Ex Parte Stack

Board of Patent Appeals and Interferences

Jun 30, 2003

08923293 (B.P.A.I. Jun. 30, 2003)

1 Application for patent filed September 4, 1997, entitled
"System and Method for Providing Recommendation of Goods or
Services Based on Recorded Purchasing History."
- 1 -
The opinion in support of the decision being
entered today was not written for publication
and is not binding precedent of the Board.
_______________
Paper No. 24
UNITED STATES PATENT AND TRADEMARK OFFICE

BEFORE THE BOARD OF PATENT APPEALS
AND INTERFERENCES

Ex parte CHARLES STACK

Appeal No. 2001-2051
Application 08/923,2931

ON BRIEF

Before THOMAS, BARRETT, and GROSS, Administrative Patent Judges.
BARRETT, Administrative Patent Judge.
DECISION ON APPEAL
This is a decision on appeal under 35 U.S.C. § 134(a) from
the final rejection of claims 1-16.
We affirm-in-part.
Appeal No. 2001-2051
Application 08/923,293
- 2 -
BACKGROUND
The invention relates to a system, method, and computer
program product for providing recommendations of goods and/or
services of potential interest to customers based upon the
customer's selection of other goods and/or services, in
conjunction with a database of the purchasing history of other
customers of the merchant. For example, if a user selects a
specific book that he or she is interested in purchasing, the
system searches the database to determine all previous customers
who have purchased that book and, if those previous customers
have also purchased other books in common, the books will be
displayed. A common example today, although not known to this
panel to be prior art to the present application, is found on
websites such as "www.barnesandnoble.com" where selection of a
book results in information on the book and a list of books under
the heading of "People who bought this book also bought:."
Claim 16 is reproduced below.
16. A computer program product having a computer readable
medium having computer readable code recorded thereon for
the recommendation of goods or services in response to user
input, comprising:
input means for receiving user commands representing
any of a plurality of goods or services to be used as filter
data;
database storage means for the retention of data
concerning goods or services purchase decisions of prior
users; and
Appeal No. 2001-2051
Application 08/923,293
- 3 -
means for filtering said database storage means using
said inputted user commands to obtain recommendations to a
user based on said inputted user commands.
The examiner relies on the following references:
Robinson 5,790,426 August 4, 1998
(filed April 30, 1997)
Payton 5,790,935 August 4, 1998
(filed January 30, 1996)
Claims 1-16 stand rejected under 35 U.S.C. § 103(a) as being
unpatentable over Robinson and Payton.
We refer to the final rejection (Paper No. 14) (pages
referred to as "FR__") and the examiner's answer (Paper No. 21)
(pages referred to as "EA__") for a statement of the examiner's
rejection, and to the brief (Paper No. 20) (pages referred to as
"Br__") and reply brief (Paper No. 22) (pages referred to as
"RBr__") for a statement of appellant's arguments thereagainst.
OPINION
Grouping of claims
Appellant states that the claims are grouped to stand or
fall together (Br5) and appears to argue mainly the limitations
of independent claim 1. We find the limitations of the
independent claims to be too different to be treated the same.
Hence, we address the independent claims separately.
The references
Robinson discloses an automated collaborative filtering
(ACF) system for recommending an item to a first user based on
Appeal No. 2001-2051
Application 08/923,293
- 4 -
ratings of items by other system users (col. 1, lines 53-56).
The items may be any product or service that can be rated by a
user (col. 3, lines 24-27). The improvement involves determining
a similarity in preference of the user as compared to other
users, which is used to determine a subgroup of users having
preferences similar to the first user, and the ratings of the
subgroup of users are then used to recommend an item to the first
user (abstract; col. 6, lines 29-35). In order to determine a
similarity value, the first user and other user must have ratings
in the database for at least one common item (col. 6, lines 25-27
& 38-41). The users in Robinson are required to actively
participate by entering their subjective ratings of items in the
group, in contrast to the disclosed invention which uses previous
customer activity history to reduce the degree of customer input
in prior art systems (specification, p. 2, lines 6-16).
Payton is directed to a virtual on-demand information system
(such as for movies, games, computer applications) wherein the
information is downloaded to a storage device at the subscriber
during non-peak hours based on an information filtering system
that predicts items the subscriber might like, and therefore
request. A subscriber database stores a subscriber profile which
includes a rating vector in which the user subscriber has rated
each of the items he or she has previously requested (col. 5,
lines 6-12). Alternatively, the system could just record the use
Appeal No. 2001-2051
Application 08/923,293
- 5 -
of a requested item as a positive vote (col. 6, lines 41-42). A
collaborative filtering system acts on the subscriber profiles
and predicts which of the available items the subscriber may be
interested in and request, and produces a list of those
recommended items for each subscriber (col. 5, lines 12-16). A
number of different collaborative filtering algorithms may be
used (col. 9, lines 4-61).
Analysis
The examiner finds that "Robinson does not expressly
disclose a database containing information pertaining to goods
and/or services purchasing history of previous customers, but
Payton teaches a database of ratings by previous customers
(column 5, lines 6-10)" (FR2; EA4). Appellant notes that the
final rejection acknowledges that Robinson does not disclose the
claimed database of customer purchase history (Br8). The
examiner responds that although Robinson does not expressly
disclose such a database (EA9):
Robinson discloses the use of a database pertaining to the
goods and/or services of previous users; since the previous
users would in many cases have purchased the movies, books,
or other goods and services they rated, Robinson's data in
fact would pertain to their purchasing history, contrary to
Appellant's assertion. Examiner wishes to note that
"pertain" is a broad word, and that independent claims 1 and
7 specify only "information pertaining to goods and/or
services purchasing history," without limitations to
detailed information on which prior users have purchased
which goods and/or services.
Appeal No. 2001-2051
Application 08/923,293
- 6 -
Appellant argues that the examiner's reasoning is contradictory,
but, in any event, the combination does not disclose using the
specification of a good/service as filter data for a previous
customer purchase history database (RBr3).
We find that Robinson does not disclose a database storing
"information pertaining to goods and/or services purchasing
history of previous customers" (claims 1 and 7) or "data
concerning goods or services purchase decisions of prior users"
(claim 16). Robinson is a recommendation service, not a sales
service, and, therefore, it does not teach maintaining purchasing
history. The examiner's reasoning that "pertaining to ...
purchase history" is broad enough to read on the users' ratings
of items in Robinson because, indirectly, the users had to have
purchased the items somewhere, is not necessary since Payton
expressly discloses that the ratings can be based on requested
(purchased) items. We conclude it would have been obvious to one
skilled in the art that the ratings in Robinson could be based on
request (purchase) history in view of Payton.
The examiner finds that "Robinson discloses receiving
customer commands specifying a particular good or service to be
used as filter data (column 6, lines 24-35; see also column 6,
line 56, through column 7, line 46)" (FR2; EA3; see also EA10).
Appellant argues that these passages deal with procedures for
determining the proper subgroup of users (RBr1) and have "nothing
Appeal No. 2001-2051
Application 08/923,293
- 7 -
to do with entering a specified item as filter data for obtaining
recommendations of other goods and/or services from a previous
customer purchasing history database, as set forth in the claims
on appeal" (RBr2) and that there is no "user input" of a
specified item in Robinson (RBr2).
The examiner does not particularly point out what
description in Robinson is relied on as the customer commands.
The only thing we see that can possibly be the customer commands
are the ratings that the first user enters into the database at
step 60 (col. 6, lines 25-27 & 38-41), which ratings form the
basis for the recommendations. Payton also contains the same
teachings about the subscriber rating each of the items he or she
has previously requested (col. 5, lines 6-12), which ratings form
the basis for the recommendation. However, these ratings do not
satisfy the claim language of claims 1 and 7.
Claim 1 recites "specifying a particular good or service to
be used as filter data." While the ratings in Robinson and
Payton are used as filter data for a recommendation, the ratings
do not specify of a particular good or service to be used as
filter data for a recommendation. That is, the user does not
specify a particular good or service, such as specifying the book
Clear and Present Danger in appellant's example of Fig. 3D, which
is then used as filter data to get a recommendation. The user
gets a recommendation based on past ratings without specifying a
Appeal No. 2001-2051
Application 08/923,293
- 8 -
particular good or service. Therefore, the rejection of claim 1
and its dependent claims 2-6 is reversed.
Claim 7 recites, inter alia, an operator interface for
inputting requests for "the purchase of goods or services,"
"means for processing inputted requests and for filtering
relevant history information regarding said inputted requests
from said database," and "whereby goods and/or services
identification information corresponding to goods and/or services
purchased by previous customers who have purchased the goods
and/or services requested by said potential customers are
transmitted to said operator interface for use by said potential
customers" (emphasis added). This requires filtering the
purchasing history of previous customers based on "the goods
and/or services requested by said potential customers," which we
interpret to be equivalent to "specifying a particular good or
service to be used as filter data," as recited in claim 1. As we
found with respect to claim 1, neither Robinson nor Payton
discloses specifying a particular good or service to be used as
filter data for a recommendation. Therefore, the rejection of
claim 7 and its dependent claims 8-15 is reversed.
Claim 16 is broader than claims 1 and 7 and, we conclude,
would have been obvious over Robinson and Payton. The limitation
of "input means for receiving user commands representing any of a
plurality of goods or services to be used as filter data" is
Appeal No. 2001-2051
Application 08/923,293
- 9 -
broad enough to read on the input means for inputting ratings in
Robinson and Payton, where the rated goods or services are used
as filter data. Note that this limitation of claim 16 does not
require "specifying a particular good or service to be used as
filter data," as recited in claim 1, or filtering the purchasing
history of previous customers based on "the goods and/or services
requested by said potential customers," as recited in claim 7.
Thus, claim 16 does not capture the idea of using a specified
good or service as a filter for obtaining recommendations of
other goods and services. Payton discloses "database storage
means for the retention of data concerning goods or services
purchase decisions of prior users" because it stores a database
of subscriber profile rating vectors which vectors contain data
concerning items which the user subscriber has previously
requested (col. 5, lines 6-12), where requested information (such
as a pay-per-view movie) is purchased information, and the
ratings can be based on the use of a requested item (col. 6,
lines 41-42), i.e., the ratings directly reflect purchase
(request) decisions of previous users. For this reason, we
disagree with appellant's argument that "neither Robinson nor
Payton disclose a database of previous customer purchase history,
as required by the claims on appeal" (RB2). It would have been
obvious to one of ordinary skill in the art that the
recommendation system in Robinson could be applied in a purchase
Appeal No. 2001-2051
Application 08/923,293
- 10 -
(request)/recommendation system as taught by Payton, although
Robinson does not appear to be necessary to the rejection. The
limitation of "means for filtering said database storage means
using said inputted user commands to obtain recommendations to a
user based on said inputted user commands" reads on the
recommendation systems of Robinson and Payton. For these
reasons, the rejection of claim 16 is sustained.
CONCLUSION
The rejection of claims 1-15 is reversed. The rejection of
claim 16 is affirmed.
No time period for taking any subsequent action in
connection with this appeal may be extended under 37 CFR
§ 1.136(a).
AFFIRMED-IN-PART
JAMES D. THOMAS )
Administrative Patent Judge )
)
)
)
) BOARD OF PATENT
LEE E. BARRETT ) APPEALS
Administrative Patent Judge ) AND
) INTERFERENCES
)
)
)
ANITA PELLMAN GROSS )
Administrative Patent Judge )
Appeal No. 2001-2051
Application 08/923,293
- 11 -
ROTHWELL, FIGG, ERNST & MANBECK, P.C.
1425 K STREET, N.W.
SUITE 800
WASHINGTON, DC 20005