1; publ. Aug.
31, 2006).
Appellant’s Contentions
Appellant contends that the Examiner erred in rejecting claims 1, 7, and
14 under 35 U.S.C. § 103(a) because:
Claim 1 includes “a set of intent transition probabilities, each intent
transition probability representing a probability of a transition
occurring from a first user intent to a second user intent.” As
discussed above, Lloyd is entirely unconcerned with user intent or a

2 According to Appellant, “Claim 17 is not indicated as being rejected in
either the Final Office Action mailed November 20, 2017, or the preceding
Non-Final Office Action. However, the Examiner indicated via phone on
February 12, 2018, that he intended to reject claim 17 on the same grounds
that are stated in support of the rejection of claim 12. To the extent that
claim 17 is deemed to be rejected, Appellant appeals this rejection.” Appeal
Br. 3. Therefore, we treat claims 12 and 17 together for purposes of this
appeal. See Ans. 2.
3 We note that claims 13, 16, and 20 were not listed in the heading statement
of the rejection; however, the Examiner provided the rejections for claims
13, 16, and 20 in the description in the Final Action (Final Act. 10–14). We
find that the Examiner’s failure to list the rejection of claims 13, 16, and 20
was an inadvertent and harmless error.

Appeal 2018-007780
Application 14/308,174

4
transition from a first user intent to a second user intent. Instead, at
best, Lloyd teaches a probability that an “accessed term” is a correct
transcription. Lloyd, col. 8, 11. 19-27. The probability that an
“accessed term” is a correct transcription, as in Lloyd, bears no
relevance to an intent transition probability that represents a
probability of a transition occurring from a first user intent to a
second user intent, as is recited by claim 1.

Furthermore, claim 1 states that “each intent transition probability ...
correspond[s] to a frequency of the transition from the first user intent
to the second user intent.” The Office relies on Lloyd to teach this
feature. Final Office Action mailed 11/20/17, 6. But Lloyd merely
mentions “the frequency that [a] term ... occurs in a particular
context.” Lloyd, col. 8, 11. 11-16. Counting a number of term
occurrences, as in Lloyd, falls far short of teaching “a frequency of
the transition from the first user intent to the second user intent,” as
is recited by claim 1. The mere fact that Lloyd uses the word
“frequency” is insufficient to teach this element of claim 1.

App. Br. 8 (hereinafter “intent transition probability” feature).
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. The Examiner has provided a comprehensive response,
supported by sufficient evidence, to each of the contentions raised by
Appellant. We adopt as our own (1) the findings and reasons set forth by the
Examiner in the Final Action from which this appeal is taken (see Final
Act. 2–15) and (2) the reasons set forth by the Examiner in the Examiner’s
Answer in response to Appellant’s Appeal Brief (see Ans. 3–21). However,
we highlight and address specific findings and arguments for emphasis as
follows.
Appeal 2018-007780
Application 14/308,174

5
As to Appellant’s above contentions, we disagree for the reasons set
forth by the Examiner at pages 16-19 of the Answer. The Examiner finds
Llyod explicitly discloses the disputed “intent transition probability” feature:
For example, the adjusted language model indicates a probability of
an occurrence of a term in a sequence of terms based on other terms in
the sequence. Adjusting a language model includes accessing a stored
language model and adjusting the stored language model based on the
similarity score. Adjusting the accessed language model includes
adjusting the accessed language model to increase a probability in the
language model that the accessed term will be selected as a candidate
transcription for the audio data. Adjusting the accessed language
model to increase a probability in the language model includes
changing an initial weighting value assigned to the term based on the
similarity score. The method includes determining that the accessed
term was entered by a user, the audio data encodes speech of the user,
the first context includes the environment in which the speech
occurred, and the second context includes the environment in which
the accessed term was entered. The information that indicates the first
context and the information that indicates the second context each
indicate a geographic location. The information that indicates a first
context and information that indicates the second context each
indicate a document type or application type. The information
indicating the first context and the information indicating the second
context each include an identifier of a recipient of a message.

Llyod, col. 1, ll. 41-64). See also Llyod, col. 10, line 40 – col. 11, line
30).

Thus, the Examiner finds Lloyd teaches changing from a first to a
second context based on similarity scores and changing weighting values. In
other words, Appellant’s argument ignores the actual reasoning of the
Examiner’s rejections. Based on our review of Llyod (Columns 1, 10, and
11), we find that the Examiner, giving the claim its broadest reasonable
interpretation consistent with the Specification, has properly found that the
Appeal 2018-007780
Application 14/308,174

6
disputed limitations are met by Lloyd. We observe that Appellant’s Reply
Brief does not rebut the Examiner’s findings regarding the intent transition
probability limitation.
Appellant has also presented several arguments that the Lloyd
reference does not teach or suggest “determining a likely user intent” as
recited in the pending claims. Appeal Br. 7−10; Reply Br. 2−5.
Appellant’s arguments are misdirected and are not persuasive.
Appellant is attacking the Lloyd reference, when the Examiner relies upon
Nitz for teaching the claimed “determining a likely user intent” feature. See
Final Act. 8; Ans. 9.
In particular, the Examiner finds Nitz explicitly teaches “determining
a likely user intent”:
Nitz et al[.] teaches a method of determining a likely user
intent corresponding to the event (Nitz para[.] 0022
determines the intent of the user); based on the likely user
intents determined for each event (para[.] 0022 determines
the intention of a user using VPA 112 and 170 for example
the bank transaction sequence).
Final Act. 8; Ans. 9. See also Nitz, ¶ 21:
An “intent” as used herein generally refers to a structured
representation of the VPA 112, 170’s interpretation of the
user’s natural language dialog input. For example, an intent
may include semantically meaningful information about the
user’s natural language dialog input as well as contextual
information contained in or derived from other inputs (e.g.,
sensor inputs), which the VPA 112, 170 can understand and
process in order to invoke another VPA 112, 170 on the user’s
behalf, or to execute some other action to fulfill the user’s
intended goal or objective as expressed by the user through the
dialog input.
Appeal 2018-007780
Application 14/308,174

7
Thus, Appellant is challenging the wrong reference and not
responding to the Examiner’s specific findings. Appellant’s challenge to the
references individually is not convincing of error in the Examiner’s findings.
See In re Keller, 642 F.2d 413, 426 (CCPA 1981) (“[O]ne cannot show non-
obviousness by attacking references individually where, as here, the
rejections are based on combinations of references.)” (Citations omitted).
Therefore, based on this record, we are not persuaded of error
regarding the Examiner’s underlying factual findings and ultimate legal
conclusion of obviousness for independent claim 1, and independent claims
7 and 14 argued together with claim 1.
As to Appellant’s contentions regarding claims 5, 6, 11, 12, 17, and
20 (Appeal Br. 10–13), the Examiner has rebutted each of those arguments
supported by sufficient evidence. Ans. 10–16, and 19–21. Therefore, we
adopt the Examiner’s findings and underlying reasoning, which are
incorporated herein by reference. We see no error in these unrebutted
findings. Separate patentability is not argued for the remaining dependent
claims.
Because the Appellant has not persuaded us the Examiner erred, we
sustain the rejections of claims 1–7 and 9–21.

CONCLUSION
We affirm the Examiner’s decision to reject claims 1–7 and 9–21
under 35 U.S.C. § 103 (a).
Appeal 2018-007780
Application 14/308,174

8
DECISION SUMMARY
In summary:
Claims
Rejected
35 U.S.C.
§
Reference(s)/Basis Affirmed Reversed
1–7, 9–11,
13–16, 18–
21
103 (a) Lloyd, Arellano,
Nitz
1–7, 9–11,
13–16, 18–
21

12, 17 103 (a) Lloyd, Arellano,
Nitz, Prous
Blancafort
12, 17
Overall
Outcome
1–7, 9–21
TIME PERIOD FOR RESPONSE
No time period for taking any subsequent action in connection with
this appeal may be extended under 37 C.F.R. § 1.136(a). See 37 C.F.R.
§ 1.136(a)(1)(iv) (2017).
AFFIRMED