TNS Media Research, LLC et al v. TiVo Research and Analytics, Inc.

Filing 89

OPINION AND ORDER: For the aforementioned reasons I construe the disputed terms within claim one as follows: When used with regard to clickstream and purchase data, household level data associated with multiple consumer households means data about a household that can be later aggregated into a data set including multiple consumer households." When used with regard to advertising and program data, "household level data associated with multiple consumer households" means "data related to advertising/programming tuned to by multiple consumer households." "Cleansing and editing algorithm" means "an algorithm to remove inconsistencies in, correct, or otherwise improve the reliability of data collected from a program delivery source." A status conference is scheduled for September 19,2012 at 3:00 pm. (Signed by Judge Shira A. Scheindlin on 8/29/2012) (jar)

UNITED STATES DISTRICT COURT
SOUTHERN DISTRICT OF NEW YORK

------------------------------------------------------- )(
TNS MEDIA RESEARCH, LLC (d/b/a
KANTAR MEDIA AUDIENCES) and
CAVENDISH SQUARE HOLDING B.V.,
Plaintiffs,
- againstTRA GLOBAL, INC. (d/b/a TRA, INC.),

11 Civ. 4039 (SAS)

Defendant.

------------------------------------------------------- )(
TRA GLOBAL, INC. (d/b/a TRA, INC.),
Counterclaim-Plaintiff,
- againstTNS MEDIA RESEARCH, LLC (d/b/a
KANTAR MEDIA AUDIENCES),
CAVENDISH SQUARE HOLDING B.V.,
WPP PLC, WPP GROUP USA, INC.,
KANTAR GROUP LTD., KANTAR
RETAIL AMERICA, INC.,
Counterclaim-Defendants.

------------------------------------------------------- )(
SHIRA A. SCHEINDLIN, U.S.D.J.:

I.

OPINION AND
ORDER

INTRODUCTION
-1­

TNS Media Research, LLC d/b/a Kantar Media Audiences and
Cavendish Square Holding, B.V. (collectively “Kantar”) bring this action against
TRA Global, Inc. (“TRA”) seeking a declaratory judgment that Kantar’s
RapidView for Retail product (“RVR”) did not infringe on U.S. Patent No.
7,729,940 (“the ‘940 Patent”), owned by TRA. Cavendish also brings a claim
against TRA for breach of contract.1 TRA has asserted various affirmative
defenses and counterclaims.
Claim construction – the process wherein the court determines, as a
matter of law, the meaning of disputed claim terms – is a task preferably tackled
early on in a patent infringement action. A Markman hearing – which provides the
parties the opportunity to argue for, and introduce evidence in support of, their
proposed constructions – was held on July 6, 2012. The parties dispute the
meaning of three phrases appearing throughout the ‘940 Patent. For ease of
reference, a list of the constructions I have adopted is included at the conclusion of
this Opinion.
II.

BACKGROUND
A.

The Invention
On June 1, 2010, the Patent and Trademark Office (“PTO”) issued the

1

See Complaint and Jury Demand (“Compl.”) ¶¶ 17-21.
-2-

‘940 Patent with TRA designated as the sole assignee.2 The claimed invention is a
method for collecting, matching, and analyzing television viewing and consumer
purchasing data to generate reports that help advertisers get the most bang for their
advertising bucks. Further, the claimed invention does this without the need for
supplemental equipment in homes or retail locations, without the need for the
consumer’s “opt-in” permission, and with measures designed to protect the privacy
of the consumer.3 Media TRAnalytics (“MTRA”) — TRA’s product based on the
‘940 Patent — collects and matches four kinds of data: (1) data from cable and
satellite set-top boxes to determine what channel a household is watching
(“clickstream data”); (2) general programming data from third-party providers that
allows it to determine what show was on that channel; (3) general advertising data
from third-party providers that allows it to determine what ads ran during that
show; and (4) household purchasing data collected based on frequent shopper or
“loyalty” cards from supermarkets and other sources.4 Using a combination of
outside vendors and proprietary algorithms, TRA matches individual viewer data
2

See ‘940 Patent, Ex. B to 1/20/12 Declaration of Richard Fenwick
(TRA’s expert) in Support of Claim Construction (Markman Hearing) (“Fenwick
Decl.”).
3

See Defendant/Counterclaim-Plaintiff TRA Global, Inc.’s Responsive
Claim Construction Brief (“TRA Reply Mem.”) at 4.
4

See 7/6/12 Markman Hearing Transcript (“Tr.”) at 4:14-7:20.
-3-

with individual purchase data while anonymizing the data to protect the privacy
rights of television viewers.5 TRA then generates reports that help advertisers: (1)
measure the return on their investment for previous ads; and (2) how best to invest
their future advertising dollars.6
B.

Claim Language
While the phrases at issue are found in more than one patent claim,

Claim 71 is illustrative of the invention, and both parties make their arguments
using Claim 71 as a reference. The claimed method is made up of five steps: (1)
collecting data; (2) matching with a centrally located electronic computer
processor; (3) storing the matched data in at least one centrally located electronic
data storage media; (4) applying at least one “cleansing and editing algorithm”; and
(5) calculating at least one true target index matrix. The full text of Claim 71
follows, with the disputed phrases emphasized:
A computer-implemented method for facilitating analysis of
consumer behavior in association with advertising exposure or
program delivery, the method comprising:
collecting in an advertising measurement system:
(i) clickstream data from a program delivery source of a
consumer, wherein collecting the clickstream data is
not dependent on a supplemental data collection
5

See TNS Media Research, LLC v. TRA Global, Inc., No. 11 Civ. 4039,
2011 WL 4425415, at *1 (S.D.N.Y. Sept. 22, 2011).
6

See Tr. at 7:21-8:5.
-4-

device, and also wherein the collected clickstream data
includes household level data associated with multiple
consumer households;
(ii) advertising data associated with delivery of the
program by the program delivery source, wherein
collecting the advertising data is not dependent on a
supplemental data collection device, and also wherein
the collected advertising data includes household level
data associated with multiple consumer households;
(iii) program data associated with the program delivered on
the program delivery source, wherein collecting the
program data is not dependent on a supplemental data
collection device, and also wherein the collected
program data includes household level data associated
with multiple consumer households;
(iv) purchase data from a purchase data source, wherein
collecting the purchase data is not dependent on a
supplemental data collection device, and also wherein
the collected purchase data includes household level
data associated with multiple consumer households;
matching at least portions of the collected advertising data, the
collected clickstream data, the collected purchase data, and
the collected program data in the advertising measurement
system at a household data level with a centrally located
electronic computer processor configured for centrally
processing data received from the program delivery source,
the advertising data source, the program data source, and
the purchase data source, wherein the matching further
includes:
(i) grouping the collected data in association with an
account identifier of each consumer household without
processing any personally identifiable information
associated with the consumer household, and
(ii) matching each account identifier associated with each
consumer household with other account identifiers
associated with the same consumer household without
processing any personally identifiable information
associated with the consumer household;
-5-

storing the matched advertising data, clickstream data,
purchase data, and program data in at least one centrally
located electronic data storage medium operatively
associated with the computer processor;
applying at least one cleansing and editing algorithm to the
matched and stored data; and,
calculating at least one true target index metric based on the
matched and stored data.7
The parties seek construction of the phrases “household level data associated with
multiple consumer households” and “cleansing and editing algorithm.”8 Kantar
additionally seeks construction of the sub-phrase “household level data.”
C.

Procedural History
Kantar commenced this suit on June 15, 2011, by filing a complaint

seeking a declaratory judgment that the RVR does not infringe the ‘940 Patent.9
Kantar further alleged that TRA breached the terms of a Voting Agreement when it
excluded a senior vice president from WPP PLC — the parent company of Kantar
— from attending a meeting of TRA’s Board of Directors.10 TRA answered on
July 5, 2011, and asserted four counterclaims: (1) patent infringement; (2) breach
of fiduciary duty; (3) misappropriation of trade secrets; and (4) breach of contract

7

‘940 Patent col. 46 l. 33 – col. 48 l. 9.

8

See Parties Joint Claim Construction Statement (“Joint St.”) at 3-4.

9

See Compl. ¶¶ 16-18, 22.

10

See id. ¶¶ 19-21.
-6-

relating to the terms of certain non-disclosure agreements.11
With the filing of its Answer, TRA moved for preliminary injunctive
relief. I denied injunctive relief on September 22, 2011, and, on the record then
before me, made the following determinations: (1) that the phrase in Claim 71 of
the ‘940 Patent — “household level data associated with multiple consumer
households” — had different meanings in different contexts; (2) that because
Claim 71 would have been obvious to a person reasonably skilled in the art,
Kantar met its burden of raising a substantial question about the patent’s validity;
and (3) that TRA had not shown by a preponderance of the evidence that the
obviousness defense raised by Kantar lacked substantial merit.12 On June 4, 2012,
I granted TRA leave to amend its counterclaims by: (1) asserting additional claims
of patent infringement based on U.S. Patent Nos. 8,000,993 and 8,112,301, both of
which were issued to TRA after the date it filed its original counterclaims; (2)
clarifying that its breach of fiduciary duty claim was actually an aiding and
abetting breach of fiduciary duty claim; and (3) joining additional Kantar entities

11

See Answer, Defenses, and Counterclaims for Patent Infringement,
Breach of Fiduciary Duty, Misappropriation of Trade Secrets, and Breach of
Contract ¶¶ 105-129.
12

See TNS Media Research, 2011 WL 4425415, at *5-6.
-7-

as counterclaim-defendants.13
On February 3, 2012, the parties stipulated as to the appropriate
construction for many terms in the ‘940 patent.14 Disputes remain as to three
phrases, each of which appears in every one of TRA’s asserted claims.15 A
Markman hearing was held on July 6, 2012.
III.

APPLICABLE LAW
Before a fact-finder can determine whether an accused device

infringes a patent, the terms of the asserted claims must be construed to determine
their scope. Because the proper interpretation of words in a patent claim is a
question of law, claim construction is a task for judges, not juries.16 Judges should
carry out this task with the goal of “‘elaborating the normally terse claim language

13

See TNS Media Research, LLC v. TRA Global, Inc., No. 11 Civ.
4039, 2012 WL 2052679, at *2 (S.D.N.Y. June 4, 2012); Answer, Defenses, and
Supplemental and Amended Counterclaims for Patent Infringement, Aiding and
Abetting Breach of Fiduciary Duty, Misappropriation of Trade Secrets, and Breach
of Contract ¶¶ 104-151.
14

See Joint St.

15

See Plaintiffs’ Opening Claim Construction Statement (“Kantar

Mem.”).
16

See Markman v. Westview Instruments, Inc., 517 U.S. 370, 384, 390-

91 (1996).
-8-

in order to understand and explain, but not to change, the scope of the claims.’”17
That said, “[c]laim interpretation is not always an exact science, and it is not
unusual for parties to offer competing definitions of even the simplest claim
language.”18
Although there are several canons of construction used by courts in
interpreting claim terms, the Federal Circuit has made clear that “there is no magic
formula or catechism for conducting claim construction.”19 Thus, “while these
interpretive tools can be indispensable aids to a federal district court tasked with
construing the meaning of a patent claim,”20 courts must nonetheless be mindful
that “applying [the substantive precepts of patent claim interpretation] to the facts
of a case can be . . . perplexing, particularly where precepts can be in opposition to
one another and district judges cannot always tell which one(s) apply in a

17

DeMarini Sports, Inc. v. Worth, Inc., 239 F.3d 1314, 1322 (Fed. Cir.
2001) (quoting Embrex, Inc. v. Services Eng’g Corp., 216 F.3d 1343, 1347 (Fed.
Cir. 2000)).
18

Q-Pharma, Inc. v. Andrew Jergens Co., 360 F.3d 1295, 1301 (Fed.

Cir. 2004).
19

Phillips v. AWH Corp., 415 F.3d 1303, 1324 (Fed. Cir. 2005) (en

banc).
20

Medisim Ltd. v. BestMed LLC, No. 10 Civ. 2463, 2011 WL 2693896,
at *3 (S.D.N.Y. July 8, 2011).
-9-

particular case.”21 Mindful of this challenge, the Federal Circuit has attempted to
give district courts some guidance as to where to begin:
[T]he ordinary and customary meaning of a claim term is the
meaning that the term would have to a person of ordinary skill in
the art in question at the time of the invention, i.e. as of the
effective filing date of the patent application. . . . [This inquiry]
provides an objective baseline from which to begin claim
interpretation. . . . Importantly, the person of ordinary skill in the
art is deemed to read the claim term not only in the context of the
particular claim in which the disputed term appears, but in the
context of the entire patent, including the specification.22
To determine how a person skilled in the art would have understood the meaning
of a disputed claim term, courts should look to publicly available sources,
including both intrinsic evidence — such as “‘the words of the claims themselves,
the remainder of the specification, the prosecution history’” — and “‘extrinsic
evidence concerning relevant scientific principles, the meaning of technical terms,
and the state of the art.’”23
A.

Intrinsic Evidence

21

Edward D. Manzo, Claim Construction in the Federal Circuit § 1.2
(2009). It is perhaps for this reason that the Federal Circuit’s reversal rate has
hovered between 20 and 45 percent over the last twenty years. See David L.
Schwartz, Pre-Markman Reversal Rates, 43 Loy. L.A. L. Rev. 1073, 1075, 109193 (2010).
22

Phillips, 415 F.3d at 1313 (citations omitted).

23

Id. at 1314 (quoting Innova/Pure Water, Inc. v. Safari Water
Filtration Sys., Inc., 381 F.3d 1111, 1116 (Fed. Cir. 2004)).
-10-

The Federal Circuit views intrinsic evidence as “a more reliable guide
to the meaning of a claim term than are extrinsic sources.”24 Thus, district courts
should use the intrinsic record as “the primary tool to supply the context for
interpretation of disputed claim terms.”25
1.

Claim Language

Although there is no rigid formula to claim construction, “[j]udicial
interpretation must begin with and remain focused upon the ‘words of the claims
themselves . . . to define the scope of the patented invention.’”26 On occasion, “the
ordinary meaning of claim language as understood by a person of skill in the art”
will be so apparent from the claim language itself that no further inquiry is
needed.27 However, even when the terms in a claim are not self-explanatory, “[t]he
context in which a term is used in the asserted claim can be highly instructive. To

24

Chamberlain Group, Inc. v. Lear Corp., 516 F.3d 1331, 1335 (Fed.
Cir. 2008) (citing Phillips, 415 F.3d at 1318-19).
25

V-Formation, Inc. v. Benetton Group SpA, 401 F.3d 1307, 1310 (Fed.
Cir. 2005) (citing Vitronics Corp. v. Conceptronic, Inc., 90 F.3d 1576, 1582 (Fed.
Cir. 1996)).
26

Investment Tech. Group, Inc. v. Liquidnet Holdings, Inc., No. 07 Civ.
510, 2010 WL 199912, at *2 (S.D.N.Y. Jan. 10, 2010) (quoting Vitronics, 90 F.3d
at 1582). Accord Haemonetics Corp. v. Baxter Healthcare Corp., 607 F.3d 776,
783 (Fed. Cir. 2010) (“[T]he claims perform the fundamental function of
delineating the scope of the invention . . . .”).
27

Phillips, 415 F.3d at 1314.
-11-

take a simple example, [the use of the term] ‘steel baffles’ . . . strongly implies that
the term ‘baffles’ does not inherently mean objects made of steel.”28 Thus, “the
most important indicator of the meaning of [a disputed claim term] is its usage and
context in the claim itself.”29
2.

The Specification

In addition to the claim itself, a patent is required by statute to
“contain a written description of the invention . . . .”30 This is often called the
specification,31 and it typically includes: “an abstract of the invention; a description

28

Id.

29

Middleton, Inc. v. Minnesota Mining and Mfg. Co., 311 F.3d 1384,
1387 (Fed. Cir. 2002).
30

35 U.S.C. § 112 (“The specification shall contain a written description
of the invention, and of the manner and process of making and using it, in such
full, clear, concise, and exact terms as to enable any person skilled in the art to
which it pertains, or with which it is most nearly connected, to make and use the
same, and shall set forth the best mode contemplated by the inventor of carrying
out his invention.”). The Federal Circuit has also explained that courts should rely
on intrinsic evidence because a person of ordinary skill in the field would use “the
patent specification and the prosecution history” to understand the invention
claimed by the patent. Multiform Desiccants, Inc. v. Medzam, Ltd., 133 F.3d 1473,
1477 (Fed. Cir. 1998). Accord Phillips, 415 F.3d at 1311-14.
31

See Medisim, 2011 WL 2693896, at *3 n.28 (“The terminology used
to describe the parts of a patent can be slightly confusing. Technically, the
specification includes both the claims and the written description. However, courts
typically use the term specification to refer to the written description on its own
and as distinct from the claims. For purposes of consistency, I adopt this common
usage.”).
-12-

of the invention’s background; a summary of the invention; patent drawings; and a
detailed description that discusses preferred embodiments of the invention.”32 The
Federal Circuit has said that specifications are “always highly relevant to the claim
construction analysis” and therefore “claims must be read in view of the
specification, of which they are a part.”33 However:
Adverse parties often hold significantly different views on how to
harmonize statements about the invention, its objects, or examples
of the invention in the “written description” with the claim
language, and probably the most difficult and frequently-litigated
issues of claim interpretation result from the inevitable differences
between statements in the written description as compared to the
words of the claims.34
Courts must be mindful that while “using the specification to interpret the meaning
of a claim” is permissible, “importing limitations from the specification into the
claim” is not.35 Because the words of the claim — and only the words of the claim
— “define the scope of the right to exclude,”36 the specification should generally

32

Id. at *3.

33

Phillips, 415 F.3d at 1315 (quotation marks and citations omitted). In
fact, a claim interpretation that excludes a preferred embodiment described in the
specification is “rarely, if ever, correct.” Vitronics, 90 F.3d at 1583.
34

Manzo § 2.44.

35

Phillips, 415 F.3d at 1323.

36

Renishaw PLC v. Marposs Societa’ Per Azioni, 158 F.3d 1243, 1248
(Fed. Cir. 1998).
-13-

be used to limit a claim only:
(1) if the claim “explicitly recite[s] a term in need of definition”;
or (2) if the specification unambiguously defines a term, i.e., if “a
patent applicant has elected to be a lexicographer by providing an
explicit definition in the specification for a claim term.”37
Even then, “there will still remain some cases in which it will be hard to determine
whether a person of skill in the art would understand the embodiments to define
the outer limits of the claim term or merely to be exemplary in nature.”38
3.

Prosecution History

The prosecution history of a patent is part of the “intrinsic evidence”
and “consists of the complete record of the proceedings before the PTO and
includes the prior art cited during the examination of the patent.”39 “[B]ecause the
prosecution history represents an ongoing negotiation between the PTO and the
applicant, rather than the final product of that negotiation, it often lacks the clarity
of the specification and thus is less useful for claim construction purposes.”40
Nonetheless, “[l]ike the specification, the prosecution history provides evidence of

37

Medisim, 2011 WL 2693896, at *4 (quoting Renishaw, 158 F.3d at

1248-49).
38

Phillips, 415 F.3d at 1323.

39

Id. at 1317 (citation omitted).

40

Id.
-14-

how the PTO and the inventor understood the patent,” and accordingly, “can often
inform the meaning of the claim language.”41 Further, despite the lack of clarity
inherent in the prosecution history, it “may be given substantial weight in
construing a term where that term was added by amendment.”42 This is because
changes made to the claim terms during the prosecution history “demonstrat[e]
how the inventor understood the invention and whether the inventor limited the
invention in the course of prosecution, making the claim scope narrower than it
would otherwise be.”43
B.

Extrinsic Evidence
The extrinsic record “‘consists of all evidence external to the patent

and prosecution history, including expert and inventor testimony, dictionaries, and
learned treatises.’”44 Previous district court opinions — such as this Court’s
preliminary injunction opinion — are also considered extrinsic evidence.45
41

Id.

42

Board of Regents of the Univ. of Tex. Sys. v. BENQ Am. Corp., 533
F.3d 1362, 1369 (Fed. Cir. 2008).
43

Phillips, 415 F.3d at 1317.

44

Id. (quoting Markman v. Westview Instruments, Inc., 52 F.3d 967, 980
(Fed. Cir. 1995)).
45

See MEMS Tech. Merhad v. ITC, 447 Fed. App’x 142, 153 (Fed. Cir.
2011) (“Related judicial holdings can be an appropriate form of non-binding
-15-

Although the Federal Circuit has explicitly “authorized district courts to rely on
extrinsic evidence,” it has also made clear that it views “extrinsic evidence in
general as less reliable than the patent and its prosecution history in determining
how to read claim terms.”46 This is for many reasons: (1) extrinsic evidence was
not created at the time of the patent for the purpose of explaining the patent’s scope
and meaning; (2) external publications “may not reflect the understanding of a
skilled artisan in the field of the patent”; (3) expert reports and testimony are
created at the time and for the purpose of litigation and may suffer from bias; (4)
“there is a virtually unbounded universe of potential extrinsic evidence of some
marginal relevance that could be brought to bear on any claim construction
question”; and (5) unlike intrinsic evidence, extrinsic evidence is not necessarily
part of the public record, and therefore undue reliance on it undermines the public
notice function of patents.47
Expert reports and expert testimony can be helpful to a court by:
[providing] background on the technology at issue, [explaining]
how an invention works, [ensuring] that the court’s understanding
of the technical aspects of the patent is consistent with that of a

extrinsic evidence in a claim construction analysis.”).
46

Phillips, 415 F.3d at 1317-18.

47

Id. at 1318-19.
-16-

person of skill in the art, or [establishing] that a particular term in
the patent or the prior art has a particular meaning in the pertinent
field.48
In contrast, it is unhelpful to a court if an expert “simply recites how [he or she]
would construe [a disputed term] based on his own reading of the specification.”49
Further, “a court should discount any expert testimony ‘that is clearly at odds with
the claim construction mandated by the claims themselves, the written description,
and the prosecution history, in other words, with the written record of the
patent.’”50
C.

Relevant Canons for Construing Claim Language
1.

Each Word Should Have a Distinct Meaning

Generally, “claims are interpreted with an eye toward giving effect to
all terms in the claim.”51 Thus, courts should avoid constructions which render
certain terms superfluous.52 And because “different claim terms are presumed to

48

Id. at 1318.

49

Symantec Corp. v. Computer Assocs. Int’l, Inc., 522 F.3d 1279, 1291
(Fed. Cir. 2008).
50

Phillips, 415 F.3d at 1318 (quoting Key Pharms. v. Hercon Labs.
Corp., 161 F.3d 709, 716 (Fed. Cir. 1998)).
51

Bicon, Inc. v. Straumann Co., 441 F.3d 945, 950 (Fed. Cir. 2010).

52

See Phillips, 415 F.3d at 1314.
-17-

have different meanings,”53 each word should be given a distinct meaning. Still,
although different words usually carry different meanings, any such presumption is
rebuttable.54
2.

Internal Consistency

Just as different words are presumed to have different meanings, the
same words are presumed to have the same meaning each time they appear, absent
special circumstances.55 This is true not just for words that appear multiple times,
but also for phrases that are used in a consistent fashion.56 This presumption is

53

Helmsderfer v. Bobrick Washroom Equip., Inc., 527 F.3d 1379, 1382
(Fed. Cir. 2008). Accord Applied Med. Res. Corp. v. U.S. Surgical Corp., 448 F.3d
1324, 1333 n.3 (Fed. Cir. 2006) (“[U]se of two terms in a claim requires that they
connote different meanings.”).
54

See Baran v. Medical Device Techs., Inc., 616 F.3d 1309, 1316 (Fed.
Cir. 2010) (construing two different terms to have the same meaning because
“[the] implication [that the use of different terms implies that they have different
meanings] is overcome where, as here, the evidence indicates that the patentee
used the two terms interchangeably”) (citing Tehrani v. Hamilton Med., Inc., 331
F.3d 1355, 1361 (Fed. Cir. 2003)).
55

See Rexnord Corp. v. Laitram Corp., 274 F.3d 1336, 1342 (Fed. Cir.
2001) (“[A] claim term should be construed consistently with its appearance in
other places in the same claim or in other claims of the same patent.”).
56

See Phonometrics, Inc. v. Northern Telecom Inc., 133 F.3d 1459, 1465
(Fed. Cir. 1998) (“A word or phrase used consistently throughout a claim should
be interpreted consistently.”).
-18-

likewise rebuttable,57 at least to give the same phrase a different meaning in
different claims (as opposed to giving the same phrase a different meaning in
different parts of the same claim).58
IV.

DISCUSSION
The first of the three phrases in dispute is “household level data

associated with multiple consumer households.” Kantar’s proposed construction
of the phrase is “household level data that has later been aggregated into a data set
including multiple consumer households.”59 With regard to clickstream and
purchase data, TRA proposes the following similar but distinct construction of the
disputed phrase: “data about the household that can later be aggregated into a data
set including multiple consumer households.”60 With regard to advertising and
program data, TRA proposes that the phrase should be construed as “data related to
advertising/programming tuned to by multiple consumer households.”61
57

See Felix v. American Honda Motor Co., Inc., 562 F.3d 1167, 117778 (Fed. Cir. 2009).
58

See Haemonetics, 607 F.3d at 783 (holding that a disputed phrase
consistently had the same meaning in one claim, irrespective of its different
meaning in a different claim).
59

Joint St. at 3.

60

TRA Reply Mem. at 9.

61

Id. at 9-10.
-19-

The second phrase at issue is “household level data.” Kantar argues
that this sub-phrase must be construed separately to address TRA’s overly broad
proposed construction of the longer phrase — which, according to Kantar,
eliminates the “household level” requirement.62 Kantar proposes that the phrase be
construed as “data from a single residential address or other individual consumer
location.”63 Although TRA believes that the Court need not separately construe
this sub-phrase, it proposes “data about a household.”64
The third dispute centers on step 4 of the claimed method, which
involves applying at least one “cleansing and editing algorithm” to the matched
and stored data. Kantar proposes that the phrase be construed as “an algorithm to
remove inconsistencies in, correct, or otherwise improve the reliability of data
collected from a program or delivery source,” whereas TRA proposes the
following construction: “[f]ormulas and calculations to remove inconsistencies in,
correct, or otherwise improve the reliability and/or utility of data.”65 For reference,
the disputes are summarized in the following table:

62

See Kantar Mem. at 18.

63

Joint St. at 3.

64

TRA Reply Mem. at 12.

65

Joint St. at 4.
-20-

Parties’ Disputed Constructions
Term

TRA’s Proposed Construction

Kantar Media’s
Proposed Construction

“household
level data
associated with
multiple
consumer
households”

When used with regard to
“clickstream” and “purchase” data:
data about a household that can be
later aggregated into a data set
including multiple consumer
households

household level data that
has later been
aggregated into a data
set including multiple
consumer households

When used with regard to
“advertising” and “program” data:
data related to
advertising/programming tuned to
by multiple consumer households
“household
level data”

TRA does not believe that this
phrase requires a separate
definition. However, if it must be
given an out-of-context definition,
it asks that the phrase be defined
as: data about a household.

data from a single
residential address or
other individual
consumer location

“cleansing and
editing
algorithm”

Formulas and calculations to
remove inconsistencies in, correct,
or otherwise improve the reliability
and/or utility of data

an algorithm to remove
inconsistencies in,
correct, or otherwise
improve the reliability of
data

-21-

A.

“Household Level Data Associated with Multiple Consumer
Households” When Used with Regard to Clickstream and
Purchase Data
With regard to clickstream and purchase data, the parties do not

entirely agree on the appropriate construction of the disputed phrase. The parties
agree that in both of those contexts, the phrase refers to data collected from
individual households. However, Kantar’s proposed construction refers to “data
that has later been aggregated,” whereas TRA’s definition refers to data “that can
be later aggregated.” The distinction is entirely temporal. TRA argues that the
method in the claim has distinct sequential steps, and that “Kantar’s definition
improperly collapses the collecting step and the matching step by requiring the
data to be aggregated before the [collecting step] is completed.”66
Kantar’s proposed construction — which includes the phrase “data
that has later been aggregated” — is somewhat ambiguous, but it does imply that
the data must be aggregated before it is matched. TRA’s proposed construction
has the advantages of being not only less ambiguous but also more consistent with
the step-by-step sequence implied by the claim’s paragraph structure. Thus, with
regard to clickstream and purchase data, I adopt TRA’s proposed construction of
“household level data associated with multiple consumer households” as “data
66

TRA Reply Mem. at 10.
-22-

about a household that can be later aggregated into a data set including multiple
consumer households.”
B.

“Household Level Data Associated with Multiple Consumer
Households” When Used with Regard to Advertising and
Program Data
1.

The Claim Language

The parties agree with this Court’s prior holding that the ordinary
meaning of “household level data associated with multiple consumer households”
involves data from individual households.67 TRA argues, however, that when used
in the context of advertising and program data, a person of ordinary skill in the art
would understand the disputed phrase to refer to general data regarding which
programs and advertisements were playing on a channel at a given time — not data
from individual households.68 Under Kantar’s proposed construction, the claimed
method covers only the collection of individual household data on program and
advertising viewership directly from individual homes, not the collection of
general programming and advertising data from advertisers and content-providers.
67

See TNS Media Research, 2011 WL 4425415, at *5; Kantar Mem. at
12; Defendant TRA Global, Inc.’s Claim Construction Brief (Markman Hearing)
(“TRA Mem.”), at 8; Fenwick Decl. at 7.
68

General programming and advertising data can be matched with
clickstream data to then determine which programs and ads an individual
household watched, and indeed, this is what both MTRA and RVR do.
-23-

TRA argues that the disputed phrase should be interpreted in light of
the full and distinct contexts in which it appears. In particular, TRA highlights the
contextual clues from the language of the patent as to the source of the various
types of data:
collecting in an advertising measurement system:
(i) clickstream data from a program delivery source of a
consumer, wherein collecting the clickstream data is not
dependent on a supplemental data collection device, and
also wherein the collected clickstream data includes
household level data associated with multiple consumer
households;
(ii) advertising data associated with delivery of the program by
the program delivery source, wherein collecting the
advertising data is not dependent on a supplemental data
collection device, and also wherein the collected
advertising data includes household level data associated
with multiple consumer households;
(iii)

program data associated with the program delivered on the
program delivery source, wherein collecting the program
data is not dependent on a supplemental data collection
device, and also wherein the collected program data
includes household level data associated with multiple
consumer households;

(iv) purchase data from a purchase data source, wherein
collecting the purchase data is not dependent on a
supplemental data collection device, and also wherein the
collected purchase data includes household level data
associated with multiple consumer households;69

69

‘940 Patent col. 46 ll. 36-60 (emphasis added).
-24-

Whereas the clickstream and purchase data are described as coming “from a” type
of source, the advertising and program data are referred to as being “associated
with . . . the program delivery source.”70 TRA argues that this distinction makes all
the difference, and that by using the words “associated with” instead of the word
“from,” the claim contemplates that the advertising and program data could come
not only from program delivery sources such as digital set-top boxes (“DSTBs”),
but also “‘from the television networks or the advertisers seeking to make use of
the proposed system.’”71 According to TRA, this is the only construction of the
full claim language that makes sense, and “‘claim language should be read in a
manner that causes the claim to make sense.’”72 Further, “different claim terms are

70

The parties made the following stipulations: (1) “clickstream data
from a program delivery source of a consumer” should be construed as “data
describing a consumer’s exposure to content delivered from a program delivery
source”; (2) “advertising/program data associated with delivery of the program by
the program delivery source” should be construed as “data describing
advertisements/media content delivered from a program delivery source”; and (3)
“purchase data from a purchase data source” should be construed as “data
describing the purchase of a particular product at a given time, obtained from a
purchase data source, such as a shopping loyalty card, point of sale collection
means, or other record of a sale of a product or service.” Joint St. at 2.
71

TRA Reply Mem. at 10-11 (quoting TNS Media Research, 2011 WL
4425415, at *4).
72

Brassica Prot. Prods. LCC v. Caudill Seed & Warehouse Co., Inc.,
591 F. Supp. 2d 389, 395 (S.D.N.Y. 2008) (quoting Joao v. Sleepy Hollow Bank,
418 F. Supp. 2d 578, 581 (S.D.N.Y. 2006)).
-25-

presumed to have different meanings,”73 and Kantar’s proposed construction fails
to address the differing language used to describe the source of each type of data.
Kantar argues that TRA’s proposed construction of the disputed
phrase violates the claim construction canon that “[t]he words of a claim are
generally given their ordinary and customary meaning,” unless either “a patentee
sets out a definition and acts as his own lexicographer,” or “the patentee disavows
the full scope of a claim term either in the specification or during prosecution.”74
Kantar argues that neither exception applies, and that the disputed phrase must
therefore be given its ordinary meaning in all contexts. However, the Federal
Circuit has explained that “the most important indicator” of a term’s meaning — as
would be understood by a person skilled in the art — is the “usage and context in
the claim itself.”75 The disputed phrase is used four times to describe four different
types of data. Each description must be read as an integrated whole to determine
what type of data is being described, and each description begins by indicating the
source of the data. The clickstream and purchase data descriptions begin
“clickstream/purchase data from a [program delivery/purchase data] source”
73

Helmsderfer, 527 F.3d at 1382.

74

Thorner v. Sony Computer Entm’t Am. LLC, 669 F.3d 1362, 1365
(Fed. Cir. 2012).
75

Phillips, 415 F.3d at 1314.
-26-

whereas the advertising and programming descriptions begin
“advertising/programming data associated with [delivery of the program by/the
program delivered on] the program delivery source.”76 These phrases inform how
the disputed phrase should be read77 and TRA’s proposed construction makes sense
in that it varies to match the different descriptions of the sources of the data, while
Kantar’s does not.78 Thus, I find that in the full context of the claim language, one
skilled in the art would ordinarily understand the disputed phrase to have different

76

‘940 Patent col. 46 ll. 37-51 (emphasis added).

77

See Paragon Solutions, LLC v. Timex Corp., 566 F.3d 1075, 1088
(Fed. Cir. 2009) (construing the phrase “real-time data” — used to describe several
types of data — in the context of the invention and the surrounding claim
language).
78

Kantar argues, that by failing to give a distinct effect to the subphrases “household level data” and “associated with multiple consumer
households,” TRA’s construction violates the canon that each word or phrase
should have a role to play. See Bicon, 441 F.3d at 950. Yet Kantar’s proposed
construction also fails to give a role to each word and phrase — Kantar proposes
that the sub-phrase “household level data” be construed as “data from a single
residential address or other individual consumer location.” Joint St. at 3. Were
this construction substituted into the claim language, the full text of the description
of clickstream data would read “clickstream data from a program delivery source
of a consumer, wherein collecting the clickstream data is not dependent on a
supplemental data collection device, and also wherein the collected clickstream
data includes data from a single residential address or other individual consumer
location associated with multiple consumer households.” Thus, Kantar’s proposed
construction would insert a second description of the source of the data, making
the claim language redundant. See Phillips, 415 F.3d at 1314.
-27-

meanings in different contexts.79
2.

The Specification

The specification describes the source of program data as follows:
Program data supplied to the advertising measurement system
may originate from cable operators, advertising media firms, and
other third-party data providers who communicate daily or
periodic data feeds of scheduled television program data, for
example. Such data may identify what was scheduled to air on
each television channel.80
Likewise, advertising data is described in the specification as coming directly from
television networks or the advertisers seeking to make use of the proposed
system.81 Another embodiment in the specification describes assessing the
effectiveness of “addressable” commercials — commercials targeted and sent to
the DSTBs of a small subset of households within a market.82 While these
embodiments describe a system in which neither program nor advertising data are
collected on the household-level, at least one embodiment describes a household-

79

Although phrases are presumed to have a consistent definition each
time they appear, see Rexnord, 274 F.3d at 1342, TRA has rebutted the
presumption by demonstrating that the contexts in which the phrases appear are
distinct.
80

‘940 Patent col. 20 ll. 17-22.

81

See id. col. 20 ll. 29-42.

82

Id. col. 9 ll. 48-67.
-28-

based device communicating “program viewing data” to the system.83
Although a court may not “[import] limitations from the specification
into the claim,” it is appropriate to “[use] the specification to interpret the meaning
of the claim.”84 TRA’s proposed construction of the disputed phrase is: “data
related to advertising/programming tuned to by multiple consumer households.”
This construction is sufficiently broad to include all of the embodiments in the
specification. Kantar’s proposed construction would construe the claim so as to
exclude the collection of program and advertising data from third-party providers;
the Federal Circuit has stated that claim interpretations that exclude the preferred
embodiment described in the specification are “rarely, if ever, correct.”85
Kantar argues that TRA is attempting to use the specification to give
the disputed phrase a different meaning than its ordinary one, and that the
specification may only do so if it “clearly set[s] forth” a special definition.86 While
the specification does not act as “a lexicographer by providing an explicit

83

See ‘940 Patent col. 7 ll. 1-5 (“[P]rogram viewing data . . . may be
communicated from the DSTB to an advertising measurement system configured
to receive and process such program data.”).
84

Phillips, 415 F.3d at 1323.

85

Vitronics, 90 F.3d at 1583.

86

Thorner, 669 F.3d at 1365.
-29-

definition” of the disputed phrase, it both “unambiguously defines” the various
types of data and lists the potential sources of program and advertising data.87
Thus, I find that TRA’s proposed construction of the disputed phrase is more
consistent with the specification than Kantar’s.
3.

The Prosecution History

Claim 1 of TRA’s original patent application describes the data
collection step of the claimed method as follows:
[R]eceiving at least advertising data, clickstream data, and
program data in an advertising measurement system, wherein at
least the clickstream data are obtained from a program delivery
source located in a household of a consumer, wherein the program
delivery source is not dependent on a supplemental data collection
device in the household for data collection.88
In June 2009, the PTO rejected the claims containing the above language, finding
them to have been anticipated by another patent application which described
“receiving at least advertising data, clickstream data, and program data in an
advertising measurement system.”89 TRA responded by revising its claim
language — structuring the collection step in a manner that separately described
87

Renishaw, 158 F.3d at 1248-49.

88

3/7/12 Declaration of Charles Steenburg in Support of Plaintiffs’
Opening Claim Construction Statement (“Steenburg Decl.”), Ex. B at 78.
89

Id., Ex. C at TRA 0000351.
-30-

the four different types of data and for the first time adding the phrases “from a,”
“associated with” and “household level data associated with multiple consumer
households.”90 The result became the ‘940 patent.
Kantar argues that by clarifying that the claimed method would collect
“household level data associated with multiple consumer households,” TRA
narrowed its claim and added limitations to which it should now be bound. The
Federal Circuit has held that the prosecution history merits “substantial weight”
when construing terms which were added by amendment, as the disputed phrase
was.91 Disregarding such additional language “would render claim examination in
the PTO meaningless.”92
TRA agrees that a proper construction will exclude interpretations
disavowed during prosecution. However, TRA asserts that “the disavowal must be
both clear and unmistakable to one of ordinary skill in the art.”93 Here, the
patentee’s amendment to the claim deleted 58 words and added 335 words.94 TRA
90

See Tr. 67:2-14.

91

Board of Regents, 533 F.3d at 1369-70.

92

Panduit Corp. v. Dennison Mfg. Co., 810 F.2d 1561, 1576 (Fed. Cir.

1987).
93

Liquidnet, 2010 WL 199912, at *4.

94

See TRA Reply Mem. at 5.
-31-

argues that these sweeping changes do not demonstrate any clear and unmistakable
disavowal — that because the claim language bears little resemblance to the
originally submitted claim, the prosecution history sheds little light on the meaning
of terms for claim construction purposes. I agree. The prosecution history reveals
that the PTO found the initial patent application to lack sufficient specificity, and
indeed, the approved patent describes the types of data to be collected in far more
detail than the denied application, thereby adding limitations to win the PTO’s
approval. However, given the extent to which the claim language was rewritten, it
is unclear whether the added limitations are those that Kantar suggests. Because
the “from” and “associated with” language was only added in the final and
approved claim language,95 it is plausible that TRA’s proposed construction was
both sufficiently specific to win the PTO’s approval and in accordance with the
PTO’s understanding of the claim language. Thus, the prosecution history does
not provide clear guidance as to how the disputed phrase should be construed.96
4.

Extrinsic Evidence
a.

95

The Preliminary Injunction Decision

See Tr. 67:2-14.

96

See Phillips, 415 F.3d at 1317 (“[The prosecution history] often lacks
the clarity of the specification and thus is less useful for claim construction
purposes.”).
-32-

Although “[r]elated judicial holdings can be an appropriate form of
non-binding extrinsic evidence in a claim construction analysis,”97 I am now
conducting a full claim construction analysis, taking into consideration all of the
facts and law presented at the Markman hearing and in the claim construction
briefing. As such, I have no need to rely on my earlier decision.
b.

Expert Testimony of Richard Fenwick

TRA has submitted a declaration from Richard Fenwick, its expert
witness, who also testified at the Markman hearing. Fenwick is well-qualified to
opine on the technology at issue,98 and indeed, Kantar cites to statements made by
Fenwick to support its assertion that collection of program and advertising data
from the DSTBs is theoretically possible.99 At the Markman hearing, Fenwick
testified as follows:
THE COURT: [Is it] also possible to get [program data] from the
set top box in the home?
[FENWICK]: It’s certainly possible to do that. It would require
a set top box with some intelligence that knows channel 2 is
playing The Simpsons right now.
THE COURT: Does that exist?

97

MEMS Tech., 447 Fed. App’x at 153.

98

See Fenwick Decl., Ex. A.

99

See Kantar Reply Mem. at 8.
-33-

[FENWICK]: I’m not aware of any that exist.
THE COURT: You’re still saying it’s not possible to get that data
from the set top box today?
[FENWICK]: I don’t know whether anybody is doing that or not.
It is certainly possible. Whether anybody is doing it or not, I do
not know. . . . I am aware of no [application as of yet] —
DirecTV, Comcast cable, they don’t do it.100
Both sides ultimately argue that it is irrelevant whether or not Kantar’s proposed
construction describes a device that is theoretically possible.101 The Federal Circuit
has repeatedly held that it is impermissible to redraft claims during litigation even
where the literal scope of those claims makes them inoperable in practice.102
Kantar argues that it would be improper for this Court to consider
Fenwick’s opinions on the appropriate construction of the disputed phrase because:
(1) the intrinsic evidence is so clear that extrinsic evidence such as expert
testimony need not be considered;103 and (2) Fenwick’s declaration and testimony

100

Tr. at 64:4-23.

101

See Kantar Reply Mem. 11-13; TRA Reply Mem. at 15-16.

102

See Chef Am. v. Lamb-Weston, Inc., 358 F.3d 1371, 1373-74 (Fed.
Cir. 2004) (holding that claim language that required heating “dough to a
temperature in the range of about 400º F. to 850º F.” meant exactly what the
ordinary meaning suggested even though all agreed that heating dough itself —
rather than the oven — to that temperature would burn the dough to a crisp and
leave it as inedible as a “charcoal briquet”).
103

See Phillips, 415 F.3d at 1317-18 (“We have viewed extrinsic
evidence in general as less reliable than the patent and its prosecution history in
-34-

are not helpful to the Court because they merely set forth his opinions on how the
claim terms should be construed.104 Because I agree with both arguments, I did not
consider Fenwick’s opinion regarding the appropriate construction of “household
level data associated with multiple consumer households.”
5.

Claim Construction

With regards to program and advertising data, I adopt TRA’s
proposed construction of “advertising/programming tuned to by multiple consumer
households.” This construction is the correct one because, as set forth above: (1) it
is the most consistent with the claim language; (2) it is supported by the
specification; and (3) the prosecution history is ambiguous.105
B.

“Household Level Data”
Whereas Kantar requests that this Court construe the sub-phrase

“household level data,” TRA argues that the claim language should not be parsed

determining how to read claim terms.”).
104

See Symantec, 522 F.3d at 1291 (clarifying that it is unhelpful to a
court if an expert “simply recites how [he or she] would construe [a disputed term]
based on his own reading of the specification”).
105

Because I find the intrinsic evidence as to the appropriate construction
of this phrase to be sufficiently clear, I need not consider the extrinsic evidence.
-35-

out into sub-phrases.106 Because my construction of the longer phrase has already
resolved all relevant questions as to the meaning of the claim words, I decline to
construe the sub-phrase.
C.

“Cleansing and Editing Algorithm”
Under the claimed method, once the various types of data are

collected, stored, and matched, at least one “cleansing and editing algorithm” is
applied to the matched and stored data.107 Following the “cleansing and editing” of
the data, at least one “true target index metric based on the matched and stored
data” is calculated.108 Thus, the purpose of these algorithms is to prepare the data
for the calculation step. The parties’ dispute centers on whether the word “utility”
has any place in a proper description of what a “cleansing and editing algorithm”

106

See Liquidnet, 2010 WL 199912, at *13 (declining to construe subphrases when the construction of the larger phrases had already resolved all
relevant questions as to the meaning of the claim words).
107

During the preliminary injunction phase, TRA argued that “applying
at least one cleansing and editing algorithm to the matched and stored data” did not
impose a sequential limitation regarding when the data must be matched and
stored. See Reply Memorandum in Support of Preliminary Injunction Against
Plaintiff/Counterclaim-Defendant TNS Media Research, LLC d/b/a Kantar Media
Audiences, at 7-8. I rejected that argument, explaining that TRA’s suggestion “did
violence to the ordinary grammatical understanding of the past tense.” TNS Media
Research, 2011 WL 4425415, at *5.
108

‘940 Patent col. 48 ll. 8-9.
-36-

does. In its original claim construction brief, Kantar proposed that the phrase be
construed as “an algorithm to remove inconsistencies in, correct, or otherwise
improve the reliability of data from a program delivery source.” At TRA’s
request, Kantar dropped the “from a program delivery source” requirement “to
promote judicial economy and narrow the claim construction issues in dispute.”109
1.

Claim Language

Kantar argues that the word “utility” is so broad that TRA’s proposed
construction encompasses “virtually anything.”110 Kantar defines utility as “the
condition or quality of being useful,”111 and argues that because anyone who
applies an algorithm to data “plainly” does so to make the data more useful, TRA’s
proposed construction makes the phrase “cleansing and editing” superfluous.112
Kantar further argues that TRA’s construction is so broad that it would include the
generation of reports and the calculation of metrics,113 and that such a construction
109

Kantar Mem. at 20 n.38.

110

Id. at 24.

111

The American Heritage Dictionary, Second College Edition,
Steenburg Decl., Ex. K.
112

Kantar Mem. at 21.

113

In his deposition, Fenwick — TRA’s expert — explicitly agreed with
Kantar’s assertion that TRA’s proposed construction of “applying [a] cleansing
and editing algorithm” could include the creation of a true target index and return
-37-

cannot be correct because the generation of reports and the calculation of metrics is
described as a separate and distinct step in the claimed method.
TRA responds that “utility” is not an overly broad word and that it can
be defined as “actionability for the user’s intended purposes,”114 with the user’s
intended purpose being defined by the rest of the claim. Even were “cleansing and
editing” given a broad definition, “[t]he patentee is free to choose a broad term and
expect to obtain the full scope of its plain and ordinary meaning unless the patentee
explicitly redefines the term or disavows its full scope.”115 TRA also argues that
Kantar’s proposed construction improperly limits the claim to covering only those
algorithms which improve the “reliability” of data — and that the word
“reliability” (or “reliable) is found nowhere in the specification.116

on investment reports. See Steenburg Decl., Ex. J at 36-39. Fenwick testified
similarly at the Markman hearing. See Tr. 91:14-16 (“I believe that part of
generating various reports can utilize editing algorithms as well as cleansing
algorithms.”).
114

TRA Reply Mem.. at 18.

115

Thorner, 669 F.3d at 1367.

116

TRA Reply Mem. at 17. Even though the word “reliable” does not
appear in the specification, the parties agreed that a portion of the definition should
reference reliability based on the intrinsic evidence. See Fenwick Decl. at ¶¶ 3844. TRA’s argument is that while “reliability” does have a proper place in the
definition of a “cleansing and editing algorithm,” the phrase should be construed to
cover algorithms which improve both the reliability and the utility of data.
-38-

2.

Specification

The specification refers to “algorithms” as “formulas and
calculations.”117 According to both proposed constructions, the following
algorithms which are described in the specification can properly be considered
“cleansing and editing” algorithms:
!

An algorithm that corrects for the “false positive” problem of data
collected from a set-top box that remains on even though the
associated television has been turned off.118

!

An algorithm that determines whether to exclude a particular
household’s data entirely because it is “insufficiently complete” —
due to, for example, cable malfunctions.119

!

An algorithm that ensures that a household containing multiple
DSTBs is not counted more than once.120

The distinction between the parties’ proposed construction is that TRA’s proposed

117

TRA Mem. at 11 (citing ‘940 Patent col. 4 ll. 51-56, col. 27 ll. 44-45,
col. 28 ll. 23-29, col. 38 ll.1-3).
118

‘940 Patent col. 5 ll. 43-46.

119

Id. col. 5 ll. 46-49.

120

See id. col. 14 ll. 43-47.
-39-

construction would also include an algorithm described in the specification as
“geographic and demographic weighing.” Such a weighting algorithm would not
improve the reliability of the data, but it would improve the data’s utility.
3.

Prosecution History

The claim language in TRA’s original application referred only to
“calculating at least one metric based on the cross-correlated data” — it did not
include the step of applying algorithms to the data (nor did it mention the word
“algorithm” at all).121 This was rejected by the PTO as anticipated by prior art.122
TRA amended the claim to require applying an “algorithm” to “facilitate” the
calculation of the metric.123 This, too, was rejected by the PTO, which then
discussed with TRA possible amendments to “specify the applied algorithm and
the calculated metric.”124 The “cleansing and editing” language was added in
TRA’s subsequent revision and incorporated into the approved patent.
4.

Claim Construction

A person of ordinary skill in the art would not apply an algorithm to

121

Steenburg Decl., Ex. B.

122

See id., Ex. C.

123

Id., Ex. D.

124

Id., Ex. E at 2.
-40-

data unless doing so served a purpose. Thus, the ordinary meaning of an
“algorithm” implies something that it is used to improve the utility of data. The
words “cleansing and editing” must be understood as words of limitation, which
limit the patent to algorithms which improve the data’s utility in specific ways.
Kantar’s proposed construction captures this by including only algorithms that
improve the reliability of data. In contrast, TRA’s proposed construction does not
give any independent meaning to the words “cleansing and editing.” Moreover, by
proposing a construction so broad as to include the calculation of metrics, TRA
ignores the context of the full claim, which sets out calculation as a distinct and
separate step.
TRA’s proposed construction would also render meaningless the
amendments TRA added after the PTO rejected the original versions of the claim
language. In making its amendments and incorporating the “cleansing and editing”
language, TRA gave notice that only the application of “cleansing and editing”
algorithms were covered by its patent. TRA should not now be permitted to adopt
a construction that undoes this prosecution history. Thus, I adopt Kantar’s
proposed construction of “cleansing and editing algorithm” as “an algorithm to
remove inconsistencies in, correct, or otherwise improve the reliability of data
collected from a program delivery source.”

-41-

V.

ADDENDUM
In my preliminary injunction ruling, based on some ambiguous

wording in a declaration, I incorrectly implied that Kantar’s DirectView was prior
art.125 At the Markman hearing, TRA asserted that this was in error, as the patent
application predated DirectView, which launched in 2009.126 Kantar concedes that
this was “an inaccuracy.”127 Although I am not revisiting my preliminary
injunction ruling, I now clarify for the record that Kantar’s DirectView was not
prior art.
VI.

CONCLUSION
For the aforementioned reasons I construe the disputed terms within

claim one as follows:
When used with regard to clickstream and purchase data, “household
level data associated with multiple consumer households” means “data about a
125

See 2011 WL 4425415, at *5 (citing 8/2/11 Declaration of George
Shabab, President of Kantar, in Opposition to TRA’s Motion for Preliminary
Injunction [Docket No. 35] ¶ 22).
126

See Tr. at 95:20-97:7.

127

See 8/27/12 Letter from Michael A. Albert, counsel for Kantar, to the
Court, at 1. Kantar argues that the statement “one of Kantar’s own products taught
[a specific prior art],” 2011 WL 4425415, at *5, was accurate and that only its
citation was in error. As I am not revisiting my prior ruling, I see no reason to
consider whether a different Kantar product taught the relevant prior art.
-42-

household that can be later aggregated into a data set including multiple consumer
households."
When used with regard to advertising and program

data~

"household

level data associated with multiple consumer households" means "data related to
advertising/programming tuned to by multiple consumer households."
"Cleansing and editing algorithm" means "an algorithm to remove
inconsistencies in, correct, or otherwise improve the reliability of data collected
from a program delivery source."
A status conference is scheduled for September 19,2012 at 3:00 pm.

Dated:

August 29, 2012
New York, New York

-43­

- Appearances For Plaintiffs/Counterclaim-Defendants:
Marc Joseph Rachman, Esq.
Christopher Andrew Keisner, Esq.
Davis & Gilbert, LLP
1740 Broadway
New York, NY 10019
[phone]
Charles Thomas Steenburg, Esq.
John L. Strand, Esq.
Michael A. Albert, Esq.
Robert M. Abrahamsen, Esq.
Wolf, Greenfield, & Sacks, P.C.
600 Atlantic Avenue
Boston, MA 02210
[phone]
For Defendant/Counterclaim-Plaintiff:
Shireen Anneke Barday, Esq.
John Paul Del Monaco, Esq.
Jay Philip Lefkowitz, Esq.
Kirkland & Ellis LLP
601 Lexington Ave
New York, NY 10022
[phone]
Richard P. Doyle, Esq.
Jaime B. Herren, Esq.
Janssen Doyle LLP
140 Brookwood Road
Suite 102
Orinda, CA 94653
[phone]

-44-

Matthew Russell Maron, Esq.
Tannenbaum Helpern Syracuse & Hirschtritt LLP
900 Third Avenue
New York, NY 10022
[phone]

-45-