Ex Parte Nandakumar et al

Patent Trials and Appeals Board

Jun 7, 2019

13755767 - (D) (P.T.A.B. Jun. 7, 2019)

UNITED STA TES p A TENT AND TRADEMARK OFFICE
APPLICATION NO. FILING DATE
13/755,767 01/31/2013
89885 7590 06/07/2019
FERENCE & ASSOCIATES LLC
409 BROAD STREET
PITTSBURGH, PA 15143
FIRST NAMED INVENTOR
Vikrant Nandakumar
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
ATTORNEY DOCKET NO. CONFIRMATION NO.
IN920120197US 1(790.182) 1512
EXAMINER
JAMI,HARES
ART UNIT PAPER NUMBER
2162
MAIL DATE DELIVERY MODE
06/07/2019 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 VIKRANT NANDAKUMAR, NA VEEN K. PRATHAP ANENI,
NITHYA RAJAMANI, and L. VENKATA SUBRAMANIAM
Appeal2018-007248
Application 13/755,767 1
Technology Center 2100
Before JOSEPH L. DIXON, JAMES W. DEJMEK, and
STEPHEN E. BELISLE, Administrative Patent Judges.
DEJMEK, Administrative Patent Judge.
DECISION ON APPEAL
Appellants appeal under 35 U.S.C. § 134(a) from a Final Rejection of
claims 1, 3, 8, 10, 11, 13, 18, 20-24, and 26. Appellants have canceled
claims 2, 4--7, 9, 12, 14--17, 19, and 25. App. Br. 25. We have jurisdiction
over the remaining pending claims under 35 U.S.C. § 6(b ).
We affirm.
1 Appellants identify International Business Machines Corporation as the
real party in interest. App. Br. 3.
Appeal2018-007248
Application 13/755,767
STATEMENT OF THE CASE
Introduction
Appellants' disclosed and claimed invention generally relates to
"gathering and managing crowd-sourced information." Spec. ,r 2. As an
example, crowd-sourced information may relate to an event such as a car
accident. Spec. ,r 26. If a car accident occurs, crowd-sourced information
may provide information such as the location of the accident, the number of
cars involved, effects of the accident ( e.g., traffic disruption), from a
plurality of different individuals (i.e., members of the crowd). Spec. ,r 26.
Crowd-sourced information may take the form of text, speech, audio, image,
or video. Spec. ,r 29.
The disclosed approach for gathering and managing crowd-sourced
information comprises the four steps of (i) event detection; (ii) event
classification; (iii) parts identification; and (iv) event completion. Spec.
,r 26. More particularly, in the event detection step, crowd-sourced
information is received and, for example, keywords are extracted from
received text messages. Spec. ,r 29. The extracted keywords are aggregated
to detect that an event has occurred (i.e., determining a number of messages
from a particular location mentioning keywords such as "car" and
"accident"). Spec. ,r,r 26, 29. During event classification, a rule- or
learning-based classifier classifies the messages into one of many known
classes and a semantic model is determined. Spec. ,r,r 26, 31. Additional
information to complete the semantic model is identified during the parts
identification step. Spec. ,r,r 26, 32. In the event completion step,
individuals may be identified and questions directed to those identified
individuals to elicit more complete information. Spec. ,r,r 26, 33.
2
Appeal2018-007248
Application 13/755,767
Figure 7 is illustrative of a semantic ( event) model and associated
parts identification and is reproduced below:
..................................................
'
. '
'. t;::'t~ ,~ ... ~:::~hi(~t1$ ~
:.:~w,":;:~v!<:{~
' ......................
_.'!t
.. ····
137
. -• ............. ·-·. ~· ·- ..................... .
'
' ~HJ~~~l'X..~:C' ~){ ~~":},.::·('::$ :
FIG .. 7
,.- ...... - .................. .
'
'
Figure 7 illustrates "an event model [(also referred to as a semantic model)]
and associated parts identification" for an event classified as a "Road
Accident." Spec. ,r,r 13, 32. In the embodiment depicted in Figure 7, the
associated parts that make up the model related to the event "Road
Accident" are shown in dotted lines ( e.g., type of injury, level of traffic
disruption, and location). According to the Specification, "event models can
be predefined or they can be developed over time by learning important
facets of information for a given event." Spec. ,r 32. Additionally, the
Specification describes that facet discovery relates to the process of
identifying different facets of an event. Spec. ,r 32.
Claim 1 is representative of the subject matter on appeal and is
reproduced below with the disputed limitations emphasized in italics:
3
Appeal2018-007248
Application 13/755,767
1. A method of gathering and managing crowd-sourced
information, said method comprising:
rece1vmg, at an electronic device, crowd-sourced
information, wherein the crowd-sourced information comprises
text content and at least one of: audio content, image content, and
video content; and
utilizing a processor to execute computer code configured
to perform the steps of:
identifying an event using the crowd-sourced information,
wherein the identifying comprises extracting keywords from the
text content of the crowd-sourced information and using at least
one of: audio processing on the audio content, image processing
on the image content, and video processing on the video content;
based on said identifying, classifying the event into a
predefined class using the extracted keywords;
thereafter, invoking a semantic model which is related to
the predefined class, wherein the semantic model identifies
predefined event characteristics, wherein the predefined event
characteristic comprise characteristics corresponding to an
event of the predefined class;
using the semantic model to identify missing information
by:
identifying, from the crowd-sourced information,
event characteristics matching the predefined event
characteristics of the semantic model for the predefined
class of the identified event; and
identifying the missing information by identifying
the predefined event characteristics of the semantic model
that are not matched to any event characteristics from the
crowd-sourced information;
locating one or more individuals in proximity to the event,
and harvesting from the located individuals additional crowd-
sourced information on the event related to the identified missing
information;
4
Appeal2018-007248
Application 13/755,767
said harvesting comprising sending directed questions to
the individuals, the directed questions relating to the identified
missing information; and
completing identification of the event using the harvested
missing information, wherein the completing comprises filling in
the missing information using the harvested information into the
semantic model and identifying the event when the semantic
model is complete.
The Examiner's Rejection
Claims 1, 3, 8, 10, 11, 13, 18, 20-24, and 26 stand rejected under pre-
AIA 35 U.S.C. § I03(a) as being unpatentable over Scofield et al. (US
2013/0132434 Al; May 23, 2013 (filed Nov. 22, 2011)) ("Scofield");
Schultz et al. (US 2013/0222133 Al; Aug. 29, 2013 (filed Feb. 29, 2012))
("Schultz"); and Runciman (US 2006/0282291 Al; Dec. 14, 2006). Final
Act. 6-15. 2
ANALYSIS 3
Appellants assert the Examiner erred in finding the cited references
teach or reasonably suggest invoking a semantic model related to a
2 We note in the body of the rejection, the Examiner mistakenly identifies
the claims are rejected under pre-AIA 35 U.S.C. § I02(e). See Final Act. 6.
From the section heading ("Claim Rejections - 35 USC§ 103") and body of
the rejection, it is clear the Examiner intended the obviousness rejection to
be under 35 U.S.C. § I03(a). See Final Act. 5. Further, Appellants do not
allege any prejudice due to the error and acknowledge the claims stand
rejected under 35 U.S.C. § I03(a). See App. Br. 25. Accordingly, we treat
the Examiner's typographical error as harmless.
3 Throughout this Decision, we have considered the Appeal Brief, filed
February 19, 2018 ("App. Br."); the Reply Brief, filed July 9, 2018 ("Reply
Br."); the Examiner's Answer, mailed May 8, 2018 ("Ans."); and the Final
5
Appeal2018-007248
Application 13/755,767
predefined class into which an event has been classified and wherein the
semantic model identifies predefined event characteristics that comprise
characteristics corresponding to an event of the predefined class. App.
Br. 28-31. In particular, Appellants argue Scofield "makes no mention of
classifying the event and invoking a semantic model." App. Br. 29-30.
Rather, Appellants argue Scofield is limited to identifying characteristics
about a user, telemetry details, and requesting clarification regarding a
previously input location condition. App. Br. 29-30 (citing Scofield ,r,r 38,
40-41 ). Regarding Schultz, Appellants assert that although Schultz teaches
using models to assign tags that represent the provided data, Schultz does
not describe the model as being a "semantic model" or that the model is
related to the predefined class. App. Br. 30.
Additionally, Appellants dispute the Examiner's interpretation of the
claimed semantic model "as a series of inquiries" about an incident. App.
Br. 29 ( quoting Final Act. 4). Rather, Appellants assert a semantic model
( also referred to as an event model) "is a model that includes a hub and
associated parts that make up the model." App. Br. 29 (citing Spec. ,r,r 28,
32, Fig. 7). Moreover, Appellants assert a semantic model "is a specific
model used to identify the necessary parts to provide a complete picture of
the event and to identify which parts are unknown." App. Br. 29.
As an initial matter, we note that Appellants' arguments and
description of a semantic model are not inconsistent with the Examiner's
interpretation. When construing claim terminology during prosecution
before the Office, claims are to be given their broadest reasonable
Office Action, mailed September 18, 201 7 ("Final Act."), from which this
Appeal is taken.
6
Appeal2018-007248
Application 13/755,767
interpretation consistent with the Specification, reading claim language in
light of the Specification as it would be interpreted by one of ordinary skill
in the art. In re Am. Acad. of Sci. Tech Ctr., 367 F.3d 1359, 1364 (Fed. Cir.
2004). We are mindful, however, that limitations are not to be read into the
claims from the Specification. In re Van Geuns, 988 F.2d 1181, 1184 (Fed.
Cir. 1993).
As discussed above, and as set forth in the Specification, an event may
be identified and classified into a predefined class using extracted keywords
from received crowd-sourced information. Spec. ,r 31, Fig. 6. For example,
based on received crowd-sourced information, it may be determined that an
accident (i.e., a general class) is being reported and, further, based on
extracted words from the information, the type of accident is determined to
be a road accident (i.e., a predetermined class of accidents). Spec. ,r 31,
Fig. 6. Once classified as a road accident, an associated semantic model is
invoked. Spec. ,r 32, Fig. 7. Appellants assert (and the Specification
describes) the semantic model (i.e., an event model) comprises a hub
classification and associated parts. Spec. ,r 32, Fig. 7; App. Br. 29. In the
example embodiments illustrated in Figures 6 and 7, the class into which the
event is classified (i.e., "road accident") serves as the "hub classification" of
the semantic model. See Spec., Figs. 6, 7. Surrounding the hub are
"different parts that are needed to provide a complete picture of the event."
App. Br. 29. The Specification describes these different parts as different
facets of an event. Spec. ,r 32. For the example of a "road accident" event,
facets may include "the type of accident, the number of vehicles involved,
whether there are injuries, etc." Spec. ,r 32. The process of identifying these
different facets for an event is referred to as facet discovery and, in facet
7
Appeal2018-007248
Application 13/755,767
discovery, "user inputs are received and processed" so as to identify the
particular information relevant to that facet. Spec. ,r 32.
Consistent with the claim language, a semantic model is associated
with the predefined class into which the event has been classified and further
comprises one or more characteristics (i.e., facets) that correspond to an
event of the predefined class. The characteristics provide additional details
describing the event and may be determined from user input. Thus,
invoking a series of inquiries and questions about an event, as interpreted by
the Examiner, corresponds to identifying appropriate facets (i.e.,
characteristics) relevant to a particular event. Accordingly, the Examiner's
interpretation is consistent with the Specification and how one of ordinary
skill in the art would understand related characteristics of the event would be
gathered.
We next provide a brief review of Scofield and Schultz.
Scofield generally relates to receiving and gathering user reports of
location-based conditions. Scofield ,r 4, Abstract. Scofield describes that
information related to location-based conditions may be received from users
and parsed to extract the desired information (i.e., related to the conditions
of the location). Scofield ,r,r 4, 24. Additionally, Scofield describes a server
receiving such information "may be configured to confirm, clarify, or
identify additional details about a location condition by generating and
presenting queries to users in the proximity of the location." Scofield ,r 4.
Scofield also describes the location-based conditions may include natural
and/or weather conditions ( e.g., fog, visibility, ice formation), natural or
artificial inanimate objects ( e.g., potholes, landslides, oil, downed power
lines), animals, and individuals (e.g., condition of those involved in an
8
Appeal2018-007248
Application 13/755,767
accident). Scofield ,r 33. Further, Scofield discloses that the system may
confirm, clarify, or supplement the information received from the user.
Scofield ,r 41. Scofield teaches detecting various keywords that have a
"known semantic meaning[]" and further detecting and extracting other
keywords necessary to infer a location condition. Scofield ,r 43.
Additionally, Scofield teaches the use of query template sets for confirming
and gathering additional information regarding a location condition.
Scofield ,r 44, Fig. 8.
Schultz generally relates to generating emergency notification based
on aggregate event data. 4 Schultz ,r 1. Schultz receives information ( e.g.,
text, audio, and video) from a plurality of user devices. Schultz ,r,r 1, 14.
Schultz describes an event processor that is configured to correlate the event
data received from the different user devices. Schultz ,r 14. The event
processor also determines an event type based on the received data. Schultz
,r 19.
To recognize an event type from the received data, Schultz teaches the
event processor uses one or more of "pattern detection, heuristic analysis,
inference or deductive processing, object matching, ontological schemes,
and other analysis models." Schultz ,r,r 19, 39. Schultz teaches that event
types may include a broad category of an event, a sub-category of an event,
a specific object type, and any other tag for characterizing the event. Schultz
,r 20. The tagging of an event may continue as more data is received and
event data is correlated and aggregated. Schultz ,r 20. "Tags are generated
to represent an event type to which the event data relates or a description of
4 Schultz also contemplates "embodiments [that] have applicability to non-
emergency events as well." Schultz ,r 13.
9
Appeal2018-007248
Application 13/755,767
one or more elements of the event data." Schultz ,r 38. Additionally,
Schultz teaches providing the results of the analysis (i.e., event tag
determination) to the user and may further query the user for additional
information, such as whether to send help. Schultz ,r,r 69-70.
We do not find Appellants' arguments persuasive of Examiner error.
The Examiner finds, and we agree, Scofield teaches extracting keywords
from the text content of crowd-sourced information to identify an event.
Final Act. 7; see also Scofield ,r,r 24, 33 (extracting keywords from location
condition reports to identify a particular condition (e.g., weather condition or
inanimate objects affecting the roadway). Moreover, after identifying the
incident based on extracted keywords, Scofield teaches the use of (i.e.,
invoking) a query template to solicit information related to the identified
incident from users near the incident. See Scofield ,r,r 24, 40, 41, 44, Fig. 8;
see also Final Act. 7. Although Scofield's "query template" may suggest the
claimed semantic model, the Examiner relies on Schultz to teach presenting
a series of questions and inquiries related to an event (i.e., event
characteristics) to a user. Final Act. 7-8 (citing Schultz ,r,r 39, 69-70,
Figs. 5A, 5B); see also Ans. 20 (the Examiner explaining "Schultz discloses
the feature that based on class/type of an event (e.g.[,] a car accident), event
characteristics (e.g.[,] Location of the event, identity of users, or time of the
event) are identified and a series of questions and inquiries are determined
and presented to a user.") As discussed above, this corresponds to invoking
a semantic model related to a predefined class into which an event has been
classified and wherein the semantic model identifies predefined event
characteristics that comprise characteristics corresponding to an event of the
predefined class, as recited in Appellants' claims.
10
Appeal2018-007248
Application 13/755,767
Appellants also dispute the Examiner's proffered reasoning to
combine the references as articulated in the Final Rejection. App. Br. 26-
28. In particular, Appellants argue the Examiner has not provided
articulated reasoning with rational underpinnings-but rather mere
conclusory statements-in support of the proffered motivation to combine
the cited references. App. Br. 26-28 ( citing KSR Int 'l v. Teleflex Inc., 550
U.S. 398, 418 (2007)). Moreover, Appellants argue "Schultz, Runciman,
and Scofield are directed to completely separate problems and therefore are
directed to completely separate solutions" and, accordingly, an ordinarily
skilled artisan would not have been motivated to combine the references as
proposed by the Examiner. App. Br. 28.
Our reviewing court guides it is irrelevant that the prior art and the
present invention may have different purposes. See Nat 'l Steel Car, Ltd. v.
Canadian Pac. Ry., Ltd., 357 F.3d 1319, 1339 (Fed. Cir. 2004) ("A finding
that two inventions were designed to resolve different problems ... is
insufficient to demonstrate that one invention teaches away from another.").
That is, it is sufficient that references suggest doing what Appellants did,
although the Appellants' particular purpose was different from that of the
references. In re Heck, 699 F.2d 1331, 1333 (Fed. Cir. 1983) (citing In re
Gershon, 372 F.2d 535, 538-39 (CCPA 1967)). "Obviousness is not to be
determined on the basis of purpose alone." In re Graf, 343 F.2d 774, 777
(CCPA 1965). Thus, Appellants' arguments that the cited references have
different purposes (i.e., "are directed to completely separate solutions") are
not persuasive of Examiner error.
The key to supporting any prima facie conclusion of obviousness
under 35 U.S.C. § 103 is the clear articulation of the reason(s) why the
11
Appeal2018-007248
Application 13/755,767
claimed invention would have been obvious. The Supreme Court in KSR
(550 U.S. at 418) noted that the analysis supporting a rejection under
35 U.S.C. § 103 should be made explicit. The Federal Circuit has stated that
"rejections on obviousness grounds cannot be sustained by mere conclusory
statements; instead, there must be some articulated reasoning with some
rational underpinning to support the legal conclusion of obviousness." In re
Kahn, 441 F.3d 977, 988 (Fed. Cir. 2006), cited with approval in KSR, 550
U.S. at 418.
"The presence or absence of a motivation to combine references in an
obviousness determination is a pure question of fact." Alza Corp. v. Mylan
Labs., Inc., 464 F.3d 1286, 1289 (Fed. Cir. 2006) (quoting In re Gartside,
203 F.3d 1305, 1316 (Fed. Cir. 2000)). Moreover, the law does not require
that the references expressly articulate a motivation to combine. See, e.g.,
Arctic Cat Inc. v. Bombardier Recreational Prods. Inc., 876 F.3d 1350, 1359
(Fed. Cir. 2017). The "[m]otivation to combine may be found in many
different places and forms .... " Allergan, Inc. v. Sandoz, Inc., 726 F.3d
1286, 1292 (Fed. Cir. 2013); see also Alza, 464 F.3d at 1294 (stating that the
motivation to combine does not have to be explicitly stated in the prior art).
"[ A ]ny need or problem known in the field of endeavor at the time of
invention and addressed by the patent can provide a reason for combining"
references. KSR, 550 U.S. at 420; see also DyStar Textilfarben GmbH v.
C.H. Patrick Co., 464 F.3d 1356, 1368 (Fed. Cir. 2006) ("[A]n implicit
motivation to combine" may result from a desire to make a product or
process "stronger, cheaper, cleaner, faster, lighter, smaller, more durable, or
more efficient.").
12
Appeal2018-007248
Application 13/755,767
Here, the Examiner finds an ordinarily skilled artisan would have
been motivated to combine the teachings of Schultz with those of Scofield
"to improve classifying of events based on event information and messages
using an event model." Final Act. 8. Additionally, the Examiner finds an
ordinarily skilled artisan would have been motivated to combine the
teachings of Runciman with Scofield and Schultz "to improve efficiency of
the method by ensuring that sufficient amount of event information to
identify an incident has been gathered." Final Act. 9.
Moreover, in response to Appellants' arguments, the Examiner notes
Schultz expressly discloses "'a knowledge base may be maintained for
enabling the ongoing and subsequent analysis of events to support response
action improvement, event classification improvement, situational/event
model development and the like."' Ans. 15 ( quoting Schultz ,r 7 6 ( emphasis
omitted)). The Examiner explains the advantage of classifying events using
a data model would be to enable real-time analysis of event data to support
response action improvement. Ans. 15. The Examiner notes the motivation
to combine Schultz and Scofield, therefore, is in the reference itself (i.e.,
Schultz) as well as in the knowledge generally available to one of ordinary
skill in the art. Ans. 15 ( citing Schultz ,r,r 14, 34, 44, 68, 7 6). Similarly,
regarding Runciman, the Examiner explains the motivation to combine
Runciman with Scofield and Schultz comes from the teachings of Runciman
(as well as the knowledge of one skilled in the art). Ans. 16. In particular,
the Examiner finds Runciman teaches in order to properly classify an event,
related data may be collected by asking a series of questions. Ans. 16 ( citing
Runciman ,r,r 34, 145, 148, 149). As the Examiner explains, "selecting
13
Appeal2018-007248
Application 13/755,767
questions and collecting [a] sufficient amount of data and answers, increase
the chance of proper classification of the incident." Ans. 16.
Scofield teaches "receiving and aggregating reports of location-based
conditions received from users, either spontaneously ('I just witnessed an
accident') or in response to a query (e.g., 'did you encounter road ice one
kilometer ago?')." Scofield, Abstract. As identified by the Examiner,
Schultz teaches a data analysis module that "employs various pattern
detection, heuristic analysis, inference or deductive processing, object
matching, ontological schemes, and other analysis models .... " Schultz
,r 39; see also Final Act. 8. Schultz describes that the data analysis module
executes in connection with one or more models for enabling the event to be
recognized. Schultz ,r 39. Moreover, as identified by the Examiner, Schultz
teaches using a knowledge base (of the determined data) for use in
improving responses to events or "defining more accurate event models."
Schultz ,r 44; see also Ans. 15. Thus, the Examiner has set forth articulated
reasoning with rational underpinning that one of ordinary skill in the art
would have been motivated to combine Schultz's teaching of identifying
event characteristics with Scofield' s aggregation of location-based condition
reports to "improve classifying of events based on event information and
messages using an event model." Ans. 15; see also KSR, 550 U.S. at 418.
Runciman, as identified by the Examiner (see, e.g., Ans. 16),
describes a system "to facilitate incident data collection and classification
.... " Runciman ,r 145. Runciman describes a classification model and
hierarchy comprising levels of questions (i.e., Main Questions at a high level
followed by "specific questions to achieve the desired level of
classification"). Runciman ,r,r 148-149. We agree with the Examiner's
14
Appeal2018-007248
Application 13/755,767
explanation that one of ordinary skill in the art would have been motivated
to combine Runciman's teaching of "selecting questions and collecting [a]
sufficient amount of data" with the combination of Schultz and Scofield "to
improve the efficiency of [incident classification] by ensuring that [a]
sufficient amount of event information to identify an incident has been
gathered." Ans. 16.
For the reasons discussed supra, we are unpersuaded of Examiner
error. Accordingly, we sustain the Examiner's rejection of representative
independent claim 1. For similar reasons, we also sustain the Examiner's
rejection of independent claims 10, 11, and 20, which recite commensurate
limitations and were not argued separately. See App. Br. 26-31; see also
37 C.F.R. § 4I.37(c)(l)(iv) (2017). Additionally, we sustain the Examiner's
rejection of claims 3, 8, 13, 18, 21-24, and 26, which depend directly or
indirectly therefrom and were not argued separately. See 37 C.F.R.
§ 4I.37(c)(l)(iv).
DECISION
We affirm the Examiner's decision rejecting claims 1, 3, 8, 10, 11, 13,
18, 20-24, and 26 under pre-AIA 35 U.S.C. § 103(a).
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.
§ 4I.50(f).
AFFIRMED
15