I/P Engine, Inc. v. AOL, Inc. et al

Filing 512

Declaration re 511 Opposition, of Howard Chen in Support of Defendants' Opposition to Plaintiff's Daubert Motion, and Fourth Motion in Limine to Exclude Lyle Ungar's New Theory of Invalidity and Opinions Regarding Claim Construction by AOL Inc., Gannett Company, Inc., Google Inc., IAC Search & Media, Inc., Target Corporation. (Attachments: # 1 Exhibit 1, # 2 Exhibit 2, # 3 Exhibit 3, # 4 Exhibit 4, # 5 Exhibit 5, # 6 Exhibit 6, # 7 Exhibit 7, # 8 Exhibit 8, # 9 Exhibit 9, # 10 Exhibit 10, # 11 Exhibit 11, # 12 Exhibit 12, # 13 Exhibit 13)(Noona, Stephen)

EXHIBIT 11

Page 1
UNITED STATES DISTRICT COURT
EASTERN DISTRICT OF VIRGINIA
NORFOLK DIVISION
---------------------------------x
I/P ENGINE, INC.,
Plaintiff,
v.

Civil Action No.:
2:11-cv-512

GOOGLE INC.,
Defendant.
---------------------------------x
CONFIDENTIAL - ATTORNEYS' EYES ONLY
Videotaped 30(b)(6) Deposition
of
JAIME G. CARBONELL, Ph.D.
Washington, D.C.
Friday, September 21, 2012
9:04 a.m.

Reported by:

Amy E. Sikora, RPR, CRR, CSR-NY, CLR

Job No. CS1338951

Veritext Corporate Services
800-567-8658

973-410-4040
9349fef0-7eb5-423b-bd0b-89d5688fb872

Page 32
1

Q.

Do you have any estimate at all?

2

A.

I can give you an estimate, but I

3

cannot guarantee its accuracy.

4

Q.

Understood.

5

A.

It would be between one and 200 hours.

6

Q.

Okay.

7
8

MR. NELSON:

Let's mark this next one

here.

9

(Carbonell Exhibit No. 4, copy of U.S.

10

Patent No. 6,006,222, marked for identification

11

as of this date.)

12

Q.

Okay.

I'm showing you now what I've

13

had marked as Carbonell Deposition Exhibit No. 4.

14

It is a -- United States Patent No. 6,006,222.

15

Do you see that?

16

A.

Yes.

17

Q.

And the named inventor is Gary

18

Culliss.

Do you see that?

19

A.

I see that.

20

Q.

I think -- well, actually, let me just

21

ask that first.

22

before?

23

A.

Yes, I have.

24

Q.

Do you recall about how long you spent

25

Have you reviewed Exhibit No. 4

reviewing it?
Veritext Corporate Services

800-567-8658

973-410-4040
9349fef0-7eb5-423b-bd0b-89d5688fb872

Page 33
1

A.

I reviewed the asserted prior art as a

2

group as well as individually.

3

to say collectively, I mean, how my time

4

distributed among the various items of prior art.

5

Q.

Okay.

So it's difficult

But do you feel like you have a

6

good understanding of what we've marked here as

7

Carbonell Deposition Exhibit No. 4?

8
9
10

A.

I have read it more than once.

Yes,

the answer to your question.
Q.

Okay.

Now, if you look to page 24 of

11

your report that we marked as Exhibit No. 1, you

12

have a section that's headed "XII."

13

Do you see that?

14

A.

Yes.

15

Q.

And this section concerns what we've

16

marked as Carbonell Deposition Exhibit No. 4;

17

correct?

18

A.

Correct.

19

Q.

Okay.

The -- here in parentheticals

20

and throughout this section of your report, you

21

refer to it as "the Culliss reference"; is that

22

right?

23

A.

Just as "Culliss."

24

Q.

Okay.

25

So I'll -- is it okay for the

deposition, if I just refer to it as Culliss,
Veritext Corporate Services

800-567-8658

973-410-4040
9349fef0-7eb5-423b-bd0b-89d5688fb872

Page 34
1

then, so that we use the same terminology used in

2

your report?

3

A.

Yes.

4

Q.

Okay.

Now, in paragraph 102, you

5

express a summary of your opinions concerning the

6

Culliss reference with respect to the '420

7

patent; correct?

8

A.

Correct.

9

Q.

And there are two elements that you

10

identify in -- let's just start with independent

11

claim 10 that you believe are not found in the

12

Culliss reference; correct?

13

A.

Correct.

14

Q.

Okay.

Now, with respect to inde --

15

independent claim 10, at least, of the '420

16

patent, and you have it there in front of you, if

17

you want to look at it, in your report you didn't

18

express any opinions concerning elements of

19

claim 10 beyond the two that you addressed here

20

or you list here in paragraph 102; correct?

21

A.

That's correct.

22

Q.

So, then --

23

A.

Let me -- let me put claim 10 up here

24
25

so that I can refresh my memory.
Q.

Yup.
Veritext Corporate Services

800-567-8658

973-410-4040
9349fef0-7eb5-423b-bd0b-89d5688fb872

Page 81
1

Q.

Okay.

Well, let's -- let's -- let's

2

go back to column 3 towards the bottom, about

3

line 61 for a moment.

4

You see it says, "These articles" --

5

that are given the generic labels A1, A2, A3.

6

You see that?

7

A.

Right.

8

Q.

"These articles are each associated

9

with one or more of these key terms by any

10

conceivable method of association, such as

11

through indexing all words or through meta tag

12

headers containing keywords selected by the

13

author or the editor."

14

A.

Right.

15

Q.

Okay.

16

A.

Yes, I see that.

17

Q.

So you agree that one of the ways that

You see that?

18

Culliss teaches that the key terms can be

19

selected is by indexing words in the article;

20

correct?

21

A.

22

tables.

23

Q.

That's how we would initialize the
Okay.

Now, if we go over to column 4,

24

about line 10, it says, "The invention will

25

accept a search query from a user, and a search
Veritext Corporate Services

800-567-8658

973-410-4040
9349fef0-7eb5-423b-bd0b-89d5688fb872

Page 82
1

engine will identify key terms which match the

2

search query"?

3

A.

Yes.

4

Q.

"These key terms which match the

5

search query are called matched key terms.

6

search engine then identifies in any conceivable"

7

matter -- "manner, the articles which are

8

associated with the matched key terms."

9

The

Do you see that?

10

A.

I see that.

11

Q.

So do you understand that to mean that

12

the key term scores are used by the search engine

13

to do that matching in response to the search

14

query that we just looked at at the top of

15

column 4?

16

A.

I do not know.

If the key term scores

17

reach zero or some very small number, it would be

18

normal to -- they would be ranked very low, and

19

Culliss could return them or could not return

20

them.

21

has an entry in the table, then the scores are

22

not material.

23

it.

24
25

If Culliss were to return everything that

Q.

I do not know which way he does

Okay.

So you don't know which -- one

way or the other is what you're saying?
Veritext Corporate Services

800-567-8658

973-410-4040
9349fef0-7eb5-423b-bd0b-89d5688fb872

Page 83
1

A.

Yeah.

2

Q.

Okay.

3

A.

That's why I can't agree when you say

4

he does it one specific way.

5

know.

6

Q.

Okay.

7

I don't -- I don't

Understood.

So now, let's go back to your report,

8

which we've marked as Exhibit 1, and look at

9

paragraph 108.

There you say, "Third, Culliss

10

does not disclose filtering each informon for

11

relevance to the query or filtering the combined

12

information for relevance to at least one of the

13

query and the first user."

14

Do you see that?

15

A.

I see that.

16

Q.

And then further down, you say that --

17

and I'm paraphrasing here because I don't want to

18

just read the whole thing -- that Culliss

19

discloses ranking, but not filtering; correct?

20

A.

That's correct.

21

Q.

Okay.

So can you explain to me what

22

you mean by the difference between ranking and

23

filtering?

24

A.

Yes.

25

Q.

Please do.
Veritext Corporate Services

800-567-8658

973-410-4040
9349fef0-7eb5-423b-bd0b-89d5688fb872

Page 84
1

A.

Ranking is putting a set of elements,

2

whether they be documents or other items, in

3

order.

4

from a score, a relevant score.

5

popularity score.

6

criteria.

That order would typically be derived

7

It can be a

It can be based on other

So ranking is -- is essentially you

8

start with a set, and you end up with an ordered

9

set.

Same set in order.

10

Filtering is the process where you

11

examine elements of a set one at a time, and you

12

determine whether or not they qualify according

13

to some filtering criterion.

14

Then that set is divided into two

15

sets, one of which is the filtered or accepted

16

set, one of which is the rejected or filtered out

17

set.

18

Q.

So let me have an example.

If you

19

have a -- a criterion that says, I want to review

20

the top 10 something, then you rank whatever the

21

candidates are, and you display the top 10, is

22

that ranking or is that filtering?

23

A.

That is ranking.

24

Q.

Okay.

25

Even though you used a

criterion to decide which ones were going to be
Veritext Corporate Services

800-567-8658

973-410-4040
9349fef0-7eb5-423b-bd0b-89d5688fb872

Page 85
1

displayed and which ones were not going to be

2

displayed; correct?

3

A.

That -- that criterion is not based

4

on -- it's not an absolute criterion.

5

based on -- it is based on relative properties of

6

the members of the set where they -- where they

7

belong in the ranking.

8
9

Q.

It is not

So you think that in order for

something to be filtering, it has to -- there has

10

to be a decision being made concerning only the

11

properties of that individual member of the set?

12
13
14

A.

Right.

Say yea or nay, depending on

those properties.
Q.

So if you're including some decision

15

concern -- some criteria concerning properties of

16

other members of the set in your decision to say

17

yea or nay, then you're not filtering?

18

A.

If you're comparing this one to the

19

other members of the set in your -- in your

20

criteria, these comparative criterion, then

21

you're not filtering.

22

subsetting.

Bowman calls that

23

Q.

What -- what was that last part?

24

A.

There are -- one of the other

25

references uses a different term for it.
Veritext Corporate Services

800-567-8658

973-410-4040
9349fef0-7eb5-423b-bd0b-89d5688fb872

Page 86
1

Q.

Oh, okay.

2

A.

We can get to it later.

3

Q.

Okay.

Good.

Okay.

So let's look at -- back to the

4

I understand what you're

saying now.

5
6

Culliss reference, which is Exhibit No. 4 to your

7

deposition, and I want to look at column 11 here.

8

So you see here there's a section that

9
10

begins about line 6 or 7, something like that,
the heading is "Ratings"?

11

A.

Yes.

12

Q.

And it says, "The invention can also

13

be used to organize articles by ratings.

14

end, the key terms of the index may additionally

15

or alternatively comprise rating key terms

16

represented by the generic labels G rated and X

17

rated, for example."

18

To this

Do you see that?

19

A.

Yes.

20

Q.

So you understand that Culliss is

21

saying, in addition to the key terms used in the

22

index, you can include a rating index?

23
24
25

A.

You can include rating labels.

It

didn't say index, but -Q.

Oh, that's fair.

You can include in

Veritext Corporate Services
800-567-8658

973-410-4040
9349fef0-7eb5-423b-bd0b-89d5688fb872

Page 87
1

your index, along with the key terms, a rating

2

label; correct?

3
4
5

A.

You can include a rating label, yes.

It could be in the index.
Q.

Okay.

So then it goes -- the Culliss

6

reference goes on to say, "The rating key term is

7

considered appropriate for all ages while the

8

rating key term X rated is considered appropriate

9

only for adults."

10
11
12

Do you see that?
A.

Where does he say that?

There it is.

I found it.

13

Q.

It's like line 12, maybe.

14

A.

Yes, I see that.

15

Q.

Then at line 15 it says, "The articles

16

are initially associated with one or more of

17

these key terms by any possible manner, such as

18

by human judgment or default association."

19

Do you see that?

20

A.

I see that.

21

Q.

So do you understand a human could

22

initially decide this is G-rated content or this

23

is X-rated content in the article?

24

A.

Yes.

25

Q.

Okay.

And then if we go down a little

Veritext Corporate Services
800-567-8658

973-410-4040
9349fef0-7eb5-423b-bd0b-89d5688fb872

Page 88
1

bit further about line 23, which is the next

2

paragraph there in column 11.

3

the rating key terms can be incorporated into the

4

index of key terms and included in the

5

association of the comparison score and, if used,

6

the key term probability score."

7

It says, "Moreover

Do you see that?

8

A.

I see that.

9

Q.

So the -- the rating can be an

10

additional value in the key term index; correct?

11

A.

Yes.

12

Q.

And then in the example that's

13

provided here, if we go down about line 39, it

14

says, "The invention operating separately from or

15

in addition to the manner described above would

16

permit or require the user to enter a rating key

17

term in the search query."

18

Do you see that?

19

A.

I see that.

20

Q.

So there you understand that you could

21

have an example where you put in key terms alpha

22

and gamma, and in addition you say, I want G

23

rated; right?

24
25

A.

You would -- not exactly, but close to

what you said.

You would provide three key terms

Veritext Corporate Services
800-567-8658

973-410-4040
9349fef0-7eb5-423b-bd0b-89d5688fb872

Page 89
1
2

in the query:
Q.

Alpha, gamma, G or G rated.

Okay.

And -- and we talked about

3

earlier, alpha and gamma can be associated with

4

words that are in the article; correct?

5

A.

Yes, they can be.

6

Q.

So then, the next sentence there,

7

beginning about line 41, says, "The invention

8

would operate in a similar manner for the rating

9

key terms as described above for the key terms

10

alone, whereby the search activity of the user

11

would alter the key term scores and key term

12

total scores for the rating key terms."

13

Do you see that?

14

A.

I see that.

15

Q.

So, in other words, users -- you'd

16

have a key term score associated with the rating

17

that's initialized at some particular value;

18

right?

19

A.

Yes.

20

Q.

Yeah.

It seems to be once here.
In the example that's shown,

21

for example, in the index, you know, surrounding

22

line 35 of column 11, they're initialized to the

23

values of one; correct?

24

A.

Yes.

25

Q.

And then continuing with the example,
Veritext Corporate Services

800-567-8658

973-410-4040
9349fef0-7eb5-423b-bd0b-89d5688fb872

Page 90
1

one of the key terms that the user would enter in

2

the query in this embodiment is to include a

3

rating key term; correct?

4

A.

You're referring to 47 through 55?

5

Q.

Correct.

6

A.

Yeah, that's what it says there.

7

Q.

And so that rating key term score

8

would be altered by both whether an article is

9

returned, as well as whether it's selected by the

10

user as per the previous examples that we

11

discussed; correct?

12

A.

Yes.

Under the embodiment where it

13

keeps both scores with a slash notation as -- as

14

exemplified here, that would be correct.

15

Q.

Okay.

So then, if we look toward the

16

bottom of column 11, continuing onto the top of

17

column 12, it says, "In this manner," peo --

18

"people looking for X-rated material will

19

identify and effectively label that material as X

20

rated.

21

screened entirely from the rating key term of

22

G rated by precluding articles entirely from the

23

search results which have a key term probability

24

score or comparison score for the rating key term

25

X rated above a predetermined threshold."

Such X-rated material can then be

Veritext Corporate Services
800-567-8658

973-410-4040
9349fef0-7eb5-423b-bd0b-89d5688fb872

Page 91
1

Right, you see that?

2

A.

I see what it says here, yes.

3

Q.

So in what's described there, the --

4

the rating key term will be evaluated for each

5

article individually; correct?

6

A.

It -- he's talking about putting in

7

the ratings as -- in the same way you would put

8

other data associated with the -- with the

9

article such as key terms.

10

Q.

Understood.

But here in the part that

11

I just read at the bottom of column 11, the

12

rating key term for each of the articles is

13

evaluated to determine whether the article will

14

be displayed independently of any of the other

15

articles in the set; correct?

16

A.

It doesn't say that it's evaluated

17

with respect to whether the article will be

18

displayed.

19

Q.

Well, it says, "In this manner, people

20

looking for X-rated material will identify and

21

effectively label that material as X" -- "as X

22

rated.

23

screened entirely from the rating key term of

24

G rated by precluding articles entirely from the

25

search results which have a key term probability

Such X-rated material can then be

Veritext Corporate Services
800-567-8658

973-410-4040
9349fef0-7eb5-423b-bd0b-89d5688fb872

Page 92
1

score or comparison score for the rating key term

2

X rated above a predetermined threshold"; right?

3

A.

That's what it says.

4

Q.

So isn't that saying that the rating

5

key term score will be evaluated independently

6

for each article to determine whether that

7

article will be screened entirely from the search

8

results?

9

A.

The -- the X rated or the G-rated key

10

term will have a score, and then that score can

11

be modified over time, depending on the feedback

12

from the user and the key term probability score

13

or comparison score above a predetermined

14

threshold, and then that -- that key term, then,

15

can be used as -- as a factor or criterion in

16

determining what to show or what to -- what to --

17

not to show, it doesn't say show.

18

say?

19

What does he

From the search results.
Q.

Right.

And if we look -- let's just

20

look further.

21

more light on this in the example.

22

Maybe we can shed a little bit

Continuing in column 12, about line 6,

23

it says, "For example, suppose article A3

24

contained adult content, and articles A1 and A2

25

contained not adult content, which would not be
Veritext Corporate Services

800-567-8658

973-410-4040
9349fef0-7eb5-423b-bd0b-89d5688fb872

Page 93
1

of interest to users searching with the rating

2

key term of X rated."

3

Do you see that?

4

A.

Right.

5

Q.

So in this example, the users that are

6

performing the query, one of the key terms they

7

include is a rating key term; correct?

8

A.

That's right.

9

Q.

Okay.

Then it goes on to say, "After

10

several users have performed searches, the index

11

may look like this," and then provides an

12

example.

13
14
15
16

Do you see that?
A.

Provides an example where X rated was

preferred for -- for article A3.
Q.

17

For article A3, correct.
And then it continues in the

18

description, beginning at line 20 of column 12,

19

"While the article A3 containing adult content

20

was clearly of interest to the G-rated crowd, it

21

was also clearly of interest to the X-rated

22

crowd.

23

the article A3 (i.e., prevent the squib from

24

being displayed from search queries containing

25

the rating key term of G rated)."

Accordingly, the invention would screen

Veritext Corporate Services
800-567-8658

973-410-4040
9349fef0-7eb5-423b-bd0b-89d5688fb872

Page 94
1
2

Do you see that?
A.

I see what he says, but I don't -- I

3

don't see how the invention would -- would screen

4

it, because it also has a high -- high

5

probability value under his example.

6

Q.

Right.

Well, what it says is when it

7

has a high probability value of being X rated, it

8

will be screened from key queries containing a

9

G-rated key term; correct?

10

A.

Under this example, it would have a

11

high probability value for both X rated and

12

G rated.

13

it's 45 over 45 for one.

14

over 22 or, you know, .95 or whatever that number

15

actually is.

16

Q.

Slightly higher for X rated, because

Right.

In the other case, 21

And what -- what it's saying

17

is because the article A3 has a high probability

18

for X rated, it will be screened from queries --

19

meaning not displayed, from queries that contain

20

the G-rated term; correct?

21
22
23

A.

The -- the method disclosed or taught

here does not -- would not have that effect.
Q.

Well, that's exactly what it says;

24

right?

25

content was clearly of interest to the G-rated

"While the article A3 containing adult

Veritext Corporate Services
800-567-8658

973-410-4040
9349fef0-7eb5-423b-bd0b-89d5688fb872

Page 95
1

crowd."

2

In other words, little kids may be

3

wanting to look at it, but they're not supposed

4

to be looking at it; right?

5

A.

Right.

Like the -- even if you divide

6

the -- the users into two groups, the G rated and

7

the X-rated crowd here, what's taught in the

8

invention would not have that -- that result.

9

Q.

Even where it says, "Accordingly, the

10

invention would screen the article A3 (prevent

11

the squib being displayed from search queries

12

containing the rating key term of G rated)";

13

right?

14

A.

The methods that are described before

15

will not have the effect of performing that

16

screening.

17

Q.

Sure they will.

If your -- if your

18

threshold is -- if it has a high probability of

19

being X rated, I'm not going to show it to

20

G rated; correct?

21
22
23

MR. JACOBS:

Objection.

Asked and

answered.
A.

Yeah.

The -- the method described

24

here where you show by -- by the scores will not

25

have that effect.
Veritext Corporate Services

800-567-8658

973-410-4040
9349fef0-7eb5-423b-bd0b-89d5688fb872

Page 96
1

Q.

You don't think that this will screen

2

article A3 from -- from queries that include the

3

G-rated key term?

4
5
6

MR. JACOBS:

Objection.

Asked and

answered.
A.

The method described of generating

7

these tables, calculating the probability or the

8

incremental values, will not have the desired

9

effect or the stated effect of screening one and

10
11

permitting the other.
Q.

Well, sure it will, if your threshold

12

is -- does it have a high probability of being

13

X-rated content; correct?

14

A.

That is not the way the method here

15

works.

16

calculating the scores between what the terms in

17

the user preference is and ranking according to

18

those scores.

19

The method here works is -- works by

Q.

Right.

But this screening of X rated

20

from G rated is based solely upon the score

21

associated with the rating key term; correct?

22

A.

Yeah.

The -- the part here that is

23

inconsistent with the earlier teachings is,

24

accordingly, the invention would screen the

25

article A3.

The way that the invention is

Veritext Corporate Services
800-567-8658

973-410-4040
9349fef0-7eb5-423b-bd0b-89d5688fb872

Page 97
1

described, it will not have the effect of

2

screening that article.

3

Q.

Okay.

So you -- you think that both

4

the G-rated probability and the X-rated

5

probability have to be used in order to determine

6

whether you're going to screen the article in the

7

embodiment described in column 11 and 12 here,

8

that's what you're saying?

9

A.

It's essentially by looking at all of

10

the terms associated with the article.

11

need to devise an additional or new or different

12

mechanism to have the desired effect.

13

Q.

You would

Well, let's look at the next one.

14

the other hand" -- this is -- begins at

15

column 26.

16

terms for articles A1 and A2 under the X-rated

17

key term are low and suggests that those

18

articles, A1 and A2, do not contain adult

19

content.

20

displayed in response to a search query

21

"On

containing a rating key term of G rated."

22

"On the other hand, the rating key

Accordingly, these articles could be

Do you see that?

23

A.

Yes, I see what it says there.

24

Q.

So, in other words, the screening of

25

X rated content, meaning determining whether to
Veritext Corporate Services

800-567-8658

973-410-4040
9349fef0-7eb5-423b-bd0b-89d5688fb872

Page 98
1

display it to G-rated users, is based upon the

2

probability that a particular article has X-rated

3

content in this example; correct?

4

A.

The -- the method described here of

5

tabulating and calculating sums or probabilities

6

and then using that to score which items have a

7

higher score and which ones have a -- have a

8

lower score would not have the effect of using

9

the presence of some item to serve as a way of

10
11

excluding that item.
Q.

Well, of course it would.

We have --

12

we have the first example that's given that says

13

article A3 has a high probability, 45 out of 45,

14

right, of having X-rated content; correct?

15
16
17

A.

And that's what it says, yes, in the

example.
Q.

Right.

And so in that example,

18

it's -- in column 12, the Culliss reference says,

19

I'm not going to show that to people who included

20

a G-rated key term in their query; correct?

21

A.

The -- the mechanism -- that's what it

22

says what he wants to do.

23

disclosed here will not have the effect that he

24

desires.

25

Q.

The mechanism

Well, if you -- if you base -- well,
Veritext Corporate Services

800-567-8658

973-410-4040
9349fef0-7eb5-423b-bd0b-89d5688fb872

Page 99
1

let's just continue with the example.

2

Then we have the example of articles

3

A1 and A2 that have a low probability of

4

including X-rated content; correct?

5

A.

According to the example.

6

Q.

Right.

And in that instance, that low

7

probability of including X-rated content leads to

8

the conclusion that I can show those articles or

9

display those articles to a user, including the

10

G-rated key term in their query; correct?

11

A.

That's what it says here.

The Culliss

12

mechanism for generating the scores and so on

13

would not have that effect, either.

14

Q.

Well, the mechanism that's described

15

here is, take the X-rated probability and

16

determine whether you're going to display that

17

particular article to someone who does a G-rated

18

search; correct?

19
20
21

A.
here.

There's no real mechanism described
There's a desired outcome described here.

Q.

But you agree that in the example

22

that's shown here that we just looked at in

23

column 11 and 12, that the -- the determination

24

of whether to screen an article, meaning not

25

display it, is done on an article-by-article
Veritext Corporate Services

800-567-8658

973-410-4040
9349fef0-7eb5-423b-bd0b-89d5688fb872

Page 100
1

basis?

2
3

MR. JACOBS:
A.

Objection as to form.

The -- the -- no, not necessarily.

4

The -- Culliss discloses a mechanism for ranking

5

articles based on user feedback with some

6

initialization step, then he goes on to add

7

additional features like G rated and X rated and

8

collects statistics on those.

9

So if people who put X-rated items

10

liked a particular item a considerable number of

11

the time, and people who liked G rated also liked

12

it, that does not mean that that article

13

necessarily is -- is -- well, first of all, it

14

doesn't mean it's an X-rated article, it just

15

simply means it is liked by both crowds.

16

Q.

It says here, right, "While the

17

article A3 containing adult content was clearly

18

of interest to the G rated" cloud -- "crowd, it

19

was also clearly of interest to the X-rated

20

crowd.

21

the article A3 (i.e., prevent the squib from

22

being displayed from search queries containing

23

the rating key term of G rated."

24
25

Accordingly, the invention would screen

Do you see that?
A.

Right.

I agree that it says that.

Veritext Corporate Services
800-567-8658

973-410-4040
9349fef0-7eb5-423b-bd0b-89d5688fb872

Page 101
1

Q.

Right.

So the article has a high

2

probability of being X rated, so it's not shown

3

in response to a query including a G-rated search

4

term; correct?

5
6
7

MR. JACOBS:

Objection as to form.

Asked and answered.
A.

Right.

So Culliss would need a

8

different mechanism beyond the one described here

9

to have the desired effect.

10

Q.

The mechanism being determining

11

whether there's a high probability that it's X

12

rated; correct?

13

A.

The mechanism being how the -- the

14

ranking and selection are done compared to his

15

method of keeping these tables about all of the

16

other terms.

17

Q.

Yeah.

The probability associated with

18

whether the article has X-rated content is

19

determined at least in part by the feedback from

20

users that included an X-rated rating key term in

21

their query; correct?

22

A.

The way that the -- that the Culliss

23

works, that would be one of the criterion that

24

would be combined with -- with the others to

25

determine the pertinence of that article to the
Veritext Corporate Services

800-567-8658

973-410-4040
9349fef0-7eb5-423b-bd0b-89d5688fb872

Page 102
1
2
3

query.
Q.

That would be the only criterion;

correct?

4
5

MR. JACOBS:
A.

Objection as to form.

Culliss is -- developed a method where

6

he combines criteria from all the terms

7

associated with the article.

8

12, he mentions an example where one particular

9

criterion may dominate or maybe dominate over

Then at the top of

10

others.

11

describe under what cases when a criterion would

12

be used instead of the combination of the

13

criterion.

14

combination of the criterion.

15

He doesn't provide any mechanism to

Q.

His description throughout is a
So you don't think that beginning at

16

line 20, where he says article A3's got a high

17

level of interest to the X-rated crowd, so I'm

18

not going to show it to the G-rated crowd, and

19

conversely, articles A1 and A2 have a low level

20

of interest to the X-rated crowd, so I am going

21

to show it to the G-rated crowd describes a

22

mechanism?

23
24
25

MR. JACOBS:
A.

Objection as to form.

Article A2, for example, has no

ratings with respect to the G-rated crowd, so
Veritext Corporate Services

800-567-8658

973-410-4040
9349fef0-7eb5-423b-bd0b-89d5688fb872

Page 103
1

it's not clear whether they would like to see

2

item A2 or not.

3

Q.

Right.

Because I'm only -- in the

4

example that's shown here, I'm only basing it on

5

the interest that the X-rated crowd had in the

6

article; correct?

7

A.

So you're saying that the lack of

8

interest of the X-rated crowd on A1 and A2 makes

9

that a higher candidate for -- more eligible

10
11

candidate for showing it to other people.
Q.

To the G-rated crowd, that's what it

12

says right here.

13

key terms for articles A1 and A2 under the

14

X-rated key term are low and suggests those

15

articles A1 and A2 do not contain adult-rated

16

content.

17

displayed in response to a search query

18

containing a rating key term of G rated; right?

19

A.

On the other hand, the rating

Accordingly, these articles could be

Right.

So what you're describing here

20

is using a low probability as a positive

21

criterion for showing something.

22

that he had described here had a way of favoring

23

low probabilities over high probabilities under

24

certain cases and not under others, and there was

25

a mechanism for selecting these things, then I

If the method

Veritext Corporate Services
800-567-8658

973-410-4040
9349fef0-7eb5-423b-bd0b-89d5688fb872

Page 104
1

would -- then that would make sense, it would be

2

consistent with his invention.

3

He's describing here a desired

4

outcome, but he's not providing a modification of

5

his mechanism that -- that would work -- that

6

would work in general.

7

Q.

Well, sure he is, isn't he?

Isn't he

8

saying that I'm going to look at the probability

9

results associated with a particular key term for

10

a particular article and determine whether that

11

has a high probability of being X rated or a low

12

probability of being X rated?

13

A.

So let's take another -- another term

14

here like table or wood table.

15

people that like wood tables and other people

16

that don't seem to -- another group of people

17

that also like wood tables, but they happen to

18

have another property, like they -- they like

19

classic furniture, and the first one likes modern

20

furniture, and they both like wood tables.

21

There are some

So in this case, because the people

22

who like classic furniture like wood tables, you

23

would never show that to somebody who likes

24

modern furniture, even if they also like wood

25

tables.

The mechanism you're describing would
Veritext Corporate Services

800-567-8658

973-410-4040
9349fef0-7eb5-423b-bd0b-89d5688fb872

Page 105
1

have that effect.

2

Q.

Wood tables are X rated?

3

A.

Substitute for X rated modern

4
5
6
7

furniture versus classical furniture.
Q.

Well, I'm trying to stick with the

example that's here, X rated and G rated.
A.

Yeah.

And I'm giving you another

8

example that is structurally equivalent to show

9

that there's no mechanism that is being offered

10

to address this -- this -- to address this

11

problem.

12

So if I have -- you're the classic

13

furniture guy, and I'm the modern furniture guy,

14

or the other way around, and we both like wood

15

tables, just like the G and the X rated both

16

liked A3 here, like some -- some article that

17

talks about wood tables, the fact that you are

18

now a member of a club different from mine or a

19

crowd different from mine, to use your word, that

20

means that I would no longer be able to see wood

21

tables because you got wood tables or vice versa.

22

Q.

Sure.

If you -- if you decided

23

that -- in your example that you were going to

24

screen -- in your implementation, you were going

25

to screen on the basis of whether you were a
Veritext Corporate Services

800-567-8658

973-410-4040
9349fef0-7eb5-423b-bd0b-89d5688fb872

Page 106
1

member of the classic furniture club or the

2

modern furniture club; right?

3

A.

4

tables.

5

Q.

Yeah, but they could both like wood
So what?

I mean, we can see right

6

here in article A3, little kids can like

7

pornography, too; right?

8
9

A.

I agree it's a laudable goal to not

display pornography to little kids.

10

Q.

Right.

11

A.

This is -- this is not a part of what

12

the Culliss method would necessarily do by

13

combining all of the -- all of the elements.

14

you were to add to Culliss the fact that having

15

some items then precludes -- having a high value

16

on some item precludes showing -- a high value in

17

some terms, excuse me, or feature precludes

18

showing items that were preferred by people who

19

had a different value for that term, then you

20

would get an invention that works very

21

differently and not very well in general than

22

what Culliss actually describes.

23

Q.

If

So what you're saying is because you

24

can think of a bad implementation that Culliss

25

doesn't describe -Veritext Corporate Services

800-567-8658

973-410-4040
9349fef0-7eb5-423b-bd0b-89d5688fb872

Page 107
1

A.

Well --

2

Q.

Let me just finish so the record is

3

clear.

4

So you're saying because you can think

5

of a bad implementation not using the G-rated-

6

and the X-rated example, but using wood furniture

7

example, that Culliss doesn't describe a means

8

for screening articles based upon the results

9

associated with the rating key term?

10
11
12

MR. JACOBS:

Objection.

Misstates

testimony.
A.

Okay.

So these are not -- I'm not

13

describing an implementation.

14

outcome or a desired outcome, in this case an

15

undesired outcome that is analogous to the one

16

that is suggested here.

17

I am describing an

And I'm saying, in fact, that neither

18

one of these would be -- would result from an

19

implementation of the Culliss invention or the

20

Culliss patent describing these particular tables

21

that associate terms with items based on -- based

22

on popularity.

23

Q.

But in terms of the screening that's

24

described here in columns 11 and 12, the

25

screening's not based on any results associated
Veritext Corporate Services

800-567-8658

973-410-4040
9349fef0-7eb5-423b-bd0b-89d5688fb872

Page 108
1
2
3
4
5
6

with the other key terms; correct?
A.

The screening is -- that is described

here is essentially ignoring the other key terms.
Q.

Right.

So it's based upon -- solely

based upon the rating key term; correct?
A.

So, in general, in the invention how

7

would the system know which key term to associate

8

in order to do this kind of screening?

9
10
11

Q.

Because it -- you're saying that the

invention doesn't know one key term from another?
A.

The method described in the invention

12

does not use one key term and ignore all the

13

others throughout the description of the patent.

14

Q.

So are you saying that one of ordinary

15

skill in the art in 1998 couldn't implement a

16

system that looked at a particular key term like

17

a rating key term and made a judgment based upon

18

that key term?

19
20

MR. JACOBS:
A.

Objection as to form.

That would -- implementing a system

21

that -- that essentially looks at a single key

22

term, and then looks at the value of that key

23

term, and then decides whether to block it or not

24

is a mechanism which is separate from the

25

mechanisms described in the rest of Culliss.
Veritext Corporate Services

800-567-8658

973-410-4040
9349fef0-7eb5-423b-bd0b-89d5688fb872

Page 109
1

Q.

Right.

So the coding would be if

2

rating X-rated key term is above a threshold,

3

then block from G-rated queries; correct?

4
5
6

MR. JACOBS:

Objection as to form.

Incomplete hypothetical.
A.

The -- the method here would -- would

7

be, ignore everything else that -- that Culliss

8

does and then look at a single attribute; in

9

fact, look at the specific attributes that you

10

just described, and then make a decision based on

11

that particular attribute.

12

This is a little bit like saying that

13

if you have an automobile, and describe the

14

automobile invention, and -- and how to rate it,

15

and how to drive it, and so forth, and then you

16

also have -- add on to the automobile like if you

17

want to take a walk, put your shoes on, ignore

18

the automobile, jump out of the automobile, and

19

now whether you can walk or not depends upon

20

whether you have shoes.

21
22

Q.

Okay.

So that -- that isn't an

accurate example at all, is it?

23

So how about this example, would be a

24

better one to stick with your automobile example.

25

You have all sorts of factors:

Drivability,

Veritext Corporate Services
800-567-8658

973-410-4040
9349fef0-7eb5-423b-bd0b-89d5688fb872

Page 110
1

handling, engine power, model, make.

2

kinds of things; right?

3
4

All those

You also have color.

Now, the user says, I don't want a red
car.

It could happen all the time; right?

5

A.

That's right.

6

Q.

So in that instance, the user could

7

say any color but red.

8

to see it because I'm not going to buy it; right?

9
10
11
12
13
14

If it's red, I don't want

A.

Uh-huh.

Q.

So that would be filtering on the

criteria of the color of the car; correct?
A.

That would be the user doing that

filtering, not the system.
Q.

Right.

But, yes.

But you could implement a

15

system, they do this on -- they've done this on

16

search systems for quite some time now, haven't

17

they, where one of the criterion will be a

18

particular aspect of the product, in other words,

19

maybe the color in terms of cars, and then it

20

will display to you cars either only of that

21

color or, if you say, I don't want that color,

22

cars not of that color; correct?

23

MR. JACOBS:

24

Calls for speculation.

25

A.

Objection as to form.

There could be such a system.

That is

Veritext Corporate Services
800-567-8658

973-410-4040
9349fef0-7eb5-423b-bd0b-89d5688fb872

Page 111
1
2

not the mechanism described under Culliss.
Q.

Well, it is the mechanism described as

3

Culliss; right?

4

is indicative of whether the article has X-rated

5

content; correct?

6
7
8
9
10

You include a rating term that

MR. JACOBS:

Objection as to form.

Argumentative.
A.

One of the terms that is added to the

system is a rating term, yes.
Q.

Right.

And as described here in

11

columns 11 and 12, the purpose of that is to

12

determine whether an article is appropriate for

13

display to the G-rated audience; correct?

14

A.

15

rating term.

16

appropriate to display or not is -- is -- it's

17

not contained within that particular table.

18

Q.

The term that you're adding is the
Whether it's appro -- who it is

Well, but it says right at the

19

beginning of the example, going back to

20

column 11, about line 12, it says, "The rating

21

key term G rated is considered appropriate for

22

all ages, while the rating key term X rated is

23

considered appropriate only for adults"; right?

24

A.

That's what it says, right.

25

Q.

Right.

So G rated, I can show that to

Veritext Corporate Services
800-567-8658

973-410-4040
9349fef0-7eb5-423b-bd0b-89d5688fb872

Page 112
1

everybody.

2

adults; correct?

3
4
5
6

X rated, I can only show that to
That's what it's saying?

MR. JACOBS:
A.

Right.

Objection as to form.

So how would the system know

which are the adults and which are the G rated?
Q.

Well, because we know that one of the

7

key terms that you have to include is whether

8

you're searching for X-rated content or G-rated

9

content; correct?

10
11
12
13

MR. JACOBS:
A.

Objection as to form.

You could have a term in the query

that would be X rated or G rated, yes.
Q.

Right.

So you can use that as a proxy

14

for determining whether someone's an adult in

15

this instance; correct?

16
17

MR. JACOBS:
A.

Objection as to form.

If someone who's an adult was to

18

search for G rated or someone who's a child

19

states X rated when they shouldn't, you -- you

20

don't control that.

21
22

Q.

25

So what you're saying is

the system might not be perfect; right?

23
24

Understood.
MR. JACOBS:

Objection as to form.

Misstates testimony.
A.

Yeah.

I'm saying that this depends on

Veritext Corporate Services
800-567-8658

973-410-4040
9349fef0-7eb5-423b-bd0b-89d5688fb872