Ex Parte Robinson et al

Patent Trial and Appeal Board

Jul 27, 2018

12789493 (P.T.A.B. Jul. 27, 2018)

UNITED STA TES p A TENT AND TRADEMARK OFFICE
APPLICATION NO. FILING DATE
12/789,493 05/28/2010
23117 7590 07/31/2018
NIXON & V ANDERHYE, PC
901 NORTH GLEBE ROAD, 11 TH FLOOR
ARLINGTON, VA 22203
FIRST NAMED INVENTOR
Stephen R. Robinson
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
ATTORNEY DOCKET NO. CONFIRMATION NO.
1266-55 8738
EXAMINER
AMORIN, CARLOS E
ART UNIT PAPER NUMBER
2498
NOTIFICATION DATE DELIVERY MODE
07/31/2018 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):
PTOMAIL@nixonvan.com
pair_nixon@firsttofile.com
PTOL-90A (Rev. 04/07)
UNITED STATES PATENT AND TRADEMARK OFFICE
BEFORE THE PATENT TRIAL AND APPEAL BOARD
Ex parte STEPHEN R. ROBINSON, TONY ROBINSON,
and
ROB BURSON
Appeal2018-001327
Application 12/789,493
Technology Center 2400
Before: CARLA M. KRIVAK, JEREMY J. CURCURI, and
JON M. JURGOV AN, Administrative Patent Judges.
KRIVAK, Administrative Patent Judge.
DECISION ON APPEAL
Appellants appeal under 35 U.S.C. § 134(a) from the Examiner's
Final rejection of claims 1-5 and 12-21. We have jurisdiction under 35
U.S.C. § 6(b).
We affirm.
Appeal2018-001327
Application 12/789,493
STATEMENT OF THE CASE
Appellants' invention is directed to "using predictive analysis based
on a data set of previous undesirable accesses to detect and interdict further
undesired accesses" (Spec. ,r 3).
Independent claim 1, reproduced below, is exemplary of the subject
matter on appeal.
1. A computer arrangement connected to a computer
network, said computer arrangement reducing an impact of
undesired server accesses, the computer arrangement
compnsmg:
a storage device storing an historical server access data
set including server transactions that are known to have
constituted non-human scraping accesses and server
transactions that are known to have constituted human accesses;
a computer processor operatively coupled to the storage
device, the computer processor configured to apply machine
learning against the stored historical server access data set to
automatically train an automated classifier predictive model, the
trained automated classifier predictive model enabling an
associated predictive classifying agent to predict whether future
server accesses are non-human scraping accesses, the computer
processor performing training based on data in the data set
indicating which server transactions constituted non-human
scraping accesses and which server transactions constituted
human accesses, the same or different computer processor
communicating with plural monitoring computers and
executing said predictive classifying agent to detect non-human
scraping accesses in real time, the predictive classifying agent
also learning and using identified signatures of undesired server
accesses by particular non-human scraper agents, a first
signature indicating characteristic ways that a particular first
non-human scraper agent accesses webpages, a second
signature indicating characteristic ways that a particular second
non-human scraper agent accesses web pages, to help predict
non-human scraping accesses;
2
Appeal2018-001327
Application 12/789,493
the plural monitoring computers monitoring accesses to
corresponding plural servers, the plural monitoring computers
communicating with the executing predictive classifying agent
over a computer network, the monitoring computers performing
online monitoring of associated corresponding servers and
reporting monitoring results to the predictive classifying agent
over the computer network, the predictive classifying agent
using said reported results and said trained predictive model to
classify between human server accesses and non-human
scraping accesses, and also to further iteratively train the
predictive model;
wherein the historical data used to train the classifier
predictive model includes historical server access data collected
before a given said monitoring computer began monitoring an
associated one of said servers and information said predictive
classifying agent harvests from a first server is used to predict
or detect undesired accesses of a second server different from
said first server;
the same or different computer processor being further
configured to analyze said monitored accesses to recognize
patterns and/or characteristics associated with scraping
accesses;
the same or different computer processor being further
configured to, based at least in part on the trained predictive
model, predict a likelihood that a reported access constitutes a
scraping access being made by a non-human agent, said
predicting including using identified first and second signatures
of particular first and second non-human agents that make
undesired scraping accesses; and
the same or different computer processor being further
configured to, if said analyzing predicts that a particular
monitored access is likely to be a scraping access made by a
non-human agent, select, based on a predictive certainty factor
determined for the particular monitored access, at least one
interdiction action in substantially real time response to said
particular monitored access to prevent further scraping.
3
Appeal2018-001327
Application 12/789,493
REFERENCES and REJECTIONS 1
The Examiner rejected claims 1-5 and 12-14, 17, and 19-21 under 35
U.S.C. § I03(a) based upon the teachings of Broder (US 2008/0147456 Al;
pub. June 19, 2008), Petta (US 2009/0288169 Al; pub. Nov. 19, 2009) and
W ebcrawler, Wikipedia,
http://web.archive.org/web/2008030706561 O/http://en.wikipedia.org/wiki/Webcrawler,
last visited June 2014, (hereinafter "Wikipedia").
The Examiner rejected claims under 35 U.S.C. § I03(a) based upon
the teachings of Broder, Petta, Wikipedia, and Britton (US 2009/0157875
Al, pub. June 18, 2009).
The Examiner rejected claim 16 under 35 U.S.C. § I03(a) based upon
the teachings of Broder, Petta, Wikipedia, and Husic (US 2009/0282062 Al;
pub. Nov. 12, 2009).
ANALYSIS
Appellants contend the Examiner erred in finding the combination of
Broder, Petta, and Wikipedia teach or suggest "the predictive classifying
agent using said reported results and said trained predictive model to classify
between human server accesses and non-human scraping accesses, and also
to further iteratively train the predictive model" as claimed (Br. 6). That is,
Appellants contend, Broder discloses "click fraud" using bots to "'click
through' an advertising link on a web page to inflate advertising revenues,"
thus, distinguishing "between clicks performed by users and clicks
performed by bots" using an "intermediate page" (CAPTCHA), and not a
1 The Examiner withdrew the rejection of claims 1-5 and 12-21 under 35
U.S.C. § 112(a) in the Advisory Action mailed January 20, 2017.
4
Appeal2018-001327
Application 12/789,493
predictive model using known scraper data (Br. 6-7). Appellants further
contend "Broder does not disclose []a predictive model using known scraper
access" (Br. 9 ( emphasis omitted)), "selecting interdiction based on a
predictive certainty factor" (Br. 10 ( emphasis omitted)), "using predictive
analysis to predict scraping access" (Br. 13 ( emphasis omitted)), and "using
historical data collected before the current computer began collecting data"
(Br. 14 (emphasis omitted)); rather, Broder merely distinguishes human
"click throughs" from non-human click throughs, and "has no way to
classify non-human scraper accesses" (Br. 8). We do not agree.
We agree with and adopt the Examiner's findings as our own (Ans.
23-32). Specifically, the Examiner finds although Broder does not use the
term "scraping accesses," scraping accesses are similar to web crawling (see
Wikipedia) and Petta specifically teaches "web scraping" (Petta Title
"Systems and Methods to Control Web Scraping" (see Petta ,r 30; see also
Ans. 23-24 ). Wikipedia states web crawlers "gather specific types of
information for Web pages" and Petta states "Web scraping generally
includes activities to extract data or content from a website through manual
or automated processes" (Petta ,r 2). Appellants' claimed "scraping
accesses" reads on the term web crawling and web scraping. The Examiner
has provided articulated reasoning with a rational underpinning for
combining the references supported by evidence, which Appellants do not
rebut. Therefore, we agree with the Examiner's broad but reasonable
interpretation of Appellants claimed term "scraping accesses" as no express
definitions are provided in Appellants' Specification. Appellants own
Specification states "[ o ]n a more detailed technical level, plagiaristic
webcrawlers often perform an operation known as 'web scraping' or 'page
5
Appeal2018-001327
Application 12/789,493
scraping."' (Spec ,r,r 11, 39 ("[S]craper/webbot/webcrawler computer or
other non-human browser agent.")). Appellants have not provided sufficient
evidence to contradict the Examiner's findings. Further, Appellants are
arguing the references separately and not as a combination (see In re Keller,
642 F.2d 413,426 (CCPA 1981) ("[O]ne cannot show non-obviousness by
attacking references individually where, as here, the rejections are based on
combinations of references.")).
As to Appellants' remaining arguments we agree with the Examiner
that Broder' s paragraphs 73 and 7 4 teach and suggest machine learning
algorithms determining/predicting whether a web view is human or non-
human (Ans. 24--25; see also Broder ,r 74 ("[A] Bayesian classifier may be
created in order to help identify non[-]human web viewer entities.")). 2
Further, contrary to Appellants argument that Broder classifies
interactions with the intermediate pages (CAPTCHA pages) (Br. 10-11),
Appellants' invention appears to do the same (Spec. ,r 29 ("Scraper
remediation (from low-impact to high-impact interdiction) can include for
2 A Bayesian classifier is defined as follows:
Bayesian classification is based on Bayes' Theorem. Bayesian classifiers are
the statistical classifiers. Bayesian classifiers can predict class membership
probabilities such as the probability that a given tuple belongs to a particular
class.
Baye's Theorem
Bayes' Theorem is named after Thomas Bayes. There are two types of
probabilities -
• Posterior Probability [P(H/X)]
• Prior Probability [P(H)]
where Xis data tuple and His some hypothesis. According to Bayes'
Theorem, P(H/X)= P(X/H)P(H) I P(X)
https://www.tutorialspoint.com/ data_mining/ dm_bayesian_classification.htm
6
Appeal2018-001327
Application 12/789,493
example: []Display of a 'captcha' page (page requiring human interpretation
and action) to the scraper."); see also Ans. 28-29). Thus, we concur with
the Examiner's conclusion the combination of Broder, Petta, and Wikipedia
teaches and suggests Appellant's claimed invention. We therefore sustain
the Examiner's rejection of independent claim 1 and independent claim 12
argued for substantially the same reasons as claim 1, dependent claims 4, 15,
and 18 dependent from claims 1 or 12, and claims 2, 3, 5, 13, 14, 16, 17, and
19-21 for which no arguments were presented.
Thus, on this record, we are not persuaded the Examiner's reading of
the claims on the cited combination of references is overly broad,
unreasonable, or inconsistent with the Specification. We find the weight of
the evidence supports the Examiner's ultimate legal conclusion of
obviousness, and sustain the Examiner's rejection of claims 1-5 and 12-21.
DECISION
The Examiner's decision rejecting claims 1-5 and 12-21 is affirmed.
No time period for taking any subsequent action in connection with
this appeal may be extended under 37 C.F.R. § 1.136(a)(l)(iv).
AFFIRMED
7