International Business Machines Corporation

Patent Trials and Appeals Board

Dec 31, 2020

2019003203 (P.T.A.B. Dec. 31, 2020)

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.
14/573,240 12/17/2014 Donna K. Byron AUS920140373US1 9276
50170 7590 12/31/2020
IBM CORP. (WIP)
c/o WALDER INTELLECTUAL PROPERTY LAW, P.C.
445 CRESTOVER CIRCLE
RICHARDSON, TX 75080
EXAMINER
CHEN, YING YU
ART UNIT PAPER NUMBER
2125
MAIL DATE DELIVERY MODE
12/31/2020 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 DONNA K. BYRON, ALEXANDER PIKOVSKY, and
MARY D. SWIFT
Appeal 2019-003203
Application 14/573,240
Technology Center 2100
Before MICHAEL J. STRAUSS, JEREMY J. CURCURI, and
PHILLIP A. BENNETT, Administrative Patent Judges.
CURCURI, Administrative Patent Judge.
DECISION ON APPEAL
STATEMENT OF THE CASE
Pursuant to 35 U.S.C. § 134(a), Appellant1 appeals from the
Examiner’s decision to reject claims 1, 4–11, and 14–23. We have
jurisdiction under 35 U.S.C. § 6(b).
We REVERSE.

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 International
Business Machines. Appeal Br. 2.
Appeal 2019-003203
Application 14/573,240
2
CLAIMED SUBJECT MATTER
The claims are directed to “mechanisms for expanding training
questions through contextualizing feature search.” Spec. ¶ 1. Claim 1,
reproduced below, is illustrative of the claimed subject matter:
1. A method, in a data processing system having a
processor and a memory and implementing a Question and
Answer (QA) system, the method comprising:
receiving, by the QA system, a training question for
processing by the QA system;
processing, by the QA system, the training question to
generate an answer to the training question, from a portion of
content in a corpus of information;
identifying, by the QA system, a repeatable pattern of
content present in the portion of content in association with the
answer to the training question;
applying, by the QA system, the repeatable pattern of
content to other portions of content to generate at least one
additional training question and at least one additional entry in
a ground truth data structure to thereby expand a set of training
questions and expand the ground truth data structure;
storing the expanded set of training questions and
expanded ground truth data structure; and
training the QA system using the expanded set of training
questions and expanded ground truth data structure, wherein the
portion of content is an unstructured natural language portion of
content and the repeatable pattern of content is identified by
identifying a structure within the unstructured natural language
portion of content that comprises a plurality of entries, each
entry having the repeatable pattern of content.

Appeal 2019-003203
Application 14/573,240
3
REFERENCES
The prior art relied upon by the Examiner is:
Name Reference Date
Brill US 2006/0173834 A1 Aug. 3, 2006
Agichtein US 7,873,624 B2 Jan. 18, 2011
Baker US 8,346,792 B1 Jan. 1, 2013

REJECTIONS
Claims 1, 4–9, 11, 14–19, and 21–23 are rejected under 35 U.S.C.
§ 103 as obvious over Baker and Agichtein. Final Act. 9–61.
Claims 10 and 20 are rejected under 35 U.S.C. § 103 as obvious over
Baker, Agichtein, and Brill. Final Act. 61–69.
OPINION
The Obviousness Rejection of Claims 1, 4–9, 11, 14–19, and 21–23 over
Baker and Agichtein
The Examiner finds Baker and Agichtein teach all limitations of claim
1. Final Act. 9–15.
In particular, the Examiner finds Baker teaches “identifying, by the
QA system, a repeatable pattern of content present in the portion of content
in association with the answer to the training question” (claim 1) as recited
in claim 1. See Final Act. 10–11 (citing Baker, Fig. 4, col. 10, ll. 7–9, 15–21)
(“a text fragment [in Fig. 4, box 404] containing the key term ‘Rye’ and this
pattern is repeated at least three times”); see also Ans. 4.
The Examiner further finds Baker teaches “applying, by the QA
system, the repeatable pattern of content to other portions of content to
generate at least one additional training question and at least one additional
entry in a ground truth data structure to thereby expand a set of training
Appeal 2019-003203
Application 14/573,240
4
questions and expand the ground truth data structure” (claim 1) as recited in
claim 1. See Final Act. 11–13 (citing Baker, Fig. 4, col. 5, ll. 11–22, 66–67,
col. 10, ll. 23–30, col. 11, ll. 1–2, 16–19, 34–37, col. 16, ll. 16–21)
(“generate query templates, which are then applied to other portion of
content to generate at least one additional question and at least one
additional answer”); see also Ans. 5–7.
Among other arguments, Appellant presents the following principal
arguments:
Baker does not teach “identifying, by the QA system, a repeatable
pattern of content present in the portion of content in association with the
answer to the training question” (claim 1) because
[w]hile Baker teaches the generation of a template based on the
tags of a document associated with a term corresponding to a
search query, Baker does not teach the identification of a
repeatable pattern of content present in the portion of content
associated with an answer to a training question. The templates
in Baker are merely the tags associated with a portion of a
document that has matching terms to that of the search query.
The generation of a template in Baker is not based on the
identification of a repeatable pattern of content in a portion of
content associated with an answer to a training question. That is,
there is no identification in Baker of any repeatable pattern
within a portion of content associated with an answer to a
training question. All Baker does is match terms of the search
query with terms in the document and extract the embedded tags
associated with those terms. There is no consideration as to
whether or not the tags in the document are part of a repeated
pattern of content in the portion of content associated with an
answer to a training question.
Appeal Br. 27; see also Reply Br. 2–4.
Baker does not teach “applying, by the QA system, the repeatable
pattern of content to other portions of content to generate at least one
Appeal 2019-003203
Application 14/573,240
5
additional training question and at least one additional entry in a ground
truth data structure to thereby expand a set of training questions and expand
the ground truth data structure” (claim 1) because
[n]owhere in [Baker’s] process is there any teaching or
suggestion to actually identify a repeatable pattern of content in
the portion of content associated with an answer to a training
question and then generating at least one additional training
question, and at least one additional entry in a ground truth data
structure. To the contrary, Baker only teaches generating
synthetic search queries which are just search terms.
Appeal Br. 28.
The template extraction rules do not in fact apply any identified
repeatable pattern of content to other portions of content to
generate at least one additional training question and at least
one additional entry in a ground truth data structure to thereby
expand a set of training questions and expand the ground truth
data structure.
Appeal Br. 28.
Appellant’s arguments persuade us that the Examiner erred in finding
Baker teaches the following key disputed limitations:
identifying, by the QA system, a repeatable pattern of
content present in the portion of content in association with the
answer to the training question;
applying, by the QA system, the repeatable pattern of
content to other portions of content to generate at least one
additional training question and at least one additional entry in
a ground truth data structure to thereby expand a set of training
questions and expand the ground truth data structure.
Claim 1.
Regarding the “repeatable pattern of content,” Appellant’s
Specification describes
[e]xamples of such repeating patterns of content (e.g., repeatable
contexts) include tabular data, lists/bulleted data, and outlined
Appeal 2019-003203
Application 14/573,240
6
data with headings and subheadings. If such a repeatable pattern
is found, the structure of the repeatable pattern is used to
automatically identify sibling elements and question/answer
pairs are generated for each sibling of the repeatable pattern.
Spec. ¶ 17.
Although the Specification does not provide an explicit definition for
“repeatable pattern of content,” we interpret the claim language in light of
the Specification.
We agree with Appellant that Baker’s repetition of the term “Rye”
(Baker, Fig. 4, block 404) and Baker’s query template (Baker, Fig. 4, block
406) do not teach a “repeatable pattern of content” as recited in claim 1.
Baker discloses locating terms in a document that match terms in a
search query. Baker, Fig. 4, block 404. Thus, Baker discloses repeated
content (repetition of the term “Rye”). However, we do not readily discern
any consideration as to a repeated pattern of tags that provide a repeatable
context for the term “Rye.” We only see the locating of each occurrence of
the term “Rye,” and counting the number of occurrences of the term “Rye.”
See Baker, Fig. 4, blocks 404, 406. Put another way, we interpret the
claimed “repeatable pattern of content” (emphasis added) as requiring some
repeated context for the repeated content, that is, as requiring something
more than merely repeated content (repetition of the term “Rye”). See Spec.
¶ 17 (“Examples of such repeating patterns of content (e.g., repeatable
contexts) include tabular data, lists/bulleted data, and outlined data with
headings and subheadings.”).
Regarding the “applying” step, because we do not readily see any
consideration as to a repeated pattern of tags that provide a repeatable
context for the term “Rye,” we also do not see any application of such a
Appeal 2019-003203
Application 14/573,240
7
pattern to generate additional questions and entries in a ground truth data
structure.
We, therefore, do not sustain the Examiner’s rejection of claim 1.
We also do not sustain the Examiner’s rejection of claims 4–9 and 22,
which depend from claim 1.
Independent claim 11 recites the same key disputed limitations.
We, therefore, do not sustain the Examiner’s rejection of claim 11.
We also do not sustain the Examiner’s rejection of claims 14–19 and
23, which depend from claim 11.
Independent claim 21 recites the same key disputed limitations.
We, therefore, do not sustain the Examiner’s rejection of claim 21.

The Obviousness Rejection of Claims 10 and 20 over Baker, Agichtein, and
Brill
The Examiner does not find Brill cures the deficiency of Baker and
Agichtein. See Final Act. 61–69; see also Ans. 21–23.
We, therefore, do not sustain the Examiner’s rejection of claims 10
and 20.
CONCLUSION
The Examiner’s decision to reject claims 1, 4–11, and 14–23 is
reversed.

Appeal 2019-003203
Application 14/573,240
8
DECISION SUMMARY
In summary:
Claims
Rejected
35 U.S.C.
§
Reference(s)/Basis Affirmed Reversed
1, 4–9, 11,
14–19, 21–
23
103 Baker, Agichtein 1, 4–9, 11,
14–19, 21–
23
10, 20 103 Baker, Agichtein,
Brill
10, 20
Overall
Outcome
1, 4–11, 14–
23
REVERSED