Ex Parte Mason

Board of Patent Appeals and Interferences

Sep 2, 2010

10684313 (B.P.A.I. Sep. 2, 2010)

UNITED STATES PATENT AND TRADEMARK OFFICE
____________________

BEFORE THE BOARD OF PATENT APPEALS
AND INTERFERENCES
____________________

Ex parte ZACHARY J. MASON
____________________

Appeal 2009-002829
Application 10/684,313
Technology Center 2100
____________________

Before JOHN A. JEFFERY, ST. JOHN COURTENAY III, and THU A.
DANG, Administrative Patent Judges.

DANG, Administrative Patent Judge.

DECISION ON APPEAL1

1 The two-month time period for filing an appeal or commencing a civil
action, as recited in 37 C.F.R. § 1.304, or for filing a request for rehearing,
as recited in 37 C.F.R. § 41.52, begins to run from the “MAIL DATE”
(paper delivery mode) or the “NOTIFICATION DATE” (electronic delivery
mode) shown on the PTOL-90A cover letter attached to this decision.
Appeal 2009-002829
Application 10/684,313

I. STATEMENT OF THE CASE
Appellant appeals from the Examiner’s final rejection of claims 1-16
under 35 U.S.C. § 134(a) (2002). We have jurisdiction under 35 U.S.C.
§ 6(b).
We reverse.

A. INVENTION
According to Appellant, the invention relates to analyzing browse
activity data of users of a database access system, and more specifically, to
the analysis of item selection histories of users of a database access system
to predict category preferences or affinities to such users (Spec. 1, ¶ [0001]).

B. ILLUSTRATIVE CLAIMS
Claims 1 and 11 are exemplary and are reproduced below:
1. A computer-implemented method of analyzing browse
activity data of users of a database access system, the method
comprising:

providing a browse tree in which items represented
within a database are arranged within item categories over
multiple levels of item categories;

assigning individual user history scores to specific
categories of the browse tree based at least in-part on an item
selection history of a user, wherein the individual user history
scores represent the user's predicted affinities for the
corresponding item categories;

assigning collective user history scores to specific
categories of the browse tree based at least in-part on item
selection histories of a population of users, wherein the
collective user history scores represent the predicted affinities
2
Appeal 2009-002829
Application 10/684,313

of the user population for the corresponding item categories;
and

evaluating differences between the individual user
history scores and the collective user history scores to generate
a relative preference profile for the user, wherein the relative
preference profile comprises relative preference scores for
specific item categories, said relative preference scores
reflecting a degree to which the user's predicted affinity for a
category differs from the predicted affinity of the user
population for that category.

11. A method of distributing credit for a selection event
among the nodes of a browse tree, the method comprising:

determining a total amount of credit to be distributed for
the selection event in which a user selected an item within the
browse tree;

identifying each ancestor node of the selected item within
the browse tree;

dividing said total amount of credit by the number of
ancestor nodes of the selected item to determine an amount of
credit per ancestor to be distributed for the selection event; and

assigning said amount of credit per ancestor to the
ancestor nodes of the selected item within the browse tree.

C. REJECTIONS
The prior art relied upon by the Examiner in rejecting the claims on
appeal is:
Herz US 6,460,036 B2 Oct. 1, 2002
Ortega US 6,606,619 B2 Aug. 12, 2003
Ford US 6,963,867 B2 Nov. 8, 2005
(filed on Mar. 31, 2003)
3
Appeal 2009-002829
Application 10/684,313

Claims 11-14 stand rejected under 35 U.S.C. § 102(e) as anticipated
by the teachings of Ford.
Claims 1-10 and 15-16 stand rejected under 35 U.S.C. § 103(a) over
the teachings of Ortega in view of Herz.
Claims 2-3 and 16 stand rejected under 35 U.S.C. § 103(a) over the
teachings of Ortega in view of Herz and Ford.

II. ISSUES
1) Has the Examiner erred in finding that Ford teaches “dividing said
total amount of credit by the number of ancestor nodes” (claim 11), as
Appellant contends?
2) Has the Examiner erred in finding that Ortega and Herz teach or
suggest “assigning individual user history scores to specific categories of the
browse tree based at least in-part on an item selection history” and
“evaluating differences between the individual user history scores and the
collective user history scores to generate a relative preference profile for the
user” (claim 1), as Appellant contends?

III. FINDINGS OF FACT
The following Findings of Fact (FF) are shown by a preponderance of
the evidence.
Ford
1. Ford discloses a category ranking process used to generate a category
relevancy ranking for each competing category in an All Products
search (col. 21, ll. 4-6; Fig. 8), wherein a query server determines a
4
Appeal 2009-002829
Application 10/684,313

category popularity score indicative of the significance of the query
term to the category (col. 21, ll. 58-60).
2. In an embodiment, the category popularity score is determined by
taking the mean value of the constituent top-level result item
popularity scores (col. 22, ll. 50-52), wherein the mean value is
determined by diving the sum of popularity scores for the constituent
search result items in each category by the total number of constituent
search result items in each category (col. 22, ll. 52-57; Table IV).
Herz
3. Herz discloses automatically computing the likelihood of interest in a
particular target object for a specific user (col. 18, ll. 49-51; Fig. 12),
wherein the computing process includes estimating the intrinsic
quality measure for a target object (col. 18, ll. 55-58), and computing
the topical interest that users, like the specific user, generally have in
the object (col. 19, ll. 18-22).
4. Estimating the intrinsic quality measure of the object includes
selecting the attributes of the object which may include the object’s
popularity among users in general (col. 18, ll. 65-67), multiplying the
selected attributes by a stored weight indicative of the specific user’s
preference for the object (col. 19, ll. 6-11), and computing the
weighted sum of the selected attributes (col. 19, ll. 11-13).
5. Estimating topical interest at all points includes interpolating among
estimates of topical interest at selected points (col. 19, ll. 34-42) by
determining the similarity distance between (U,X) as an extended
object that bears all the attributes of target X and all the attributes of
user U and (V,Y) as an extended object that bears all the attributes of
5
Appeal 2009-002829
Application 10/684,313

target Y and all the attributes of user V, wherein the user U’s interest
can be estimated even if user U is a new user or has never provided
feedback, because the relevance feedback of users whose attributes
are similar to U’s attributes is taken into account (col. 20, ll. 51-55).

IV. ANALYSIS
35 U.S.C. § 102(e)
Claims 11-14
Appellant contends that “Ford does not disclose ‘identifying each
ancestor node of the selected item within the browse tree’” (App. Br. 5). In
particular, Appellant contends that the Office Action’s reading of the
“categories” as the “ancestor” is “improper” because “no credit from a
selection event is ever allocated to the ‘categories’” and “[c]laim 1[sic]
relates to ancestors of an item associated with a selection event, and not with
ancestors of a ‘group of related items’” (App. Br. 7).
Furthermore, Appellant contends that “Ford does not disclose
‘dividing said total amount of credit by the number of ancestor nodes of the
selected item’” (App. Br. 6). That is, Appellant argues that, in Ford, “the
divisor is a number of different items (not a number of ancestor nodes of an
item associated with a selection event)” (id.).
However, the Examiner finds that “Ford teaches receiving a search
query, interpreted as a selection event” (Ans. 12), and “the top level
categories are interpreted to be ancestor nodes from which items belong to”
(Ans. 12-13). In particular, the Examiner finds that “the limitation
[‘identifying each ancestor node of the selected item within the browse tree’]
is interpreted by the examiner as identifying the groups from which a
6
Appeal 2009-002829
Application 10/684,313

selected item belongs to” (id.). The Examiner further finds that, in Ford,
“the total score within a category is added up and divided by the number of
items, which provides a score for a category” (Ans. 13).
To determine whether Ford teaches “dividing said total amount of
credit by the number of ancestor nodes,” as recited in claim 11, we begin our
analysis by giving the claims their broadest reasonable interpretation. See
In re Bigio, 381 F.3d 1320, 1324 (Fed. Cir. 2004). Since claim 11 does not
place any limitation on what “ancestor node” means, includes, or represents,
other than reciting “dividing said total amount of credit by the number of
ancestor nodes of the selected item to determine an amount of credit per
ancestor,” we interpret “ancestor node” as meaning any node at a higher
level than a particular node.
Ford teaches a search process which includes determining the
significance of a query term to a category (FF 1), wherein a popularity score
is determined for the constituent search result items in each category (FF 2).
We find Ford to disclose an ancestor node for the particular search result
items since the category comprising the particular search result items is a
node at a higher level than the search result items.
Though Appellant contends that “no credit from a selection event is
ever allocated to the ‘categories’” and (App. Br. 7), we disagree. As the
Examiner finds, “Ford teaches receiving a search query, interpreted as a
selection event” (Ans. 12). Further, Ford discloses the determining a
category popularity score (FF 2). Thus, contrary to Appellant’s contention,
credit (category popularity score) for the selection event is allocated to the
categories.
7
Appeal 2009-002829
Application 10/684,313

Furthermore, though Appellant contends that “[c]laim 1[sic] relates to
ancestors of an item associated with a selection event, and not with
ancestors of a ‘group of related items’” (App. Br. 7), we find an ancestor of
a group of related items associated with a selection event to be an ancestor
or “an item” associated with the selection event. That is, a group of items
comprises “an item.”
Although we agree with the Examiner in finding that Ford discloses
“identifying each ancestor node of the selected item within the browse tree”
(claim 11), we agree with Appellant that there is no teaching in the sections
of Ford pointed out by the Examiner of “dividing said total amount of credit
by the number of ancestor nodes of the selected item” (App. Br. 6). That is,
in view of our finding of “categories” to be “ancestor nodes,” in Ford, the
mean value is determined by dividing the sum of item scores by the total
number of items (FF 2), and not by the number of categories. We agree with
Appellant’s argument that, in Ford, “the divisor is a number of different
items (not a number of ancestor nodes of an item associated with a selection
event)” (App. Br. 6). In fact, even the Examiner admits that, in Ford, “the
total score within a category is added up and divided by the number of
items” (Ans. 13).
As such, we find that Appellant has shown that the Examiner erred in
rejecting claim 11 and claims 12-14 standing therewith over Ford.

35 U.S.C. § 103(a)
Claims 1-10 and 15-16
Appellant contends that “Herz does not calculate the topical interest f
by evaluating the difference between the relevance feedback for a particular
8
Appeal 2009-002829
Application 10/684,313

user for the target object and the relevance feedback from a population of
users for that target object” but instead “an interpolation is done on the
function f(V,Y), which takes into account the f value for all evaluated objects
(V) and for all user (Y)” (App. Br. 9). That is, Appellant argues that Herz
does not disclose or suggest “a differencing carried out between the
relevance feedback (e.g. click and browse behavior) of an individual user
and the relevance feedback of a population of users” (App. Br. 10).
However, the Examiner finds that “the preference of a user is
computed by utilizing specific data made by a user and data from similar
user’s through feedback[] from users” (Ans. 15-16), where “the similarity
distance between user preference and other user’s preferences are calculated
to predict relevance of an object to a user” (Ans. 16). In particular, the
Examiner finds that “[t]he attribute data for a user is interpreted to be the
user history scores of the present invention and is compared to attribute data
from other users to find the difference” (id.). That is, the Examiner finds
that “the similarity distance between (U,X), representing a user’s interest in
an object and (V, Y), representing different users’ interest in an object, can
be used for topical interest” (id.).
Herz discloses computing the likelihood of interest in a particular
(target) object for a specific user by summing the estimated quality measure
for the object with the topical interest of like users for the object (FF 3),
wherein estimating the quality measure of the object includes using the
object’s popularity among users and multiplying such attributes by a value
indicative of the specific user’s preference (FF 4). In Herz, estimating the
particular user’s interest includes interpolation using the similarly distance
between 1) the attributes of the particular object with the attributes of the
9
Appeal 2009-002829
Application 10/684,313

particular user, and 2) the attributes of similar objects with the attributes of
the similar users, wherein the particular user’s interest can be estimated even
if the particular user is a new user or has never provided feedback because
the relevance feedback of users whose attributes are similar to the particular
user’s attributes is taken into account (FF 5).
We agree with Appellant that that Herz does not disclose or suggest “a
differencing carried out between the relevance feedback (e.g. click and
browse behavior) of an individual user and the relevance feedback of a
population of users” (App. Br. 10). That is, we disagree with the Examiner’s
finding that, in Herz, “the similarity distance between user preference and
other user’s preferences are calculated to predict relevance of an object to a
user” (Ans. 16). That is, in Herz, the particular user is a new user or has
never provided feedback (FF 5).
Thus, contrary to the Examiner’s finding that “[t]he attribute data for a
user is interpreted to be the user history scores of the present invention and
is compared to attribute data from other users to find the difference” (Ans.
16), we find that there is no teaching in the sections of Herz pointed out by
the Examiner of “assigning individual user history scores to specific
categories of the browse tree based at least in-part on an item selection
history” (Claim 1) since there is no selection history for the particular user
who is a new user or a user who has not provided feedback. Thus, we find
Herz also does not disclose “evaluating differences between the individual
user history scores and the collective user history scores to generate a
relative preference profile for the user” as required by claim 1.
10
Appeal 2009-002829
Application 10/684,313

We find that Ortega does not cure these deficiencies. As such, we
find that Appellant has shown that the Examiner erred in rejecting claim 1-
10 and 15-16 standing therewith over Ortega in view of Herz.

Claims 2-3 and 16
We find that Ford does not cure the deficiencies of Ortega and Herz.
As such, we will also reverse the rejection of claims 2-3 and 16 over Ortega
in view of Herz and Ford.

V. CONCLUSION
Appellant has shown that the Examiner erred in finding claims 11-14
anticipated by the teachings of Ford under 35 U.S.C. § 102(e); in holding
claims 1-10 and 15-16 unpatentable over the teachings of Ortega in view of
Herz under 35 U.S.C. § 103(a); and in holding claims 2-3 and 16
unpatentable over the teachings of Ortega in view of Herz and Ford under 35
U.S.C. § 103(a).

VI. DECISION
We have not sustained the Examiner's rejection with respect to any
claim on appeal. Therefore, the Examiner’s decision rejecting claims 1-16 is
reversed.

REVERSED

11
Appeal 2009-002829
Application 10/684,313

peb

KNOBBE MARTENS OLSON & BEAR LLP
2040 MAIN STREET
FOURTEENTH FLOOR
IRVINE, CA 92614

12