Ex Parte 6778979 et al

Patent Trial and Appeal Board

Oct 26, 2015

95001430 (P.T.A.B. Oct. 26, 2015)

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.
95/001,430 09/08/2010 6778979 3454REX.747BS 1480
86636 7590 10/26/2015
BRUNDIDGE & STANGER, P.C.
2318 MILL ROAD, SUITE 1020
ALEXANDRIA, VA 22314
EXAMINER
WOOD, WILLIAM H
ART UNIT PAPER NUMBER
3992
MAIL DATE DELIVERY MODE
10/26/2015 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
____________

GOOGLE, INC.
Requester and Respondent

v.

XEROX CORPORATION
Patent Owner and Appellant
____________

Appeal 2014-000031
Reexamination Control 95/001,430
Patent 6,778,979 B2
Technology Center 3900
____________

Before RICHARD M. LEBOVITZ, DAVID M. KOHUT, and
ANDREW J. DILLON, Administrative Patent Judges.

LEBOVITZ, Administrative Patent Judge.

DECISION ON REQUEST FOR REHEARING
Requester filed a Request for Rehearing (“Req. Reh’g”) under 37 C.F.R.
§ 41.79 of the Decision by a panel of the Patent Trial and Appeal Board (“the
panel”) mailed February 3, 2014 (“Decision”) in the above-identified inter partes
reexamination of U.S. Patent No. 6,778,979 B2 (“the ’979 patent”). In the
Decision, the panel affirmed the Examiner’s final rejection of claim 1–20. The
claims stand rejected over fourteen grounds of rejection. Decision 2. Although all
20 claims have been rejected under the different rejections, Patent Owner only
Appeal 2014-000031
Reexamination Control 95/001,430
Patent 6,778,979 B2

2

argued independent claims 1, 14, and 18 for six of the rejections, and argued in
detail only with respect to claim 1. Id.
Patent Owner contends that the panel’s Decision misapprehended or
overlooked issues in three rejections: 1) anticipation by Wieser under 35 U.S.C.
§ 102; 2) obviousness in view of Black and HyPursuit under 35 U.S.C. § 103; and
3) obviousness in view of Black and Donaldson under 35 U.S.C. § 103. Req.
Reh’g 2–3.
We did not reach the rejection based on Black and Donaldson because all the
claims rejected under it were already rejected over Black and HyPursuit.
CLAIM 11
1. A method for automatically generating a query from selected
document content, comprising:
[1] defining an organized classification of document content
with each class in the organized classification of document content
having associated therewith a classification label;
each classification label corresponding to a category of
information in an information retrieval system;
[2] automatically identifying a set of entities in the selected
document content for searching additional information related thereto
using the information retrieval system;
[3] automatically categorizing the selected document content
using the organized classification of document content for assigning
the selected document content a classification label from the
organized classification of content; and
[4] automatically formulating the query to restrict a search at
the information retrieval system for information concerning the set of
entities to the category of information in the information retrieval
system identified by the assigned classification label.

1 Numbered steps [1] to [4] have been added for reference.

Appeal 2014-000031
Reexamination Control 95/001,430
Patent 6,778,979 B2

3

WIESER
Patent Owner contends that the panel erred in finding “the product clusters
of Wieser to be labels within the scope of the claims, and therefore inherently
found them to be labels in an organized classification of document content.
(Decision, p. 12).” Req. Reh’g 2. Specifically, Patent Owner argues that in a
classification system, “the system’s categories (i.e., classes) exist independent of
any data to be categorized, and therefore enable classification of data into the
existing categories/classes of the system.” Id. at 3. Patent Owner states that in
Wieser’s system the products change in size, number and/or scope as more data is
added, and thus, are not “classification labels” as that term is used in the claims.
Id. at 3–4.
On page 10 of the Decision, Wieser’s description of “clusters” of products in
a database was determined to serve as “classification labels,” as recited in claim 1.
The following disclosure from Weiser was cited for this description:
After the vector generations, the document vector, or feature
vector, must be compared with database vectors, or feature vectors. A
naïve approach to product matching would be to compare the
document vector to the vectors for every product in the database. This
process becomes more burdensome as the number of products in the
database grows. Fortunately, the database vectors are not smoothly
distributed throughout the vector space, but rather, tend to “clump”
together, leaving vast empty spaces between the clumps, or clusters as
they are commonly known. . . . Then, when one wishes to find the
products that match a given document vector, one need only compare
the document to the products in the cluster whose bounding sphere
contains the document vector (or the nearest spheres, if no sphere
contains the document vector).
(Wieser, p. 17, 15-28; emphases added.)
As explained in the quoted passage, the “clusters” are the product
“classification labels.” The clusters, or classification labels, are determined (step
Appeal 2014-000031
Reexamination Control 95/001,430
Patent 6,778,979 B2

4

[1] of claim 1) and then compared to the document vector content. Because
Wieser teaches that the product classification labels are “match[ed]” to the
document vectors, at least some of the document vectors must bear a classification
label as in step [3] of the claim.
Patent Owner attempts to rebut this disclosure in Wieser with testimony by
Dr. V.S. Subrahmanian (Declaration dated June 7, 2011). Req. Reh’g 3. Dr.
Subrahmanian testified:
. . . the set of vectors for a particular set of documents “are not
smoothly distributed throughout the vector space, but rather, tend to
‘clump’ together, leaving vast empty spaces between the clumps, or
clusters as they are commonly known” ([Weiser] at 17, lines 19-21)
Subrahmanian Decl. ¶ 35 (emphasis added).
In fact, as quoted above from Weiser, the clumped “vectors” were a
reference to “database vectors” not document vectors:
Fortunately, the database vectors are not smoothly distributed
throughout the vector space, but rather, tend to “clump” together,
leaving vast empty spaces between the clumps, or clusters as they are
commonly known.
Wieser, p. 17, ll. 16–21 (emphases added).
However, Dr. Subrahmanian characterized the clumped vectors as vectors
“for a particular set of documents.” Because Dr. Subrahmanian’s opinion contains
this factual error, his further opinion that Wieser does not describe “classification
labels” is given little weight because it based on an apparent misreading of Wieser.
In paragraph 43–44 of Dr. Subrahmanian’s declaration, he states that “data
or documents are grouped together on an ad hoc basis using mathematical
calculations involving the spheres” and that therefore “the definition of cluster
spheres is mathematical,” making it “not possible to determine the subject matter
Appeal 2014-000031
Reexamination Control 95/001,430
Patent 6,778,979 B2

5

of the data or documents whose vectors fall within the boundaries of a given
cluster sphere without actually examining those data or documents.”
This testimony is not supported by a preponderance of the evidence. Wieser
expressly refers to products in the description of clustering. Wieser, p. 17, ll.
15-28. Dr. Subrahmanian does not explain how this explicit reference to a
“product,” as well as Wieser’s characterization of its process of matching product
offers to documents a customer is reading (Decision 10) constitutes an
incomprehensible cluster sphere. Thus, we find Patent Owner’s argument (Req.
Reh’g 6–7) that that subject matter in a cluster described in only mathematical
terms to be unsupported by the evidence and contrary to Wieser who expressly
describes product categories. See, e.g., Wieser at p. 6, ll. 4–9 describing golf.
Decision 10.
We agree with Patent Owner’s statement that the claimed system requires a
classification system that would indicate understandable labels indicative of the
subject matter in a document or database. Req. Reh’g. 7. To the extent the
Decision states or implies otherwise, we withdraw that reasoning and do not rely
on it.
The question is whether Wieser describes classification labels, which the
evidence answers in the affirmative. Specifically, as quoted in the Decision,
Wieser teaches:
For example, if a visitor or customer of client 12 selects an article
about BMW’s newest vehicle, the match server 14 of the present
invention presents a buying opportunity to the customer, i.e., an
attractively discounted book for “BMW Enthusiasts”. If the customer
is reading an article about Tiger Woods, the match server 14 displays
or presents items such as Nike golf balls or a book on the history of
the US Open to the customer. It is appreciated that the intelligence
engine 14 of the present invention understands the difference between
Appeal 2014-000031
Reexamination Control 95/001,430
Patent 6,778,979 B2

6

Tiger Woods, a tiger, and woods.
Wieser, p. 6, ll. 4-11.
Clearly, Wieser’s classifications labels constitute real-world meaning
in order that they can match a document about Tiger Woods to a product
offer about golf. Dr. Subrahmanian’s opinion to the contrary is unsupported
by evidence, fails to address the entire disclosure in Wieser, and is based on
an apparent misreading of Wieser.
On page 8 of the Request for Rehearing, Patent Owner, in a footnote,
refers to an alleged mis-description of the functionality of Wieser relied
upon in the Decision. Patent Owner’s Request does not identify what step of
the claimed method they are attempting to distinguish over Wieser.
However, referring back to the Decision, it appears the Patent Owner intends
to address Step [4] of the claimed method. We quote step [4] below:
[4] automatically formulating the query to restrict a search at
the information retrieval system for information concerning the set of
entities to the category of information in the information retrieval
system identified by the assigned classification label.
Decision 3.
In the Decision, the panel found that Wieser’s description of
automatically forming an SQL query to search cluster IDs corresponded to
step [4] of claim 1. Decision 11 (“Step [4]”). Patent Owner disputes this
finding. Req. Reh’g. 8–9. Patent Owner states:
By “always” matching multiple clusters to an input document vector
(including clusters not containing the vector), Wieser’s system is not
restricted to “the category,” and thus Wieser's disclosure does not
meet the claimed automatically formulating the query limitation of the
claims.
Appeal 2014-000031
Reexamination Control 95/001,430
Patent 6,778,979 B2

7

Id. at 9.
In Step [4] of claim 1, the query is formulated to “restrict” the search
relating to the “set of entities” to the category of information identified by the
“classification label.” The term “entity” is defined in the ‘979 patent to mean
“something recognized in a document . . . that can be in the form of an image, text,
embedded data, HTML, etc.” (id. at col. 6, l. 65 to col. 7, l. 2). Decision 4. In
other words, in simple terms, if “golf ball” is identified in the document classified
as “golf”, then a query is formed to search the category of information classified as
“golf.” Patent Owner attempts to distinguish Wieser on the basis that Wieser
allegedly always searches multiple clusters, not a single one as required by step [4]
of claim 1.
The following disclosure from Wieser is pertinent to this issue:
Then, when one wishes to find the products that match a given
document vector, one need only compare the document to the
products in the cluster whose bounding sphere contains the document
vector (or the nearest spheres, if no sphere contains the document
vector). Thus, the computation is reduced from comparing the
document vector to all the product vectors to simply comparing the
document vector to the center vectors for the spheres, followed by
comparison to the product vectors for the products in the matching
spheres.
Wieser, p. 17, ll. 24–31 (emphasis added).
In accordance with an embodiment of the present invention, the query
context vector, is compared to the center vector of each cluster and the
clusters with center vectors closest to the query context vector are
selected. It is appreciated that these clusters IDs are then used to
narrow the scope of products returned by the original metadata query
as issued by the client 12. In other words, this narrowing qualification
is added to the original metadata query to form an SQL query.
Appeal 2014-000031
Reexamination Control 95/001,430
Patent 6,778,979 B2

8

Id. at p. 18, ll. 16–23 (emphasis added).
From this disclosure, it appears that Wieser identifies “clusters” which
match the document vector, and then uses this information to form a further SQL
query of the clusters, which is done automatically. However, this disclosure in
Wieser does not exclude circumstances where only one cluster is identified.
Wieser’s method is simply designed to be inclusive so as not miss contextually-
related clusters, but does not exclude circumstances in which only one cluster is
identified. This possibility appears to be disclosed in Wieser where it refers to “the
step of comparing compares said feature vector to center vectors to determine at
least one contextually related cluster.” Wieser, p. 30, ll. 24–26.
For the reasons discussed above, we are not persuaded that the panel erred in
affirming the rejection of claims 1, 5, 6, 10–14, and under 35 U.S.C. § 102 as
anticipated by Wieser.

BLACK AND HYPURSUIT
In the obviousness rejection based on Black and HyPursuit, Black was cited
for disclosing or suggesting step [3] of claim 1. The following is excerpted from
the Decision:
Step [3] of automatically categorizing the selected document
content by assigning classification label is said by Examiner and
Requester to be met by Black’s description of a hook (Request 104).
Black teaches:
The hook is the concept, primary subject matter or main topic
for a body of text. The hook is used to define a query as
narrowly as possible on a particular topic for a selected
information source. To determine a “hook”, a content layer
must exist for which a context can be determined. There must
be a perceivable structure to the information source and each
Appeal 2014-000031
Reexamination Control 95/001,430
Patent 6,778,979 B2

9

content entry must have an associated context or place or places
within the structure of the information source.
(Black, col. 2, ll. 10-18.)
In one of the examples, Black describes automatically
generating a hook by extracting proper names from a document and
then, based on frequencies and other context, identifying a proper
name in the document to use as hook (Black, col. 2, l. 57 to col. 3, l.
24). The classification label assigned to the document in this case is
the proper name.
Decision 17.
Patent Owner contends that “there is no suggestion in Black of identifying a
hook based on a classification regime outside of the document alone. The entire
discussion of Black addressing generation of the hook is based solely on
examination of the document itself.” Req. Reh’g 11.
This argument was addressed on page 19 of the Decision, including the
suggestion to generate a hook based on predefined categories. The suggestion to
use predefined categories does not have to come from Black alone. Req. Reh’g 12.
The rejection is based on Black and HyPursuit, and, as explained in the Decision,
Black’s “disclosure reasonably suggests using HyPursuit’s method of clustering
content for the purpose of improving searching.” Decision 19. Thus, it would
have been obvious to one of ordinary skill in the art to have utilized HyPursuit’s
predefined categories in Black’s method. Id.
In regard to the statement in the Decision cited in the Request for Rehearing
that “Black meets the claimed requirement of ‘automatically formulating’ a search
query,” it clear that the Decision was referring to the automation aspect of the
claims, and not the entire limitation. Decision 20–21.
Appeal 2014-000031
Reexamination Control 95/001,430
Patent 6,778,979 B2

10

Patent Owner also contends that the Requester erred by arguing that Black’s
query would be classified by HyPursuit, which is not what is required by the claim.
Req. Reh’g 12.
The disputed limitation is step [4] of claim 1 which recites “automatically
formulating the query to restrict a search at the information retrieval system for
information concerning the set of entities to the category of information in the
information retrieval system identified by the assigned classification label.”
The Decision very expressly identified support in HyPursuit for step
[4], which we repeat here:
HyPursuit describes automatically categorizing a document
based on its content:
The abstraction function for query routing, on the other hand,
computes a manageable set of terms that are used for
identifying portions of the information space relevant to
particular queries. The abstraction function uses term and term
frequency information in the children’s content labels to
compute term weights. The abstraction function then selects
the most heavily weighted terms for generating the content
router’s content label.
(HyPursuit, p. 185, col. 1; emphasis in original.)
HyPursuit also describes restricting the search based on the
assigned labels:
HyPursuit uses query routing to support the search operations.
Query routing uses the content labels stored in the content
router to determine which of the child servers are likely to
contain documents related to the user query.
(HyPursuit, p. 186, col. 2.)

Decision 18.
Based on this disclosure, the Decision states that it would have been
obvious to implement these strategies in Black to enhance “retrieval of only
Appeal 2014-000031
Reexamination Control 95/001,430
Patent 6,778,979 B2

11

relevant information” (Black, p. 180, col. 2) and “specificity of the
information need” (id. at p. 181, col. 1). Decision 20.
Patent Owner has not adequately addressed this explicit disclosure
and reasoning in the Decision. For example, on pages 11–13 of the Request
for Rehearing, Patent Owner focuses on Black without addressing the
specific HyPursuit disclosure cited in the Decision. Consequently, we find
Patent Owner’s arguments unsupported by the evidence of record.
Patent Owner also argues that “there is no teaching or suggestion of
how the query of Black could generated with a hook identified in accordance
with the classification system of HyPursuit.” Req. Reh’g 14. As already
discussed, Black teaches automation. Dec. 17–19. Patent Owner has not
adequately explained or provided evidence to the contrary that Black would
have enabled the skilled worker to implement HyPursuit’s query routing
procedure.
REHEARING DENIED

Appeal 2014-000031
Reexamination Control 95/001,430
Patent 6,778,979 B2

12

Patent Owner:

BRUNDIDGE & STANGER, P.C.
2318 Mill Road, Suite 1020
Alexandria, VA 22314

Third Party Requester:

QUINN EMANUAL URQUHART
& SULLIVAN, LLP
50 California Street, 22nd Floor
San Francisco, CA 94111