Microsoft Technology Licensing, LLC

Patent Trials and Appeals Board

Aug 17, 2021

2020003174 (P.T.A.B. Aug. 17, 2021)

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.
15/413,180 01/23/2017 Victoria Newcomb Podmajersky 14917.3070US01/361704.01 9310
27488 7590 08/17/2021
MERCHANT & GOULD (MICROSOFT)
P.O. BOX 2903
MINNEAPOLIS, MN 55402-0903
EXAMINER
PAULA, CESAR B
ART UNIT PAPER NUMBER
2177
NOTIFICATION DATE DELIVERY MODE
08/17/2021 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):
USPTO27488@merchantgould.com
usdocket@microsoft.com
PTOL-90A (Rev. 04/07)

UNITED STATES PATENT AND TRADEMARK OFFICE
____________

BEFORE THE PATENT TRIAL AND APPEAL BOARD
____________

Ex parte VICTORIA NEWCOMB PODMAJERSKY,
BUGRA OKTAY, and TRACY CHILDERS

Appeal 2020-003174
Application 15/413,180
Technology Center 2100
____________

Before MAHSHID D. SAADAT, RICHARD M. LEBOVITZ, and
JASON J. CHUNG, Administrative Patent Judges.

LEBOVITZ, Administrative Patent Judge.

DECISION ON APPEAL
The Examiner rejected the claims under 35 U.S.C. § 103 as obvious.
Pursuant to 35 U.S.C. § 134(a), Appellant1 appeals from the Examiner’s
decision to reject the claims. We have jurisdiction under 35 U.S.C. § 6(b).
We AFFIRM but designate it a new ground of rejection pursuant to
37 C.F.R. § 41.50(b).

1 We use the word “Appellant” to refer to “applicant” as defined in
37 C.F.R. § 1.42. Appellant identifies the real party in interest as Microsoft
Technology Licensing, LLC. Appeal Br. 3.
Appeal 2020-003174
Application 15/413,180

2
STATEMENT OF THE CASE
The Examiner rejected claims 1–5 and 8–22 in the Final Action under
35 U.S.C. § 103 as obvious in view of Kim et al. (US 2016/0308794 A1,
published Oct. 20, 2016) (“Kim”) and Mulvey (US 2004/0201607 A1,
published Oct. 14, 2004). Final Act. 3.
Independent claim 1 is representative and copied below (bracketed
numbering added for reference to the limitations in the claim):
1. A computer-implemented method for a virtual input system,
the method comprising:
[1] obtaining user data from one or more data sources,
the user data indicative of a personal communication style of a
user;
[2] generating a user communication model based, in
part, on the user data, wherein generating the user
communication model comprises:
[2a] generating a diction model for the user; and
[2b] generating a syntax model for the user;
[3] obtaining data regarding a current communication
context, the data comprising data regarding a communication
medium;
[4] generating a plurality of sentences for use in the
current communication context based, in part, on [4a] the user
communication model and [4b] the data regarding the current
communication context, wherein generating the plurality of
sentences comprises:
[4c] for each sentence of the plurality of sentences,
selecting a word for inclusion in the respective sentence using:
[4ci] the diction model of the user and [4cii] the syntax model
of the user; and
[5] causing the plurality of sentences to be
provided to the user for use over the communication medium.

CLAIM 1
Claim 1 is directed to a “computer-implemented method for a virtual
input system” that generates sentences to be provided to the user for use over
Appeal 2020-003174
Application 15/413,180

3
a communication medium (step [5] of claim 1). The sentences are generated
based in part on “[4a] the user communication model and [4b] the data
regarding the current communication context.” The “user communication
model” comprises a diction model ([2a], [4ci]) and a syntax model ([2b],
[4cii]) which are used [4c] to select the words in the sentences.
As explained in the Specification, “[t]raditional virtual input systems
are often limited to, for example, letter-by-letter input of text or include
simple next-word-prediction capabilities.” Spec. ¶ 24. The claimed method,
however, according to the Specification, improves input systems and
methods and “can provide the user with, for example, context-aware
communication options presented at a sentence or phrase level that are
customized to the user’s personal communication style.” Id.
The Specification explains that a diction model:
can define a particular vocabulary of words that can be used,
including the use of slang, jargon, profanity, and other words.
The diction model data 312 can also describe the use of words
common to particular dialects. For example, the dialect data can
describe regional dialects (e.g., British English) or activity-group
dialects (e.g., the jargon used by players of a particular video
game).
Spec. ¶ 54.
With respect to a “syntax model,” the Specification explains:
The syntax model data 310 is data for a syntax model, describing
how the syntax of a communication can be formulated, such as
how words and sentences are arranged. For example, where the
communication model 122 is a model of the communication for
a user of the input system, then the syntax model data 310 is data
regarding the user's use of syntax.
Spec. ¶ 53.
Appeal 2020-003174
Application 15/413,180

4
The Specification also discloses that a syntax model:
can include data regarding the use of split infinitives, passive
voice, active voice, use of the subjunctive, ending sentences with
propositions, use of double negatives, dangling modifiers,
double modals, double copula, conjunctions at the beginning of
a sentence, appositive phrases, and parentheticals, among others.
Spec. ¶ 53.
DISCUSSION
The Examiner found that Kim teaches substantially all the steps of
claim 1, but not selecting words based on diction and syntax models. Final
Act. 8–9. To meet this deficiency in Kim, the Examiner further cited
Mulvey, which the Examiner found describes generating and using diction
and syntax models in generating sentences as required by steps [2] and [4] of
claim 1. Final Act. 9–10. The Examiner found it would have been obvious to
one of ordinary skill in the art to incorporate Mulvey’s teachings into Kim’s
system “because both systems predict a user’s intended message or
alphanumeric information based on a user context.” Id. at 11. The Examiner
explained that the “combination would enable Kim’s system to generate a
personal context model that correlates various examples of user context
against a unique personal language model for the user as Mulvey suggested
in the Abstract.” Id.
Appellant contends that “Mulvey actually teaches away from
generating entire sentences for display to the user at least by describing that
‘[t]o minimize cognitive load on the user, pursuant to a preferred
embodiment, only one-word predictions are presented at a time’. [0030].”
Appeal Br. 19. Appellant states that this teaching weighs against the
combination of Mulvey with Kim to reach the claimed recitations.” Id.
Appeal 2020-003174
Application 15/413,180

5
Appellant also argues that Mulvey does not describe a syntax model
as asserted by the Examiner. Appeal Br. 19. Appellant states the paragraph
22–24 in Mulvey cited by the Examiner “describe merely analyzing word
frequency” and describes “communication context,” rather than syntax. Id. at
20.
Because Appellant only challenges the Examiner’s finding with
respect to the “syntax model” of claim 1, we focus our discussion on this
element of the claim. We begin with Kim.
Kim describes methods for analyzing a received message from a
communication medium and providing a recommended reply message based
on the content of the analyzed message. Kim ¶ 2. The Examiner found that
Kim analyzes the personal communication style of a user (citing Kim ¶¶ 66,
67) and recommends a reply message to a received message, e.g., a formal
reply message when the received message is formal (citing Kim ¶¶ 100,
102). Final Act. 3–5. In rejecting dependent claims 2 and 3, the Examiner
also cited Kim as selecting different styles in which to reply to a received
message (Kim ¶¶ 114, 115, 277–290). Final Act. 12–14.
As indicated by the Examiner, Kim describes generating reply
sentences in specific styles and also changing the style of a response by a
user. For example, Kim discloses:
Examples of the style of the recommended reply messages may
include a basic style, a friendly style, a cool style, and a user’s
style, but are not limited thereto. The user’s style may mean a
style the user usually uses, and the device 100 may determine a
style of the user based on messages directly input by the user.
Spec. ¶ 284.
Kim also discloses that styles may be selected by the user (brackets in
original to show numbered paragraphs):
Appeal 2020-003174
Application 15/413,180

6
[0288] The device 100 may also display another style on the
button 2310. For example, the device 100 may display “my style”
on the button 2310 to indicate that the recommended reply
messages may be changed to the user’s style.
[0289] Referring to FIG. 23C, upon receiving a user input of
selecting the button 2310 indicating “my style”, the device 100
may change the style of the recommended reply messages to the
user’s style.
[0290] For example, the user may tend to add an emoticon at the
end of a sentence. Accordingly, the device 100 may generate
recommended reply messages in the user’s style by adding
emoticons at the end. For example, the device 100 may change a
recommended reply message “sorry” to “sorry :-( ”.
The term “style” is not expressly defined in Kim, but examples of
styles are given of adding an honorific form to sentences to indicate a formal
style (Kim ¶ 281), choosing specific words (Kim ¶ 115, 287) (changing
“yes” to “yup”), and adding emoticons at the end of sentences when that is a
characteristic style of the user (Kim ¶ 290). Kim discloses that different
styles may be used for a reply message, such as the user’s style, a friendly
style, a cool style, etc. (Kim ¶ 284). In essence, a style is the “voice” of the
writer, such as a friendly voice, a cool voice, or an informal voice.
Appellant’s Specification, as discussed above (supra. 4), expressly
discloses “voice” as an example of data used to formulate the syntax model
(Spec. ¶ 53: “passive voice, active voice”). It is implicit in using a specific
voice to reply to a message that not only word choice would be considered,
but also the arrangement of the specific words in the sentence, such as
structuring a sentence to add an emoticon at the end of it (Kim ¶ 290). Table
5 of Kim (at ¶ 96) gives examples of formal and informal replies to the same
message which illustrate this point:
Appeal 2020-003174
Application 15/413,180

7
For a reply asking “who,” the informal reply can be “Who?” and the
formal reply can be “Who are you?” Kim ¶ 96, Table 5. In this example, the
formal reply ends with “you,” which is a pronoun, while the informal style
uses a different sentence structure that lacks the pronoun at the end,
indicating informality in the user’s voice. For a reply asking a location, the
informal reply can be “In XX,” while the formal reply can be “It is in XX.”
Id. In this example, the informal reply lacks the noun “it” at the sentence’s
beginning and has a different sentence syntactical structure than the formal
reply which uses a noun and verb at its beginning.
It is therefore reasonably understood by one of ordinary skill in the art
that the use of style in Kim evokes a syntax model to formulate a reply.
When a style is selected by a user, the reply message will employ different
sentence structures depending upon the user’s choice of style/voice.2
Therefore Kim’s disclosure about selecting a style and responding in the
selected style, reasonably suggests to one of ordinary skill in the art, and
makes obvious, utilizing a syntax model to generate the reply sentences by
choosing sentence structures (see Kim, Table 5) that reflect the specific
writing style. The Examiner, in fact, found that a syntax model would have
been obvious based on Mulvey’s teaching of using certain word usage to
generate sentences. Final Act. 48–49.

2 Kim, as cited by the Examiner in an earlier Office Action (Non-Final
Office Action), discloses “determin[ing] by the morphologic/syntactic
analyzer 214, a named entity, a sending intention, formality, and emotion of
a sender, and determine a reply type.” Kim ¶ 80. Thus, Kim determines the
formality of a reply using a syntax analyzer, providing additional evidence
that whether a reply is generated in a formal or informal voice reflects the
syntax of it.
Appeal 2020-003174
Application 15/413,180

8

Equating “style” with “syntax” is also consistent with Appellant’s
disclosure which teaches that “communication model input data 110 can
include information regarding or indicative of a specific style or pattern of
communication, including information regarding grammar, syntax,
vocabulary, and other info.” Spec. 34 (emphasis added). In other words,
“style” includes “syntax.”
Thus, we find that Kim describe using a syntax model when
generating a reply message to a received message.
While we agree with Appellant that Mulvey is largely concerned with
word usage (Appeal Br. 19–20), and therefore diction, the rejection is not
defective because Kim describes utilizing a syntax model, and for the
reasons given by the Examiner, it would have been obvious to apply
Mulvey’s teachings to Kim’s system.
Appellant also argues, as indicated above, that “Mulvey actually
teaches away from generating entire sentences for display to the user at least
by describing that ‘[t]o minimize cognitive load on the user, pursuant to a
preferred embodiment, only one-word predictions are presented at a time’.
[0030].” Appeal Br. 19. Appellant states that this teaching weighs against
the combination of Mulvey with Kim to reach the claimed recitations.” Id.
This argument does not persuade us that the Examiner erred in finding
the claims obvious over the combination of Kim and Mulvey. Mulvey
teaches that “[p]referably one word only predictions are presented as part of
an overall strategy to minimize cognitive load on the user.” Mulvey ¶ 16.
Mulvey further discloses “[t]o minimize cognitive load on the user, pursuant
to a preferred embodiment, only one-word predictions are presented at a
Appeal 2020-003174
Application 15/413,180

9
time.” Id. ¶ 30. However, while one word at a time is presented to a user,
Mulvey teaches“[w]ith each acceptance of a correct prediction, the user will
be offered the next prediction. In this manner a user can conceivably
complete an entire sentence or message without entering additional specific
text.” Id. Thus, Mulvey generates sentences, but one word at a time. The
claim does not exclude generating sentences one word at a time as long as
the sentence generation selects words “for inclusion in the respective
sentence using: [4ci] the diction model of the user and [4cii] the syntax
model of the user” as recited in step [4] of the claim.
For the foregoing reasons, the obviousness rejection of claim 1 is
affirmed. Claims 2–4 and 8–21 were not argued separately and therefore fall
with claim 1. 37 C.F.R. 41.37(c)(1)(iv).

Claim 5
Claim 5 depends from claim 1, and further comprises:
receiving a selection of a communication model other
than a current communication model; and
wherein generating the plurality of sentences for use in
the current communication context is further based, in part, on
the communication model other than the current
communication model.

The Examiner found that Mulvey describes the limitations recited in
claim 5 and that it would have been obvious to one of ordinary skill in the
art to apply Mulvey to Kim. Final Act. 15–16. The Examiner cited paragraph
25 of Mulvey as teaching the limitation recited in claim 5. Paragraph 25 of
Mulvey is copied below.
Referring again to FIG. 2, when making 22 the statistical
analysis, the process can additionally consider other material and
Appeal 2020-003174
Application 15/413,180

10
or analysis 23 as may be available. For example, prior raw
material for this individual could be processed anew in
conjunction with the newly received 21 information. . . . In the
above-suggested examples, the additional material relates to the
user himself (or herself). If desired, however, the additional
material could constitute other information that, although not
originating with the user, may still relate to the user in some
appropriate way. For example, if standard language models are
available for various professions, hobbies, cultures, and so forth,
such models could be used to supplement and enrich the
statistical analysis (in this way, for example, a few occurrences
of the word “baby” in the user’s materials could be statistically
weighted more heavily if it were known that the user was a
pediatrician and the word “baby” had an overall high frequency
of use in a pediatrician-specific language model).
Mulvey ¶ 25.
The Examiner determined it would have been obvious to incorporate
Mulvey’s teaching into Kim’s system to “enable Kim’s system to generate a
personal context model that correlates various examples of user context
against a unique personal language model for the user as Mulvey suggested
in the Abstract.” Final Act. 16.
Appellant contends that Mulvey “rather than describing the selection
and use of a communication model other than a current communication
model to, for example, generate sentences, the paragraph describes factors
relevant to generating the user’s model.” Appeal Br. 22–23.
This argument is not persuasive. Mulvey specifically states that
“standard language models . . . for various professions, hobbies, cultures,
and so forth, such models could be used to supplement and enrich the
statistical analysis” (see Mulvey 25 copied above). This disclosure is an
express teaching of supplementing the model in Mulvey (e.g., a diction
model based on word usage frequency) with additional models, for example,
Appeal 2020-003174
Application 15/413,180

11
a model based on a user’s profession. Appellant did not persuasively
demonstrate an error in the Examiner’s fact-finding or reasoning.
Accordingly, the rejection of claim 5 is affirmed for the reason given by the
Examiner. Appellant make the same argument for claim 22. Appeal Br. 23.
The rejection of claim 25 is therefore affirmed for the same reasons given by
the Examiner.

Claim 22
Claim 22 depend from claim 5 and further recites:
wherein the communication model other than the current
communication model is a communication model of a regional
dialect different from that of the user; or
wherein the communication model other than the current
communication model is a communication model of a celebrity,
author, or fictional character.

Appellant argues that “recitations are further patentable over the
proposed combination of references because Mulvey does not describe, for
example, communication models of regional dialects or a celebrity, author,
or fictional character as claimed in claim 22.” Appeal Br. 23.
This argument is not persuasive.
Appellant appears to be attempting to distinguish the claim, not on the
function of the communication model and how it is implemented, but rather
on its content, for example, of using word choice to match a regional dialect
or the voice of a fictional character. We consider this to be “nonfunctional
descriptive matter” which is an inadequate basis to establish patentability.
Patentable weight is not given to the content of information where there no
functional relationship between the information and the computer, device, or
any other element recited in the claim. In re Gulack, 703 F.2d 1381, 1385
Appeal 2020-003174
Application 15/413,180

12
(Fed. Cir. 1983); In re Ngai, 367 F.3d 1336 (Fed. Cir. 2004); Ex parte Nehls,
88 USPQ2d 1883, 1889 (BPAI 2008) (precedential) (“[T]he nature of the
information being manipulated does not lend patentability to an otherwise
unpatentable computer-implemented product or process.”). The specific
elements of claim 22 are not functional, but rather represent only content
that is intended for human perception, for example, picking words and
sentences to a match dialect of a region of the United States. In fact, Kim
suggests this concept in describing messages that may be tailored to a
boyfriend of the user, using the terms “bzzy, “sup?”, and “dunno” (Kim
¶ 115), where the choice of a word operates in the same way that words
would be chosen to match a specific regional dialect. Consequently,
Appellant’s argument is inadequate to distinguish claim 22 from Kim and
Mulvey.

New ground of rejection
We designate the affirmance of the rejection of clam 1 as a new
ground of rejection, because while we relied on the same combination of
references cited by the Examiner, our reasoning differed from the
Examiner’s. We adopt the Examiner’s uncontested findings of fact from the
Final Rejection and Examiner’s Answer for the remaining claims.

DECISION SUMMARY
In summary:
Claims
Rejected
35
U.S.C.
§
Reference(s)/Basis Affirmed Reversed New
Ground
1–5, 8–
22
103 Kim, Mulvey 1–5, 8–22 1–5, 8–
22
Appeal 2020-003174
Application 15/413,180

13
TIME PERIOD FOR RESPONSE
This decision contains a new ground of rejection pursuant to
37 C.F.R. § 41.50(b). 37 C.F.R. § 41.50(b) provides “[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. . . .

Further guidance on responding to a new ground of rejection can be
found in the Manual of Patent Examining Procedure § 1214.01.
AFFIRMED; 37 C.F.R. 41.50(b)