International Business Machines Corporation

Patent Trials and Appeals Board

Aug 2, 2021

2020002642 (P.T.A.B. Aug. 2, 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.
15/010,141 01/29/2016 Artem Barger IL920150149US1 8644
48915 7590 08/02/2021
CANTOR COLBURN LLP-IBM YORKTOWN
20 Church Street
22nd Floor
Hartford, CT 06103
EXAMINER
SECK, ABABACAR
ART UNIT PAPER NUMBER
2122
NOTIFICATION DATE DELIVERY MODE
08/02/2021 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):
usptopatentmail@cantorcolburn.com
PTOL-90A (Rev. 04/07)

UNITED STATES PATENT AND TRADEMARK OFFICE
________________

BEFORE THE PATENT TRIAL AND APPEAL BOARD
________________

Ex parte ARTEM BARGER, ROY LEVIN, and
HAGGAI ROITMAN
________________

Appeal 2020-002642
Application 15/010,141
Technology Center 2100
________________

Before BRADLEY W. BAUMEISTER, JASON V. MORGAN, and
RUSSELL E. CASS, Administrative Patent Judges.

Opinion for the Board filed by Administrative Patent Judge MORGAN.

Opinion Dissenting filed by Administrative Patent Judge BAUMEISTER.

MORGAN, 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 1–6, 8–13, and 15–20. Claims 7 and 14
are canceled. Appeal Br. 23, 26. We have jurisdiction under 35 U.S.C.
§ 6(b). We reverse.

1 “Appellant” refers to “applicant” as defined in 37 C.F.R. § 1.42. Appellant
identifies the real party in interest as International Business Machines
Corporation. Appeal Br. 3.
Appeal 2020-002642
Application 15/010,141

2

SUMMARY OF THE DISCLOSURE
Appellant’s claimed subject matter relates to a pseudo-relevance
feedback system
that determines an optimized relevance model for a search
query by utilizing a posterior relevance model to estimate the
likelihood that an initial set of top-K retrieved documents
would be retrieved given the posterior relevance model, re-
ranking the top-K documents based on their respective
estimates of likelihood of retrieval, determining a rank
similarity between the initial ranking of the top-K documents
and the re-ranking of the top-K documents, updating one or
more model parameters of the posterior relevance model based
on the rank similarity, and iteratively performing the above
process until the rank similarity is maximized, at which point,
the optimized relevance model is obtained.
Abstract.
ILLUSTRATIVE CLAIM
1. A method for enhancing robustness of pseudo-relevance
feedback models using query drift minimization, the method
comprising:
determining, by a computer processor, a first set of search
results returned for a search query, wherein the first set of
search results is ranked in accordance with a first ranking;
determining, by the computer processor, a first relevance
model;
determining, by the computer processor and based at least in
part on the first relevance model, a respective probability of
retrieval of each search result in the first set of search results;
determining, by the computer processor, a second ranking for
the first set of search results based at least in part on the
respective probability of retrieval of each result in the set of
search results;
determining, by the computer processor, a rank similarity
between the first ranking and the second ranking;
Appeal 2020-002642
Application 15/010,141

3

determining, by the computer processor, a second relevance
model by updating at least one model parameter of the first
relevance model based at least in part on the rank similarity;
determining, by the computer processor, that the second
relevance model is an optimized relevance model for the search
query; and
determining, by the computer processor, a second set of search
results for the search query using the second relevance model,
wherein the first set of search results and the second set of
search results comprise a set of common search results, and
wherein determining the second set of search results using the
second relevance model causes a rank similarity between a
ranking of the set of common search results in the first set of
search results and a ranking of the set of common search results
in the second set of search results to be maximized and query
drift to be minimized without performing query anchoring of
the search query.
REJECTION
The Examiner rejects claims 1–6, 8–13, and 15–20 under 35 U.S.C.
§ 101 as being directed to patent-ineligible subject matter. Final Act. 2–12;
Ans. 3–19 (newly rejecting the claims under 35 U.S.C. § 101 in light of the
2019 Revised Guidance discussed below).
PRINCIPLES OF LAW
To constitute patent-eligible subject matter, an invention must be a
“new and useful process, machine, manufacture, or composition of matter,
or [a] new and useful improvement thereof.” 35 U.S.C. § 101. There are
implicit exceptions to the categories of patentable subject matter identified
in 35 U.S.C. § 101, including: (1) laws of nature; (2) natural phenomena;
and (3) abstract ideas. Alice Corp. v. CLS Bank Int’l, 573 U.S. 208, 216
(2014). The U.S. Supreme Court has set forth a framework for
Appeal 2020-002642
Application 15/010,141

4

distinguishing patents with claims directed to these implicit exceptions
“from those that claim patent-eligible applications of those concepts.” Id. at
217 (citing Mayo Collaborative Servs. v. Prometheus Labs., Inc., 566 U.S.
66 (2012)). The evaluation follows a two-part framework: (1) determine
whether the claim is directed to a patent-ineligible concept, e.g., an abstract
idea; and (2) if so, then determine whether any element, or combination of
elements, in the claim is sufficient to ensure that the claim amounts to
significantly more than the patent-ineligible concept itself. See id. at 217–18.
Under U.S. Patent and Trademark Office USPTO guidance, we first
look to whether the claim recites:
(1) any judicial exceptions, including certain groupings
of abstract ideas (i.e., mathematical concepts, certain methods
of organizing human activity such as a fundamental economic
practice, or mental processes) (MPEP § 2106.04(a)(2); USPTO,
2019 Revised Patent Subject Matter Eligibility Guidance, 84
Fed. Reg. 50, 52 (Jan. 7, 2019) (“2019 Revised Guidance”)
(step 2A, prong one); USPTO, October 2019 Update: Subject
Matter Eligibility, 3–9, available at https://www.uspto.gov/
sites/default/files/documents/peg_oct_2019_update.pdf (Oct.
17, 2019) (“Oct. 2019 Update”)); and
(2) additional elements that integrate the judicial
exception into a practical application (MPEP §§ 2106.04(d),
2106.05(a)–(c), (e)–(h); 2019 Revised Guidance, 84 Fed. Reg.
at 54–55; Oct. 2019 Update at 10–14).
Appeal 2020-002642
Application 15/010,141

5

Only if a claim (1) recites a judicial exception and (2) does not
integrate that exception into a practical application, do we then look to
whether the claim:
(3) adds a specific limitation beyond the judicial
exception that is not “well-understood, routine, conventional”
in the field; or
(4) simply appends well-understood, routine,
conventional activities previously known to the industry,
specified at a high level of generality, to the judicial exception.
MPEP § 2106.05(d); 2019 Revised Guidance, 84 Fed. Reg. at 56; Oct. 2019
Update at 16.
ANALYSIS
35 U.S.C. § 101—Step 1
Section 101 provides that “[w]hoever invents or discovers any new
and useful process, machine, manufacture, or composition of matter, or any
new and useful improvement thereof, may obtain a patent therefor, subject to
the conditions and requirements of this title.” 35 U.S.C. § 101. Independent
claim 1 recites a “method for enhancing robustness of pseudo-relevance
feedback models using query drift minimization,” independent claim 8
recites a “system for enhancing robustness of pseudo-relevance feedback
models using query drift minimization,” and independent claim 15 recites a
“computer program product comprising a non-transitory storage medium
readable by a processing circuit, the storage medium storing instructions
executable by the processing circuit to cause a method to be performed.”
The remaining claims do not differ in statutory class from the respective
Appeal 2020-002642
Application 15/010,141

6

claims from which they depend. As such, the claims are directed to statutory
classes of invention within 35 U.S.C. § 101 (i.e., processes, machines, and
manufactures). See also Ans. 4, 9, 14.
35 U.S.C. § 101—Step 2A, prong one
The Examiner determines that claim 1 recites a mental process—i.e.,
concepts performed in the human mind including observation, evaluation,
judgment, opinion—which is a category of abstract idea. Id. at 4–5; Final
Act. 2–4; MPEP 2106.04(a)(2).III. Appellant contends the Examiner erred
because the Examiner merely makes conclusory statements, unsupported by
any analysis, “that the alleged abstract idea ‘is a mental process.’” Reply Br.
4; Appeal Br. 12. Appellant argues, in particular, that “the claim limitations,
as currently recited, cannot, as a practical matter, be performed entirely in a
human’s mind, even if aided with pen and paper.” Reply Br. 5.
We agree with the Examiner that claim 1 at least recites an abstract
idea in the form of a mental process. Ans. 4–5. In particular, claim 1 recites
“determining . . . a first set of search results” and “determining . . . a second
set of search results . . . wherein the first set of search results and the second
set of search results comprise a set of common search results.”
Searching for and retrieving data is a form of observation, evaluation,
judgment, or opinion (i.e., determining which results are most relevant to a
particular query), and thus represents an abstract idea in the form of a mental
process. MPEP § 2106.04(a)(2).III; Intell. Ventures I LLC v. Erie Indem.
Co., 850 F.3d 1315, 1327 (Fed. Cir. 2017) (“creating an index and using that
index to search for and retrieve data” is an abstract idea). Catalogues of titles
of “one or another group of compositions for any of the numerous practical
needs involved in the process of storing” works date back thousands of
Appeal 2020-002642
Application 15/010,141

7

years. Samuel N. Kramer, The Oldest Literary Catalogue: A Sumerian List
of Literary Compositions Compiled about 2000 B.C., 88 Bulletin of the
American Schools of Oriental Research 10, 13 (Dec. 1942). The oldest such
surviving catalogues even shared some, but not all titles, showing “that the
guiding principles” guiding the arrangement of each catalogue “were not
identical.” Id. at 19.
Because determining first and second search results (i.e., searching for
and retrieving data)—even when the first and second search results comprise
a set of common results—represents a form of mental process (i.e.,
observation, evaluation, judgment, opinion), we determine that claim 1
recites an abstract idea.
We further agree with the dissent’s analysis under step 2A, prong one,
to the extent that claim 1 at least recites mathematical concepts
(mathematical relationships, mathematical formulas or equations,
mathematical calculations), which are also a form of abstract idea. MPEP
§ 2106.04(a). But merely reciting an abstract idea or other judicial exception
is insufficient to make a claim patent-ineligible. See, e.g., Bascom Glob.
Internet Servs., Inc. v. AT&T Mobility LLC, 827 F.3d 1341, 1348–52 (Fed.
Cir. 2016) (claims reciting the abstract idea of filtering nonetheless were
patent-eligible because the ordered combination of claim limitations
presented an inventive concept). Further analysis is required.
35 U.S.C. § 101—Step 2A, prong two
“The Supreme Court has long distinguished between principles
themselves (which are not patent eligible) and the integration of those
principles into practical applications (which are patent eligible).” MPEP
§ 2106.04(d) (citing Mayo Collaborative Servs. v. Prometheus Labs., Inc.,
Appeal 2020-002642
Application 15/010,141

8

566 U.S. 66, 80, 84 (2012)). This distinction is important because a claim
can recite (i.e., set forth or describe) a judicial exception such as an abstract
idea, yet not be directed to the judicial exception. Id. § 2106.04.III.A.2. It is
not enough that the claim include recitations that merely render the claimed
invention novel. If a judicial exception has not been integrated into a
practical application, adding additional recitations that also fail to integrate
the judicial exception into a practical application do not make the claimed
invention patent-eligible. Synopsys, Inc. v. Mentor Graphics Corp., 839 F.3d
1138, 1151 (Fed. Cir. 2016) (“a claim for a new abstract idea is still an
abstract idea”). But a claimed invention is patent-eligible if it includes an
“inventive concept . . . in the ordered combination of claim limitations that
transform the [judicial exception] into a particular, practical application.”
Bascom, 827 F.3d at 1352. A mental process claim, in particular, can be
integrated into a practical application if it is structured to reflect a specific
implementation that a person engaged in the task being performed would not
have likely used. McRO, Inc. v. Bandai Namco Games Am. Inc., 837 F.3d
1299, 1316 (Fed. Cir. 2016) (quoting Ass’n for Molecular Pathology v.
Myriad Genetics, Inc., 569 U.S. 576, 596 (2013)).
Additionally, in the context of Step 2A, an additional element may
have integrated the exception into a practical application if it “reflects an
improvement in the functioning of a computer, or an improvement to other
technology or technical field.” 2019 Revised Guidance, 84 Fed. Reg. at 55
(citing, e.g., DDR Holdings, LLC. v. Hotels.com, L.P., 773 F.3d 1245, 1258–
59 (Fed. Cir. 2014)). On the other hand, a judicial exception has not been
integrated into a practical application where it “merely includes instructions
to implement an abstract idea on a computer, or merely uses a computer as a
Appeal 2020-002642
Application 15/010,141

9

tool to perform an abstract idea.” Id. (citing Gottschalk v. Benson, 409 U.S.
63 (1972)).
Here the Examiner determines that claim 1, by merely reciting the
additional element of a computer processor, fails to “integrate the abstract
idea into a practical application because it does not impose any meaningful
limits on practicing the abstract idea.” Ans. 6. While the courts do not
“distinguish between claims that recite mental processes performed by
humans and claims that recite mental processes performed on a computer”
(MPEP § 2106.04(a)(2).III), claim 1 includes recitations, other than just “by
a computer processor,” that go beyond those pertaining to abstract mental
processes. E.g., Ans. 4–5. In particular, as Appellant argues, claim 1 recites
“limitations directed to the practical application of iteratively updating a
relevance model until a set of search results is obtained therefrom that is
most similar in relevance ranking to an initial set of search results.” Reply
Br. 7; Appeal Br. 12.
Additionally, we agree with Appellant that claim 1 is “directed to an
improvement to computer functionality” because it is “directed to an
iterative optimization process that produces . . . an optimized relevance
model that is produced without query anchoring but that nonetheless avoids
query drift,” and thereby “represents a software-based improvement to
pseudo-relevance feedback (PRF) computer technology.” Appeal Br. 14–15.
We further agree with Appellant that the claimed iterative optimization
process recites additional improvements to computer functionality by
providing “the capability to account for potential non-uniformity in
document relevance priors and the capability to utilize the feedback inherent
in an initial ranking of a set of top-K documents by attempting to reconstruct
Appeal 2020-002642
Application 15/010,141

10

a re-ranked listing of the top-K documents that matches the initial ranking.”
Id. at 15 (citing Spec. 12).
Appellant’s characterization of claim 1 is supported by recitations that
include:
determining . . . a first relevance model [used to determine a
respective probability of retrieval of each search result in a first
set of search results];
. . .
determining . . . a rank similarity between the first ranking [of
the first set of search results] and the second ranking [of the
first set of search results determined using the respective
probability of retrieval of each search result in the first set of
search results];
determining . . . a second relevance model by updating at least
one model parameter of the first relevance model based at least
in part on the rank similarity;
determining . . . that the second relevance model is an
optimized relevance model for the search query; and
. . .
wherein determining the second set of search results using the
second relevance model causes a rank similarity between a
ranking of the set of common search results in the first set of
search results and a ranking of the set of common search results
in the second set of search results to be maximized and query
drift to be minimized without performing query anchoring of
the search query.
These recitations must be read in light of the Specification, which discloses:
1. Providing relevance model 110 “as input to the document retrieval
probability determination engine 102 . . . where each document
retrieval probability value is an estimate of the respective
probability of retrieval of a corresponding document in the PERL
Appeal 2020-002642
Application 15/010,141

11

[(pseudo effective reference list)] based on the relevance model
110” (Spec. ¶ 26 (cited in Appeal Br. 7));
2. generating “a re-ranking 114 of the top-K documents in
accordance with their respective probabilities of retrieval given the
relevance model 110” (Spec. ¶ 27 (cited in Appeal Br. 7));
3. determining “a rank similarity 118 between the initial ranking 116
of the top-K documents (e.g., in the PERL) and the re-ranked
listing 114 of the top-K documents” (id. ¶ 28 (cited in Appeal Br.
7));
4. updating “one or more model parameters of the relevance model
110 based on the rank similarity 118” (id. ¶ 29 (cited in Appeal Br.
7));
5. determining that “the ranking of the re-ranked list 114 of the top-K
documents is as close as can be achieved to the initial ranking 116
of the top-K documents (e.g., the PERL)” (id. ¶ 30 (cited in Appeal
Br. 7)), thus determining that “the updated relevance model 110
from the current iteration [is] the optimized relevance model 122”
(id. ¶ 31 (cited in Appeal Br. 7));
6. wherein optimized relevance model 122 “avoids query drift
without having to utilize query anchoring by ensuring that the
optimized relevance model results in the same ranking of the
feedback documents (e.g., the initial top-K documents retrieved for
the query) as the initial ranking of the feedback documents” (id.
¶ 16 (cited in Appeal Br. 7)).
The recitations in claim 1 thus encompass, in particular, the last
iteration of the disclosed process of updating relevance model 110 to
Appeal 2020-002642
Application 15/010,141

12

determine optimized relevance model 122. The optimized relevance model
(i.e., the second relevance model) is related to the first relevance model, but
differs because it is determined “by updating at least one model parameter of
the first relevance model.” Moreover, the claimed process maximizes the
similarity of the rankings of the results that are common (i.e., that overlap)
between the first results and the results obtained using the optimized
relevance model. These features thus accord with Appellant’s
characterization of claim 1. Reply Br. 7; Appeal Br. 12. Moreover, the
method of claim 1 is pertinent to an improved pseudo-relevance feedback
information retrieval technique that negates the need to rely on query
anchoring to minimize query drift. Spec. ¶¶ 12–13. Therefore, we agree with
Appellant that the method of claim 1 “reflects an improvement in the
functioning of a computer” and does not merely include “instructions to
implement an abstract idea on a computer” or use “a computer as a tool to
perform an abstract idea.” See 2019 Revised Guidance, 84 Fed. Reg. at 55.
The Examiner determines that all of the determining steps of claim 1
can be practically performed in the human mind. Ans. 4–5. But the
Examiner’s conclusory assertions do not show that the human mind is
equipped to optimize a relevance model in the manner recited. See SRI Int’l,
Inc. v. Cisco Systems, Inc., 930 F.3d 1295, 1304 (Fed. Cir. 2019) (declining
to identify the claimed collection and analysis of network data as abstract
because “the human mind is not equipped to detect suspicious activity by
using network monitors and analyzing network packets as recited by the
claims”).
Even if a person could manually optimize a relevance model, claim 1
is structured to reflect a specific implementation that a person engaged in
Appeal 2020-002642
Application 15/010,141

13

determining search results (i.e., the underlying abstract idea) would have
been unlikely to use. See McRO, 837 F.3d at 1316. That is, while numerous
techniques for manually searching for information are well-known or have
been documented, nothing of record suggests that the claimed relevance
model optimizing process represents an automated implementation of
techniques humans would likely use. Based on the record before us, we are
not persuaded that even with full knowledge of the described technique a
human seeking to achieve similar results would consider it practical to apply
the claimed technique manually, even for small data sets.
The claimed application of the optimized relevance model to
maximize the similarity of the rankings of the results common to first results
and to results obtained with the optimized relevance model further evinces
that claim 1 does not merely claim an improvement to an abstract technique.
SAP Am., Inc. v. InvestPic, LLC, 898 F.3d 1161, 1167–68 (Fed. Cir. 2018).
Moreover, claim 1 focuses “on a specific means or method that improves”
pseudo-relevance feedback models (i.e., that minimizes query drift) rather
than being “directed to a result or effect that itself is the abstract idea and
merely invoke generic processes and machinery.” CardioNet, LLC v.
InfoBionic, Inc, 955 F.3d 1358, 1368 (Fed. Cir. 2020), cert. denied sub nom.
InfoBionic, Inc. v. Cardionet, LLC, 141 S. Ct. 1266 (2021) (quoting McRO,
837 F.3d at 1314).
For these reasons, based on the record before us, we determine that
claim 1 includes additional recitations that integrate the underlying mental
process (or mathematical concepts) into a practical application. Accordingly,
claim 1 is directed to patent-eligible subject matter. Similarly, claims 2–6,
Appeal 2020-002642
Application 15/010,141

14

8–13, and 15–20, which contain similar recitations, are also directed to
patent-eligible subject matter.
CONCLUSION
Claim(s)
Rejected 35 U.S.C. § Reference(s)/Basis Affirmed Reversed
1–6, 8–13,
15–20
101 Eligibility 1–6, 8–13,
15–20
REVERSED

Appeal 2020-002642
Application 15/010,141

15

BAUMEISTER, Administrative Patent Judge, dissenting.
Independent claim 1 is representative of the appealed claims.2
Claim 1 is reproduced below with paragraph designators added for clarity
and emphasis added to the claim language that recites an abstract idea:
1. A method for enhancing robustness of pseudo-relevance
feedback models using query drift minimization, the method
comprising:
[(a)] determining, by a computer processor, a first set of search
results returned for a search query, wherein the first set of search
results is ranked in accordance with a first ranking;
[(b)] determining, by the computer processor, a first relevance
model;
[(c)] determining, by the computer processor and based at least
in part on the first relevance model, a respective probability of
retrieval of each search result in the first set of search results;
[(d)] determining, by the computer processor, a second ranking
for the first set of search results based at least in part on the
respective probability of retrieval of each result in the set of
search results;
[(e)] determining, by the computer processor, a rank similarity
between the first ranking and the second ranking;
[(f)] determining, by the computer processor, a second
relevance model by updating at least one model parameter of the
first relevance model based at least in part on the rank similarity;
[(g)] determining, by the computer processor, that the second
relevance model is an optimized relevance model for the search
query; and
[(h)] determining, by the computer processor, a second set of
search results for the search query using the second relevance
model, wherein the first set of search results and the second set

2 Appellant argues all of the claims together as a group. See Appeal Br. 11–
19. Accordingly, I select independent claim 1 as representative. See
37 C.F.R. § 41.37(c)(1)(iv).
Appeal 2020-002642
Application 15/010,141

16

of search results comprise a set of common search results, and
wherein determining the second set of search results using the
second relevance model causes a rank similarity between a
ranking of the set of common search results in the first set of
search results and a ranking of the set of common search results
in the second set of search results to be maximized and query
drift to be minimized without performing query anchoring of the
search query.

THE REJECTION AND CONTENTIONS
The Examiner determines that claim 1 recites a series of abstract
ideas—specifically, mental processes. Examiner’s Answer 3–5, mailed
Dec. 17, 2019 (“Ans.”). The Examiner determines that the only additional
element that claim 1 recites beyond the abstract ideas is “a processor.” Id.
at 5–6.
The Examiner determines that claim 1’s recitation of the processor
“does not integrate the abstract idea into a practical application because [the
recitation] does not impose any meaningful limits on practicing the abstract
idea.” Id. at 6. The Examiner additional determines that claim 1 does not
include additional elements that are sufficient to amount to significantly
more than the judicial exception because “the additional element of using, a
processor, . . . amounts to no more than mere instructions to apply the
exception using a generic computer component.” Id. The Examiner reasons
that “[m]ere instructions to apply an exception using a generic computer
component cannot provide an inventive concept.” Id.
Appellant’s arguments are addressed in the Analysis section, below.
Appeal 2020-002642
Application 15/010,141

17

ANALYSIS
Step 1
The claimed subject matter falls within the four statutory categories of
patentable subject matter identified by 35 U.S.C. § 101: process, machine,
manufacture, or composition of matter. Accordingly, I turn to step 2A of the
2019 Guidance.
Step 2A, Prong 1
Under step 2A, prong 1, of the 2019 Guidance, the Board first looks to
whether the claim recites any judicial exceptions, including certain
groupings of abstract ideas (i.e., mathematical concepts, certain methods of
organizing human activities such as a fundamental economic practice, or
mental processes). MPEP § 2106.04(a).
Limitation (a) recites, inter alia, “determining . . . a first set of search
results returned for a search query, wherein the first set of search results is
ranked in accordance with a first ranking.” Appellant’s Specification
explains that a pseudo-relevance feedback (PRF) technique may be used to
determine the search results and document rankings, as recited by
limitation (a). Appeal Br. 7 (citing Spec. ¶¶ 22, 25); Appellant’s
Specification further explains, “[p]sedo-relevance feedback (PRF) is an
[information retrieval (IR)] technique that may be used to improve [] the
relevance of documents retrieved in response to a search query in the
absence of explicit user feedback or implicit user feedback.”); Spec. ¶ 3; see
also Spec. ¶ 22 (“the PRF system 100 may identify a set of top-K documents
retrieved for a search query and determine an initial ranking 116 of the top-
K documents.”).
Appeal 2020-002642
Application 15/010,141

18

As such, the determination step of limitation (a) recites a step of
searching, sorting, and ranking documents with the aid of a mathematical
modeling algorithm. See Sharma, “Information Retrieval System Explained:
Types, Comparison & Components,” Mar. 10, 2021 available at
https://www.upgrad.com/blog/information-retrieval-system-explained/ (“An
information retrieval (IR) system is a set of algorithms that facilitate the
relevance of displayed documents to searched queries. In simple words, it
works to sort and rank documents based on the queries of a user.”)
Appellant’s Specification confirms this interpretation:
PRF feedback refers to various techniques that may be
used to improve the relevance of search results returned for a
search query (e.g., through query expansion, re-ranking, fusion,
etc.) in the absence of explicit or implicit relevance feedback. A
number of PRF feedback techniques have been developed over
the years. Most common among these are the Rocchio relevance
model, relevance-based language models, and model-based
feedback. Other known PRF techniques are generally based on
these common PRF techniques and suggest various
enhancements such as, for example, bootstrap sampling of
feedback documents, usage of a cluster language model (LM) for
improved smoothing, utilizing additional lists of feedback
documents, and so forth.
The Rocchio relevance model is an early relevance model
in which a query is modified to include more of those terms that
appear in feedback documents (e.g., the top-K documents
retrieved for the query). Relevance-based language models are
expansion-based (e.g., modifying the query by adding additional
terms) PRF models that attempt to estimate relevance by
assuming that the query and the top-K relevant documents both
relate to the same relevance generative model. The most
commonly used relevance-based language model is the RM3
relevance model that combines query anchoring with the RMI
relevance model. The RMI model may be formally defined as:
Appeal 2020-002642
Application 15/010,141

19

P(w, q1, ... , qk) = ∑ 𝑃𝑃(𝑀𝑀)𝑃𝑃(𝑤𝑤|𝑀𝑀) 𝑀𝑀∈ℳ ∏ 𝑃𝑃𝑘𝑘𝑖𝑖=1 (q|M). Query
anchoring may be formally defined as: θQ' = (1-α)θQ + αθF,
where Q' represents a modified query, Q represents the original
query, and F represents the feedback. The smaller the value of
α, the more similar the modified query is to the original query.
Spec. ¶¶ 12–13.
This claimed modeling procedure, then, entails a mathematical
concept including mathematical relationships, mathematical formulas or
equations, and mathematical calculations. The 2019 Guidance expressly
recognizes these mathematical concepts as constituting a patent-ineligible
abstract idea. MPEP § 2106.04(a).
Sorting and ranking searched documents also entails performing a
mental process, such as an observation, evaluation, judgment, or opinion that
can be performed either in the human mind or with the aid of paper and
pencil. For example, long before the advent of the Internet or even
computers, librarians, newspapers’ archive researchers, and the like received
requests for documents pertaining to a specified topic. And in response to
receiving such queries, the researchers would determine keywords and
search strategies based on the requested topic, synonyms, and related
information. These researchers commonly sorted and ranked the documents
obtained though their searches: e.g., they made recommendations to the
document requesters as to which documents likely would be the most
relevant or useful. The 2019 Guidance expressly recognizes such mental
processes as constituting patent-ineligible abstract ideas. MPEP
§ 2106.04(a).
Because limitation (a) recites both mathematical concepts and mental
processes, limitation (a) recites a patent-ineligible abstract idea.
Appeal 2020-002642
Application 15/010,141

20

Similar to limitation (a), each one of limitations (b) though (h) also
recite steps of an improved modeling method, which steps respectively entail
various combinations of searching, sorting, and ranking queried documents:
(b) determining . . . a first relevance model;
(c) determining, . . . based at least in part on the first relevance
model, a respective probability of retrieval of each search result
in the first set of search results;
(d) determining . . . a second ranking for the first set of search
results based at least in part on the respective probability of
retrieval of each result in the set of search results;
(e) determining . . . a rank similarity between the first ranking
and the second ranking;
(f) determining . . . a second relevance model by updating at
least one model parameter of the first relevance model based at
least in part on the rank similarity;
(g) determining . . . that the second relevance model is an
optimized relevance model for the search query; and
(h) determining . . . a second set of search results for the
search query using the second relevance model, wherein the first
set of search results and the second set of search results comprise
a set of common search results, and wherein determining the
second set of search results using the second relevance model
causes a rank similarity between a ranking of the set of common
search results in the first set of search results and a ranking of the
set of common search results in the second set of search results
to be maximized and query drift to be minimized without
performing query anchoring of the search query.
For the reasons set forth in relation to limitation (a), each one of
limitations (b) though (h), likewise, recites mathematical concepts and
mental processes that the 2019 Guidance expressly recognizes as
constituting abstract ideas. MPEP § 2106.04(a).
Appeal 2020-002642
Application 15/010,141

21

Appellant argues that none of the claimed determining steps can be
practically performed in the human mind or with the aid of paper and pencil.
Reply Br. 5. Specifically, Appellant argues,
the claims require the utilization of an information retrieval
system and serves to improve the relevance of search results
returned for a search query in the absence of explicit or implicit
relevance feedback. As described in at least paragraph 12 of
Applicant’s originally filed specification, the claimed optimized
relevance model improves over conventional platforms by
maximizing a rank similarity between a set of search results and
an initial set of search results in a manner that minimizes query
drift without performing query anchoring.
Id.
This argument is unpersuasive because, to the extent that Appellant
appears to be arguing that an information retrieval system must inherently be
computerized, Appellant does not reasonably explain which limitations of
claim 1 purportedly recite an automation component. Claim 1 only recites
that the recited search-and-ranking algorithm be carried out in response to “a
search query.” That is, aside from the additional requirement that the
claimed method be carried out by a computer processor, no language of
claim 1 indicates that the recited modeling algorithm necessarily must be
automated. Although the claims are interpreted in light of the specification,
limitations from the specification are not read into the claims. In re Van
Geuns, 988 F.2d 1181, 1184 (Fed. Cir. 1993).
Appellant further argues that the recited steps cannot be characterized
as steps of a mental process because the steps allegedly cannot be performed
in the human mind. Reply Br. 5. Appellant restates limitation (h) in support
of this assertion:
Appeal 2020-002642
Application 15/010,141

22

determining, by the computer processor, a second set of search
results for the search query using the second relevance model,
wherein the first set of search results and the second set of search
results comprise a set of common search results, and wherein
determining the second set of search results using the second
relevance model causes a rank similarity between a ranking of
the set of common search results in the first set of search results
and a ranking of the set of common search results in the second
set of search results to be maximized and query drift to be
minimized without performing query anchoring of the search
query.
Reply Br. 6 (citing claim 1, limitation (h)). Appellant then restates the
initially asserted conclusion: “Search queries, updating parameters in
relevancy models, and ranking search results require the utilization of
computer technology to access search databases for search results and
cannot be practically performed in the human mind.” Id.
This argument is unpersuasive for two reasons. First, the Examiner
did not take the position that the recited abstract idea includes performing
the improved algorithm for searching, sorting, and ranking documents
specifically with the aid of the computer processor. Rather, the Examiner
acknowledges that the claimed computer processor constitutes an additional
element beyond the abstract idea. Ans. 5. As such, Appellant’s arguments
are not commensurate with the Examiner’s position.
Second, Appellant provides no persuasive evidence that a human
would be incapable of performing the recited steps solely in the human mind
or with the aid of paper and pencil, at least for those situations covered by
the claim that entail small collections of returned search results. See
generally Reply Br. And in fact, I see no language in claim 1 that would
preclude the claimed method from being performed on a small set of
candidate documents—say five, ten or twenty documents. In such a case,
Appeal 2020-002642
Application 15/010,141

23

the process could be carried out readily in the human mind, perhaps aided by
pencil and paper.
For these reasons, I agree with the Examiner that each of
limitations (a) through (h) recites a judicial exception to patent-eligible
subject matter under step 2A, prong 1, of the 2019 Guidance. See
RecogniCorp, LLC v. Nintendo Co., 855 F.3d 1322, 1327 (Fed. Cir. 2017)
(“Adding one abstract idea . . . to another abstract idea . . . does not render
the claim non-abstract.”)
Step 2A, Prong 2
Under step 2A, prong 2, of the 2019 Guidance, I next analyze whether
claim 1 recites additional elements that individually or in combination
integrate the judicial exception into a practical application. 2019 Guidance,
84 Fed. Reg. at 53–55. The 2019 Guidance identifies considerations
indicative of whether an additional element or combination of elements
integrate the judicial exception into a practical application, such as an
additional element reflecting an improvement in the functioning of a
computer or an improvement to other technology or technical field. Id. at
55; MPEP § 2106.05(a).
As noted above, the Examiner determines that the recited “computer
processor” is the only additional element recited in claim 1. Ans. 6. And
the Examiner determines that the recitation of the computer processor
“amounts to no more than mere instructions to apply the exception [(the
abstract idea)] using a generic computer component.” Ans. 6.
I agree with the Examiner that the label “computer processor,” as
recited, merely describes a generic computer component. And I agree with
the Examiner that the term’s recitation within claim 1 amounts to mere
Appeal 2020-002642
Application 15/010,141

24

instructions to implement the abstract idea on a computer. Therefore, the
recitation of “a computer processor” is not sufficient to make the claim
patent eligible. See Alice, 573 U.S. at 226 (determining that the claim
limitations “data processing system,” “communications controller,” and
“data storage unit” were generic computer components that amounted to
mere instructions to implement the abstract idea on a computer);
October 2019 Guidance Update at 11–12 (recitation of generic computer
limitations for implementing the abstract idea “would not be sufficient to
demonstrate integration of a judicial exception into a practical application”).
Appellant argues that assuming arguendo that claim 1 recites a
judicial exception, claim 1 integrates the exception into a practical
application and imposes meaningful limitations on the exception:
Specifically, the claims recite limitations directed to the practical
application of iteratively updating a relevance model until a set
of search results is obtained therefrom that is most similar in
relevance ranking to an initial set of search results. This is clearly
a practical application having the practical benefit of obtaining
the most relevant search results for a query in a manner that
minimizes query drift without having to perform query
anchoring.
Reply Br. 7.
This argument is unpersuasive because iteratively updating a
relevance model until an improved set of search results is obtained entails
the underlying abstract idea of sorting and ranking documents. See BSG
Tech LLC v. Buyseasons, Inc., 899 F.3d 1281, 1290 (Fed. Cir. 2018) (“It has
been clear since Alice that a claimed invention’s use of the ineligible concept
to which it is directed cannot supply the inventive concept that renders the
invention ‘significantly more’ than that ineligible concept.”); Synopsys, Inc.
v. Mentor Graphics Corp., 839 F.3d 1138, 1151 (Fed. Cir. 2016) (“[A] claim
Appeal 2020-002642
Application 15/010,141

25

for a new abstract idea is still an abstract idea.”) (emphasis omitted); SAP
Am., Inc. v. InvestPic, LLC, 898 F.3d 1161, 1168 (Fed. Cir. 2018) (“What is
needed is an inventive concept in the non-abstract application realm.”).
For these reasons, Appellant does not persuade me that claim 1 is
directed to an improvement in the function of a computer or to any other
technology or technical field. MPEP § 2106.05(a). Nor does Appellant
persuasively demonstrate that claim 1 is directed to a particular machine or
transformation, or that claim 1 adds any other meaningful limitations for the
purposes of the analysis under Section 101. MPEP §§ 2106.05(b), (c), (e).
Accordingly, Appellant does not persuade me that claim 1 integrates the
recited abstract ideas into a practical application within the meaning of the
2019 Guidance.
Step 2B
Under step 2B of the 2019 Guidance, I next analyze whether claim 1
adds any specific limitations beyond the judicial exception that, either alone
or as an ordered combination, amount to more than “well-understood,
routine, conventional” activity in the field. 2019 Guidance, 84 Fed. Reg. at
56; MPEP § 2106.05(d).
Appellant’s Specification indicates that the recited additional element
—the computer processor—was well understood, routine, and conventional:
These computer readable program instructions may be
provided to a processor of a general purpose computer, special
purpose computer, or other programmable data processing
apparatus to produce a machine, such that the instructions, which
execute via the processor of the computer or other programmable
data processing apparatus, create means for implementing the
functions/acts specified in the flowchart and/or block diagram
block or blocks. These computer readable program instructions
may also be stored in a computer readable storage medium that
Appeal 2020-002642
Application 15/010,141

26

can direct a computer, a programmable data processing
apparatus, and/or other devices to function in a particular
manner, such that the computer readable storage medium having
instructions stored therein comprises an article of manufacture
including instructions which implement aspects of the
function/act specified in the flowchart and/or block diagram
block or blocks.
Spec. ¶ 60 (emphasis added).
Furthermore, Appellant’s Specification does not indicate that
consideration of the recitations of this conventional computer processor as
an ordered combination adds any significance beyond this additional
element, as considered individually. Rather, Appellant’s Specification
indicates that the invention is directed to an abstract idea that is made more
efficient with generic computer processor—improving a relevance model.
E.g., Spec. ¶ 4.
For these reasons, claim 1 does not recite additional elements that,
either individually or as an ordered combination, amount to significantly
more than the judicial exception within the meaning of the 2019 Guidance.
2019 Guidance, 84 Fed. Reg. at 52–55; MPEP § 2106.05(d).
Accordingly, I would affirm the Examiner’s rejection of claim 1 under
35 U.S.C. § 101 as being directed to an exception to patent-eligible subject
matter without reciting significantly more. I, likewise, would affirm the 101
rejection of claims 2–6, 8–13, and 15–20, which Appellant does not argue
separately. Reply Br. 4–7.