International Business Machines Corporation

Patent Trials and Appeals Board

Dec 17, 2021

2020002739 (P.T.A.B. Dec. 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.
14/297,267 06/05/2014 Corville O. Allen AUS920140054US1 2300
45725 7590 12/17/2021
Walder Intellectual Property Law PC
445 Crestover Circle
Richardson, TX 75080
EXAMINER
MCGHEE, TRACY M
ART UNIT PAPER NUMBER
2154
MAIL DATE DELIVERY MODE
12/17/2021 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 CORVILLE O. ALLEN, ALBERT A. CHUNG,
ANDREW R. FREED, and DORIAN B. MILLER
_______________

Appeal 2020-002739
Application 14/297,267
Technology Center 2100
_______________

Before ROBERT E. NAPPI, JAMES W. DEJMEK, and
JOYCE CRAIG, Administrative Patent Judges.

DEJMEK, Administrative Patent Judge.

DECISION ON APPEAL
Appellant1 appeals under 35 U.S.C. § 134(a) from a Non-Final
Rejection of claims 1–13 and 15–21. Appellant has canceled claim 14. See
Amdt. 8 (filed Nov. 2, 2016). We have jurisdiction over the remaining
pending claims under 35 U.S.C. § 6(b). See Ex parte Lemoine, 46 USPQ2d
1420, 1423 (BPAI 1994) (precedential).
We reverse.

1 Throughout this Decision, we use the word “Appellant” to refer to
“applicant” as defined in 37 C.F.R. § 1.42 (2019). Appellant identifies
International Business Machines Corporation as the real party in interest.
Appeal Br. 2.
Appeal 2020-002739
Application 14/297,267

2
STATEMENT OF THE CASE
Introduction
Appellant’s disclosed and claimed invention generally relates to
“time-based optimization of answer generation by a question and answer
system. Spec. ¶ 1. More specifically, Appellant describes generating an
answer to an input question wherein, although the input question does not
specify a temporal categorization, the question is processed “to identify a
temporal categorization of the input question indicating an intended time
context for an answer.” Spec. ¶ 5. Further,
analysis of a corpus of information based on the temporal
categorization of the input question and temporal categorizations
of sources of information in the corpus of information to
determine, for each source of information, a degree of
correspondence of the temporal categorization of the input
question to a temporal categorization of the source of
information.
Spec. ¶ 5. Based on the degrees of correspondence between the temporal
categorizations of the input question and the sources of information, an
answer is generated and presented to the user. Spec. ¶ 5.
Claim 1 is illustrative of the subject matter on appeal and is
reproduced below with the disputed limitations emphasized in italics:
1. A method, in a data processing system comprising at least
one processor and at least one memory, wherein the at least one
memory comprises instructions that are executed by the at least
one processor to specifically configure the at least one processor
to implement a question answering (QA) system, comprising a
QA pipeline and a temporal categorization engine, for generating
an answer to an input question, the method comprising:
processing, by the temporal categorization engine of the
QA system implemented by the data processing system, a
received input question to identify a temporal categorization of
Appeal 2020-002739
Application 14/297,267

3
the input question indicating an intended time context for an
answer to the input question at least by performing natural
language processing on the input question to identify a domain
of the input question, wherein the domain is a subject matter
domain specifying a subject matter of the input question, wherein
the input question does not specify a temporal categorization of
the input question, and wherein the temporal categorization of
the input question is identified based on an association of the
domain of the input question with a temporal categorization
corresponding to the domain, and wherein the data processing
system processes input questions associated with a plurality of
different subject matter domains and at least two subject matter
domains in the plurality of different subject matter domains have
different corresponding temporal categorizations;
performing, by the temporal categorization engine,
analysis of a corpus of information based on the temporal
categorization of the input question and temporal
categorizations of sources of information in the corpus of
information to determine, for each source of information, a
degree of correspondence of the temporal categorization of the
input question to a temporal categorization of the source of
information;
generating, by the QA pipeline of the QA system
implemented by the data processing system, at least one answer
to the input question based on the degrees of correspondence of
the sources of information determined by the temporal
categorization engine; and
outputting, by the QA pipeline of the QA system
implemented by the data processing system, to a source
computing device that is a source of the input question, the at
least one answer to the input question.

The Examiner’s Rejection
Claims 1–13 and 15–21 stand rejected under 35 U.S.C. § 103 as being
unpatentable over Roberts (US 2011/0153312 A1; June 23, 2011) and Jones
Appeal 2020-002739
Application 14/297,267

4
et al. (US 2010/0131538 A1; May 27, 2010) (“Jones”). Non-Final Act. 3–
16.

ANALYSIS2
In rejecting claim 1, inter alia, the Examiner relies on the combined
teachings of Roberts and Jones. See Non-Final Act. 3–16. We begin our
analysis with a brief review of these references.
Roberts generally relates to a system for answering natural language
questions. Roberts ¶ 1. Roberts describes receiving an input question and
matching the received question with a linked pair of questions and answers
stored in a database. Roberts ¶ 37. Additionally, Roberts describes the
system as configured to “determine whether a question received by the
system is time dependent.” Roberts ¶¶ 51–52, 68–69; see also Roberts
¶¶ 165 (explaining “some types of question depend upon ‘current
information’, e.g. ‘What is tomorrow’s weather forecast?’”), 176 (providing
examples of time-dependent questions relating to weather, financial markets,
and sports results). For time-dependent questions, Roberts states the system
may query a real-time data feed, rather than a database of stored question
and answer pairs, to generate an answer. Roberts ¶¶ 68–69, 176, 178.
Jones generally relates to “expanding search queries and, more
specifically, to identifying and expanding queries which are implicitly
temporally qualified.” Jones ¶ 1. In a disclosed approach, Jones describes

2 Throughout this Decision, we have considered the Appeal Brief, filed
September 25, 2019 (“Appeal Br.”); the Reply Brief, filed February 26, 2020
(“Reply Br.”); the Examiner’s Answer, mailed January 9, 2020 (“Ans.”);
and the Non-Final Office Action, mailed May 1, 2019 (“Non-Final Act.”),
from which this Appeal is taken.
Appeal 2020-002739
Application 14/297,267

5
determining whether a received query, which does not have an explicit
temporal reference, has an implicit temporal reference. Jones ¶ 4. Jones
describes this determination is made “with reference to a first query term
included in the first search query and query log data that relate a plurality of
previous query terms included in previous search queries to specific
temporal references included in the previous search queries.” Jones ¶ 4.
That is, based on a query log analysis, a query term is analyzed to determine
if it generally has been associated with a temporal reference and, if so, the
received query is modified to include an explicit temporal reference. Jones
¶¶ 15–16, 21–22. Jones describes that “[s]earch results are then generated
that are responsive to the reformulated query.” Jones ¶ 21.
Appellant disputes the Examiner’s conclusion that the combination of
Roberts and Jones teaches the particular technique of claimed invention.
Appeal Br. 6–22; Reply Br. 2–8. More particularly, Appellant argues, inter
alia, that neither Roberts nor Jones (alone or in combination) teach:
determining what the temporal categorization is of a
question based on the domain of the question and the association
of a temporal categorization with the domain, and determining
an answer to the question based on degrees of correspondence of
temporal categorizations of domains of sources of information
with the determined temporal categorization of the input
question based on the determined domain of the input question.
Reply Br. 4; see also Appeal Br. 10–12 (asserting Roberts does not describe
temporal categorizations corresponding to domains of input questions), 17–
18 (asserting Jones merely reformulates a search query based on a search
query log analysis and returns ranked results based on previous search
queries). Additionally, Appellant argues that contrary to the Examiner’s
findings (see Non-Final Act. 3–4), Roberts fails to teach performing an
Appeal 2020-002739
Application 14/297,267

6
analysis of a corpus of information based on the temporal categorizations of
the input question and that of the sources of information. Appeal Br. 10–13
(asserting that Roberts merely describes a generic “means” to determine
whether a search query is time dependent).
Based on our review of Roberts and Jones, we find Appellant’s
arguments persuasive of Examiner error. More particularly, the portions of
Roberts and Jones relied upon by the Examiner (see, e.g., Non-Final Act. 3–
7 (citing Roberts ¶¶ 51–52, 69; Jones ¶¶ 9, 16, 19–20, 22–35)) generally
relate to analyzing the received search query to determine if it is time
dependent (as in Roberts) or has an implicit temporal reference (as in Jones).
As discussed above, Roberts teaches that if a question is time dependent, a
real-time data feed, rather than a database of stored questions and answers, is
used to provide the search result. See, e.g., Roberts ¶¶ 51, 69. However, the
Examiner has not provided sufficient evidence or technical reasoning that
Roberts teaches an analysis of the corpus of information based on a temporal
categorization of the input search query. Instead, Roberts merely describes
that the disclosed system includes a means for determining whether the input
query is time dependent. See Roberts ¶ 69.
In addition, as discussed above, Jones teaches reformulating an input
search query based on an analysis of a log of previous queries that relate to a
query term so that the received search query can be reformulated. See, e.g.,
Jones ¶¶ 15–16, 19, 21. Although Jones describes ranking the results to
produce a temporally biased ranking (see, e.g., Jones ¶ 33), the Examiner
has not provided sufficient evidence or technical reasoning that Jones
analyzes a temporal categorization of sources of information in the corpus of
Appeal 2020-002739
Application 14/297,267

7
information to determine a degree of correspondence with a temporal
categorization of the input search query, as recited in the claims.
For the reasons discussed supra, we are persuaded of Examiner error.
Accordingly, we do not sustain the Examiner’s rejection of independent
claim 1. For similar reasons, we do not sustain the Examiner’s rejection of
independent claims 19 and 20, which recite similar limitations. In addition,
we do not sustain the Examiner’s rejection of claims 2–13, 15–18, and 21,
which depend directly or indirectly from independent claim 1.

CONCLUSION
We reverse the Examiner’s decision rejecting claims 1–13, and 15–21
under 35 U.S.C. § 103.

DECISION SUMMARY
Claim(s)
Rejected
35 U.S.C. § Reference(s)/Basis Affirmed Reversed
1–13, 15–
21
103 Roberts, Jones 1–13, 15–
21

REVERSED