`
`BEFORE THE PATENT TRIAL AND APPEAL BOARD
`
`GOOGLE,
`
`INC., and YOUTUBE, LLC,
`
`Petitioner,
`
`vs.
`
`NETWORK-l TECHNOLOGIES,
`
`INC.,
`
`Patent UWHEJT .
`
`Case No.
`
`IPR2D15—flU34?
`
`VIDEOTAPED DEPOSITION OF PIERRE MOULIN, PhD
`
`VOLUME I
`
`Santa Monica. California
`
`Wednesday, August 19. 2Ul5
`
`NETWORK—l EXHIBIT 2006
`
`Google Inc. 1!, Network-1 Technologies, Inc.
`IPR2{}lS-00345
`
`REPORTED BY: RICH ALOSSI, RPR, CORR, CSR NO. 13497
`
`Job No: SEEDS
`
`TSG Reporting H Worldwide - BT7—Tfl2~95Bfl
`Pagel uf3H4
`
`
`
`UNITED STATES PATENT AND TRADEMARK OFFICE
`
`BEFORE THE PATENT TRIAL AND APPEAL BOARD
`
`GOOGLE,
`
`INC., and YOUTUEE, LLC,
`
`Petitioner,
`
`vs.
`
`NETWORK-l TECHNOLOGIES,
`
`INC.,
`
`Patent Owner .
`
`Case No.
`
`IPR2D15—flU3é?
`
`‘JIDEOTAPED DEPOSITION OF PIERRE MOULIN, PhD,
`
`VOLUME I,
`
`taken on behalf of the Patent Owner, at 2G1
`
`Santa Monica Boulevard, Sixth Floor, Santa Monica,
`
`California, on Wednesday, August 19, 21315,
`
`from 9:02 AM to
`
`5:39 PM, before RICH ALOSSI, RPR, CORR, CSR NO. 13497.
`
`'.i.".|l'1|‘
`
`TSG Reporting H Worldwide - 8T7—Tfl2-958d
`Page 2 of 384
`
`
`
`APPEARANCES:
`
`For the Plaintiff:
`
`SKADDEN ARPS SLATE MEAGHER & FLOM
`
`BY:
`
`JAMES ELACQUA, Attorney at Law
`IAN CHEN, Attorney at Law
`525 University Avenue
`Palo Alto, CA 94361
`
`For the Patent Owner Network-l Technologies:
`
`DDVEL & LUNER
`
`GREGORY DOUEL, Attorney at Law
`BY:
`2fll Santa Monica Boulevard
`
`Santa Monica, CA Qfldfll
`
`-and-
`
`RUSS AUGUST & KAEAT
`
`MARC FENSTER, Attorney at Law
`BY:
`12424 Wilshire Boulevard
`
`Los Angeles, CA 9flD25
`
`Also Present:
`
`SCOTT MOHAIR, Videographer
`RICH SONNENTAG, Litigation Counsel, Googie,
`MATTHAEUS WEINHARDT
`
`Inc.
`
`TSG Reporting H Worldwide - BT7—Tfl2-95Bfl
`Page 3 1;:-F384
`
`
`
`I30--JETILHI-l‘-5L.nJf\J}—‘
`
`WITNESS
`
`PIERRE MDULIN, PhD, VOLUME I
`
`BY MR. DGVEL
`
`MARKED
`
`E X H I E I T S
`
`Exhibit E—23T— Decision re Institution of Inter
`
`Partes Review
`
`Exhibit
`
`lUU1-23- United States Patent Number
`
`8,2U5,23? E2
`
`Exhibit
`
`lflfle-lT— Declaration of Dr. Pierre Noulin
`
`in Support of Petition for Inter
`Partes Review of US Patent
`
`Number 3,EdU,1T9
`
`Exhibit
`
`lUU4—23— Declaration of Dr. Pierre Moulin
`
`in Support of Petition for Inter
`Partee Review of Us Patent
`
`Number 8,2U5,23?
`
`Exhibit
`
`lUU4-98- Declaration of Dr. Pierre Noulin
`
`in Support of Petition for Inter
`Partes Review of US Patent
`
`Number 3,fllU,93B
`
`1UU9— United States Patent Number
`
`6,9?U,E86 El
`
`1UlU- United States Patent Number
`
`5,8?d,E36
`
`1012- United States Patent Number
`
`E,lE8,U1U Bl
`
`2DU5— Handwritten List of Formulae
`
`TSG Reporting H Worldwide - 8T?—TU2-958D
`Page 4 of 384
`
`
`
`"A Fast String
`Exhibit 2UflE— Article Titled,
`Searching Algorithm," October
`19??
`
`Exhibit 2UfiT— Iwamura Data Examples
`Spreadsheet
`
`TSG Reporting H Worldwide H 8T7~Tfl2~95Bfl
`Page 5 of 384
`
`
`
`SANTA MONICA, CALIFORNIA; WEDNESDAY, AUGUST 19, 2fl15
`
`9:02 AM - 5:39 PM
`
`THE VIDEDGRAPHER: Good morning. This is the
`
`start of Tape Number 1 of the videotaped deposition
`
`of Pierre Moulin in the matter of Google,
`
`Incorporated, and YouTube, LLC, versus Network—l
`
`Technologies,
`
`Incorporated. This oase is before the
`
`US Patent and Trademark Office, Patent Trial and
`
`Appeal Board. Case number is IPR2Ul5-flD34?.
`
`This deposition is being held at 291 Santa
`
`Monica Boulevard, on the sixth floor,
`
`in
`
`Santa Monica, California. Today's date is
`
`August l9th, 2915.
`
`The time is 9:03 AM.
`
`My name is Scott MoNair from TSG
`
`Reporting,
`
`Incorporated.
`
`I‘m the legal video
`
`specialist.
`
`The court reporter today is
`
`Rich Alossi, also in association with TSG Reporting.
`
`Will oounsel please identify yourselves
`
`for the record.
`
`MR. DGVEL: My name is Greg Dovel.
`
`I
`
`represent the patent owner, Network-1.
`
`MR. ELACQUA: My name is Jim Elaogua with
`
`Skadden Arps, here to represent Google and YouTube.
`
`MR. CHEN:
`
`Ian Chen with Skadden, also on
`
`TSG Reporting H Worldwide - ST?—Tfl2-9589
`Page 15 of 384
`
`
`
`behalf of Google and YouTube.
`
`MR. SONNENTAG: Rich Eonnentag.
`
`in-house
`
`oounsel at Google.
`
`MR. FENSTER: Mare Fenster with Russ. August
`
`&
`
`Rabat, representing the patent holder in the
`
`litigation.
`
`THE UIDEOGRAPHER:
`
`Thank you. Will
`
`the court
`
`reporter p1ease swear in the witness.
`
`THE REPORTER:
`
`Please raise your right hand.
`
`Do you solemnly state that the testimony
`
`you are about
`
`to give in the matter now pending
`
`shall be the truth,
`
`the whole truth and nothing but
`
`the truth?
`
`THE WITNESS:
`
`I do.
`
`THE REPORTER:
`
`Thank you.
`
`We're on the reoord.
`
`PIERRE MOULIN, PhD,
`
`having been first duly sworn by
`
`the court reporter, was examined
`
`and testified as follows:
`
`BY MR. DOUEL:
`
`EXAMINATION
`
`Q
`
`Your name is pronounced -- is it Moulin?
`
`TSG Reporting H Worldwide - ET?—TO2-BEBO
`Page "I of3H4
`
`
`
`Yes. Almost.
`
`Dr. Moulin?
`
`Yes.
`
`Q
`
`How, Dr. Moulin, you're familiar with the
`
`concept of whether or not an algorithm is suhlinear?
`
`A
`
`Yes.
`
`Q Were you familiar with that before you
`
`started work on this case?
`
`A
`
`Yes
`
`Q When is it during the course of one's
`
`education that one would learn about
`
`the concept of
`
`suhlinear?
`
`A
`
`The concept of sublinear arises in many
`
`contexts, not just databases.
`
`The first time
`
`someone would be exposed to that would be typically
`
`in a university course, depending on their field.
`
`Q
`
`when you say "university Qourse," what
`
`course would that be?
`
`ME. ELACQUA:
`
`It depends on the field.
`
`So it
`
`could be in computer science. Often that might be
`
`TSG Reporting - Worldwide - 3TT*'?fl2~E+5Eifl
`Pngcfloffifld
`
`
`
`introduced at the senior level, or if it's
`
`electrical engineering,
`
`that might also be
`
`introduced around the same time.
`
`Each program is a
`
`different curriculum.
`
`MR. DDVEL:
`
`I'm going to mark as Exhibit
`
`Number lDU4—23T the Declaration of Dr. Moulin from
`
`the case related to the '23? patent.
`
`{Exhibit 1DD4—23T was marked for
`
`identification by the court reporter and is
`
`attached hereto.}
`
`MR. DDVEL:
`
`I've placed that in front of the
`
`witness.
`
`BY MR. DOUEL:
`
`Q
`
`In your Declaration that I've placed in
`
`front of you, you provide a definition of the
`
`concept of sublinear execution time;
`
`is that
`
`correct?
`
`A Where?
`
`Q
`
`A
`
`Q
`
`Pages 19 to 20 of your Declaration.
`
`That is correct, yes.
`
`How, on page 2D you've got a diagram that
`
`illustrates the concept of what is a sublinear
`
`search and what is a linear search; correct?
`
`A
`
`Q
`
`It illustrates it, yes.
`
`Is it the case that if we have a linear
`
`TSG Reporting H Worldwide - 8T?—TD2-958D
`Page 9 of 384
`
`
`
`search,
`
`that it will have a constant slope as you've
`
`depicted in your diagram there?
`
`A
`
`"Constant slope" in terms of ——
`
`in terms
`
`of the number of entries in the search database, you
`
`mean?
`
`Is that what you mean?
`
`Q
`
`A
`
`Q
`
`Yes. Yes.
`
`In this example, yes.
`
`Is it the case that as sublinear is used
`
`in the patent, it's going to result in a scaling of
`
`the search time compared to the size of the database
`
`such that it will have a constant slope?
`
`A
`
`Q
`
`A
`
`Q
`
`You say if it is a linear search?
`
`If it's a linear search. Right.
`
`So let me give you a different question.
`
`Only in this example, yes.
`
`Let me give you a different question so we
`
`have a clean record.
`
`If we have an algorithm that is a linear
`
`search,
`
`is it the case that it will have a constant
`
`slope if we plot the search time compared to the
`
`size of the database?
`
`A
`
`Q
`
`Yes.
`
`Is it the case that if we have a sublinear
`
`search,
`
`then the slope will curve down and approach
`
`zero if we plot search time compared to the size of
`
`TSG Reporting H Worldwide - 8??—?fl2-958$
`Page II] of 334
`
`
`
`the database?
`
`A
`
`Not necessarily.
`
`Q When wouldn't it do that?
`
`A
`
`Because this is an example that shows only
`
`how search time is as a function of number of
`
`entries in the search database. But that's not the
`
`only criterion.
`
`Q What's the other criteria?
`
`A Well, it's the size of the database that
`
`matters.
`
`So number of entries is only one factor
`
`that affects the size of the database.
`
`Q
`
`Let's -- let's talk about the size of the
`
`database,
`
`then.
`
`Is it the case that if we have a sublinear
`
`search as it's used in the Cox —— withdrawn.
`
`When we're talking about the Cox patents,
`
`do you understand we're talking about the four
`
`patents that you filed declarations for in
`
`connection with these proceedings before the PTAE?
`
`A
`
`Q
`
`Yes.
`
`If we have a sublinear search as it's used
`
`in the Cox patents,
`
`is it the case that if we plot
`
`search time compared to the size of the database
`
`over which we are searching,
`
`then the —— it will
`
`produce a curve such that it curves downward and
`
`TSG Reporting H Worldwide - 8T?—Tfl2-958$
`Pagell of3H4
`
`
`
`approaches zero?
`
`A
`
`You need to be more specific. Sublinear
`
`is mentioned in many places in all four patents in
`
`slightly different contexts.
`
`So I would like you to
`
`be more specific.
`
`Q
`
`Are you —— is it your testimony --
`
`withdrawn.
`
`We're talking about suhlinear as it's used
`
`in the claims of the patents.
`
`A
`
`Q
`
`Right.
`
`You understand that some of the claims
`
`have the phrase "sublinear search," some of them
`
`have the phrase "sublinear time search"?
`
`A
`
`Q
`
`That‘s my recollection.
`
`Is it the case that suhlinear search and
`
`sublinear time search mean the same thing in the
`
`patents?
`
`A
`
`I would have to check every instance of
`
`sublinear data. Used in different contexts,
`
`sometimes not defined precisely, you would have to
`
`refer me to a specific instance.
`
`Q
`
`As the phrase "sublinear search“ is used
`
`in the claims of the patents,
`
`is it used
`
`consistently, or does it mean different things in
`
`the claims?
`
`TSG Reporting H Worldwide - 8T?—Tfl2-95Bfl
`Page [2 of 33.4
`
`
`
`A
`
`Q
`
`It's used somewhat vaguely.
`
`I didn't ask whether it was vague.
`
`I want
`
`to know does it mean the same thing or does it mean
`
`different things?
`
`A
`
`I don't know. Because it's done vaguely,
`
`I cannot know for sure.
`
`Q
`
`You don't know one way or the other --
`
`withdrawn.
`
`When you did your analysis and decided
`
`that the prior art anticipates --
`
`A
`
`0
`
`Right.
`
`—— and you decided that it disclosed a
`
`sublinear time search, what definition did you use?
`
`A
`
`It's in relation to the size of the
`
`database, which itself depends both on the number of
`
`entries in the database and on the size of each
`
`entry, each record. That's the standard definition.
`
`Q
`
`A
`
`That's the standard definition: right?
`
`Yes.
`
`Q When you say "standard," you're talking
`
`about
`
`the standard, well—known definition in the
`
`art?
`
`A
`
`Yes
`
`TSG Reporting — Worldwide — 3T7*Tfl2~95Bfl
`Page I3 of 334
`
`
`
`Q
`
`and we can increase the size of the
`
`database either by increasing the number of entries
`
`or by increasing the size of each entry in the
`
`database?
`
`A
`
`Q
`
`That is correct.
`
`Is it the case that if we have a sublinear
`
`search —e withdrawn.
`
`Is it your understanding that the term
`
`"suhlinear search" and "sublinear time search" is
`
`used in the patent claims to have that definition
`
`that you just gave,
`
`the standard definition?
`
`A
`
`Again, it has often not been defined
`
`precisely.
`
`The word "sublinear" is used in a
`
`slightly different context
`
`in the different patents.
`
`When they say "sublinear," my
`
`understanding is, again, consistent with the art.
`
`I
`
`understand that to mean what we just discussed.
`
`So
`
`it would mean what you just discussed.
`
`So every time I read "sublinear“ in the
`
`patents,
`
`that's my understanding of it.
`
`Even though
`
`it was not clearly defined.
`
`TSG Reporting - Worldwide - 3TT*'?fl2~E+5Eifl
`F'ag¢|4of3$4
`
`
`
`MR. DDVEL:
`
`I'm going to place in front of the
`
`witness another exhibit which I'm going to mark as
`
`Exhibit
`
`lDfll—23?. This is the '23? patent.
`
`{Exhibit
`
`lflfll-23? was marked for
`
`identification by the court reporter and is
`
`attached hereto.}
`
`BY MR . DCWEL:
`
`Q
`
`I'd like you to turn to the claims of the
`
`'23? patent.
`
`Do you see Claim 1 uses the phrase
`
`“suhlinear time search"?
`
`A
`
`Q
`
`Yes.
`
`Do you see Claim 5 uses the term
`
`"suhlinear time search"?
`
`A
`
`Q
`
`res.
`
`Does the phrase “sublinear time search,"
`
`as it's used in the '23? patent claims, refer to a
`
`search where the search time decreases at less than
`
`a linear -- or increases at less than a linear rate
`
`as we increase the size of the dataset over which
`
`we're searching?
`
`A
`
`The claims do not say "as you increase the
`
`size of the database."
`
`If I read Claim 5, it reads,
`
`“A suhlinear
`
`time search of extracted features of identified
`
`TSG Reporting H Worldwide - 8T?—Tfl2-95Bfl
`Page [5 of 33.4
`
`
`
`media works," by which ene eeuld understand that it
`
`means in terms ef the database size. But it's
`
`not -— it does not read that explicitly.
`
`I
`
`F—"
`
`2 3 I
`
`C
`
`I
`
`I
`
`I
`
`I
`
`-—
`
`C I
`
`_ C
`
`I
`
`C
`
`C
`
`C
`
`_ -
`
`II
`
`C I—
`
`- I
`
`-—
`
`D I
`
`--
`
`C I-
`
`!'\J ILJ'|
`
`MR. DU?EL:
`
`I'm going to mark as Exhibit 2DU5
`
`TSG Reporting - Worldwide - 3?"!-'?C|2~E+5Eifl
`Pagclfioflfl
`
`
`
`a blank piece of notebook paper.
`
`and we‘re going to
`
`make some notations on it.
`
`{Exhibit 2DD5 was marked for identification
`
`by the court reporter and is attached
`
`hereto.}
`
`MR. DQVEL:
`
`and the first thing I'm going to
`
`write is "DEHI." I'm going to stick that in front of
`
`the witness.
`
`THE WITNESS: Qkay-
`
`BY MR. DQVEL:
`
`Q
`
`A
`
`are you familiar with that notation?
`
`Yes.
`
`Q What does that indicate to you, as one of
`
`ordinary skill in the art?
`
`A
`
`So there's a parameter, N, which is a
`
`number. And one, let's say,
`
`looks at computational
`
`time as a function of N.
`
`So if N grows, order N
`
`means that the search time will not increase faster
`
`than linearly.
`
`so it could increase linearly, or it
`
`could increase sublinearly.
`
`Q What that tells us is that the upper bound
`
`for the increase in the execution time is going to
`
`be linear; right?
`
`A
`
`Q
`
`At most linear.
`
`Is it the case that —— withdrawn.
`
`TSG Reporting H Worldwide - 8T?—Tfl2-95Bfl
`Pagal?of334
`
`
`
`5Ioouoonuu~
`
`Ir-' I\J
`
`and there. when we see that —— withdrawn.
`
`Q
`
`If N refers to the size of the dataset
`
`over which we're searching,
`
`then if we've got a
`
`database and what we know about it —— withdrawn.
`
`If N refers to the size of the database
`
`over which we're searching, and if we have an
`
`algorithm or search that performs at —— Din), as
`
`we've written there in Exhibit 2flfl5, does —— what
`
`does that tell us about whether it's linear or
`
`sublinear?
`
`A
`
`It says -- it is at most linear. That's
`
`often a source of confusion, by the way.
`
`It means
`
`at most linear.
`
`It does not mean necessarily
`
`linear.
`
`Q
`
`That would indicate -— withdrawn.
`
`Is it the case that when scientists in
`
`TSG Reporting - Worldwide - 3T":'*'?fl2~E+5Eifl
`Pagclfloflfl
`
`
`
`your field are assessing the complexity of an
`
`algorithm and how its execution time scales,
`
`that
`
`it's based upon using a hypothetical dataset?
`
`A
`
`If you apply that notation to -— to a
`
`dataset and you say N is the size of the datsset,
`
`that would mean that the function on the left side,
`
`which is a search time in your example, grows at
`
`most linearly with E.
`
`Q
`
`If we have a -- let's assume we've got
`
`a -- another algorithm and what we know about it is
`
`that its search time, compared to the size of the
`
`database, N,
`
`is that it's O{kn}, where K is a
`
`constant.
`
`What does that indicate to you?
`
`A
`
`The constant —— if K is a constant,
`
`the
`
`two equations you've written are equivalent, because
`
`U{n} or order of, say,
`
`2N is the same thing
`
`mathematically.
`
`Q
`
`Does it,
`
`in both instances,
`
`indicate that
`
`the search time is on the order of linear?
`
`A
`
`Q
`
`At most linear.
`
`If all we know about a search is that its
`
`search time oompared to the size of the database
`
`over which we're searching grows D{n} or {kn}, do we
`
`know that it's sublinear?
`
`TSG Reporting H Worldwide - 8T?—Tfl2-958$
`Page [9 of 334
`
`
`
`Pu.
`
`We don't know because the notation is not
`
`refined enough to indicate that.
`
`You would have to
`
`use another notation.
`
`Q What other notation?
`
`The theta of N notation.
`
`What's the "theta alpha notation"?
`
`"Theta of N" notation.
`
`Theta of H notation.
`
`So if instead of big 0 you had written
`
`A
`
`Q A
`
`Q A
`
`theta,
`
`that would mean linear N.
`
`Q
`
`Why don't you go ahead and write that down
`
`for me on Exhibit 2DO5.
`
`and what you've written there is,
`
`just for
`
`the record, something that looks like an H with a
`
`circle around it,
`
`in}.
`
`And that's pronounced theta
`
`of N?
`
`It's capital letter theta. Greek letter.
`
`and it's pronounced "theta N"?
`
`Theta of N.
`
`Theta of N.
`
`Now, if we have something that‘s theta of
`
`RN, what does that indicate?
`
`A
`
`Again, if K is a constant,
`
`like 2,
`
`there
`
`is no difference between theta of N and theta of KM.
`
`They mean the same thing mathematically.
`
`TSG Reporting H Worldwide - 8T?—Tfl2-95Bfl
`Page Efl of 33.4
`
`
`
`Q When we're assessing whether or not a
`
`given search is a suhlinear search or suhlinear time
`
`search, do we determine it based upon whether or not
`
`the search time grows compared to the size of the
`
`query?
`
`A
`
`You have said —— okay.
`
`We need to be
`
`specific here.
`
`You say the search time.
`
`okay. You're
`
`looking at the search time as a function of -- now
`
`this has to do with query: before you talked about
`
`the size of the database. Which one is it?
`
`Q
`
`Let me back up and ask you another couple
`
`of questions.
`
`When we're -- in this field when we're
`
`talking about a -- using a search, what we're doing
`
`is we're comparing a query or a pattern against a
`
`database of data;
`
`is that right?
`
`A
`
`Q
`
`Yes.
`
`and if we talk -- is it -- are there other
`
`words for "query" or "pattern"?
`
`What's the —— what's the best one to use?
`
`A
`
`"Query" is fine.
`
`one could ~- could say
`
`"probe" as well .
`
`TSG Reporting - Worldwide - 3".I'7"*-'?fl2~E+5Eifl
`F'ag¢2Iof3$4
`
`
`
`Q
`
`In —— in looking at some of the prior art,
`
`they refer to a w- trying ~~ having a song or a
`
`melody that is unknown and then trying to determine
`
`what that melody actually is.
`
`You're familiar with that sort of prior
`
`A
`
`Q
`
`YES.
`
`Is the unknown melody the equivalent of
`
`the query or the pattern that we're searching for?
`
`A
`
`So the —— the —— what's submitted by the
`
`user is the query; what you are comparing it against
`
`is —— is the database.
`
`So it --
`
`Q
`
`Now, if we're trying to determine whether
`
`a given algorithm is sublinear or linear. do we
`
`assume that the query size is fixed or does the
`
`query size vary when we're trying to assess whether
`
`it's sublinear?
`
`A
`
`It depends on the problem considered,
`
`the
`
`different flavors of these problems.
`
`Q
`
`If we're trying to determine whether
`
`something is sublinear as it's used in the olaims of
`
`TSG Reporting — Worldwide — 3T7*Tfl2~95Bfl
`Pag¢22of334
`
`
`
`the —— claims of the patents —— withdrawn.
`
`If we're talking about
`
`the concept of
`
`sublinear search or sublinear time search as it's
`
`used in the '23? patent, does that refer to a search
`
`time that grows at a sublinear rate as we increase
`
`the size of the pattern, or —— or does it refer to a
`
`search that increases at a sublinear rate as we
`
`increase the size of the database?
`
`A Usually,
`
`the one that matters is the size
`
`of the database, because that's a big one.
`
`Q When the —— in your definition that you
`
`set forth on page 19, where you said,
`
`"A sublinear
`
`search means a search whose execution time has a
`
`sublinear relationship to database size," you're
`
`referring to "sublinear" as it's used in the '23?
`
`patent;
`
`is that right?
`
`A
`
`Let me check the '23? patent.
`
`What I gave in my Declaration is a general
`
`definition of sublinear search.
`
`In the patent
`
`here —— for instance, if you look at Claim 5 that
`
`you -- we just discussed, sublinear time is used.
`
`It does not say what
`
`the reference is.
`
`Again,
`
`in those problems,
`
`the parameter
`
`that's very big that's of concern is the size of the
`
`database.
`
`TSG Reporting H Worldwide - 8T?—Tfl2-958$
`Page 23 of 334
`
`
`
`I-"‘ LU
`
`i--" n-'|'-‘I
`
`I--" U'|
`
`|—" CT‘:
`
`In-F --J
`
`1-‘ CO
`
`v|—|
`
`‘-53
`
`1'\J (II
`
`Ix]
`
`|~—‘
`
`l\) |'\.'|I
`
`|'\) LIJ
`
`1\J in
`
`!'\J ILJ'|
`
`Q Would it be wrong to apply a definition
`
`that said, "I'm going to determine if it's sublinear
`
`based upon whether or not
`
`the search time grows
`
`compared to the size of the query or the pattern?
`
`A Mathematically,
`
`this could be done.
`
`Whether it's relevant from an engineering point of
`
`View is a different matter.
`
`My opinion here is as an engineer. Okay?
`
`The parameter that's big that's of concern is the
`
`database size.
`
`Q
`
`In trying to determine whether or not the
`
`‘23T patent is anticipated,
`
`is it your opinion that
`
`the definition you've set forth in paragraph 53 is
`
`TSG Reporting - Worldwide - 3TT*'?fl2~E+5Eifl
`Pag¢24of3$4
`
`
`
`F—"
`
`the correct cne?
`
`A
`
`That's the definition I have applied.
`
`and
`
`1, yes, believe it's correct. That's my epinicn.
`
`I
`
`2 3 I I I I I I
`
`-
`
`Q
`
`It wculdn‘t be relevant to the ——
`
`assessing the '23? claims;
`
`is that right?
`
`A
`
`In the field of database search in
`
`- D I
`
`-"‘ LU
`
`14
`
`15
`
`16
`
`general —— and this includes the —— the patents
`
`1? we're discussing —— it's the database size that is
`
`l8
`
`the large number.
`
`and when you say "order of,“
`
`typically you refer to that.
`
`Q
`
`If we're trying to determine whether the
`
`'23? —— withdrawn.
`
`TSG Reporting — Worldwide — 3TT*Tfl2~95Bfl
`Pag¢25nf3£4
`
`
`
`IO
`
`Is it the case that for something to be --
`
`I I I U C . I C 9
`
`1D
`
`withdrawn.
`
`Q
`
`If we were doing —- withdrawn.
`
`If semeene was doing an assessment --
`
`withdrawn.
`
`TSG Reporting — Worldwide — 3??-Tfl2~95Bfl
`Page 26 of 334
`
`I I I - - - - . - I I l
`
`\) |\.'|I
`
`|'\) LIJ
`
`24
`
`
`
`lfl
`
`11
`
`12
`
`13
`
`14
`
`15
`
`lfi
`
`1?
`
`Appeal Board was presented with prior art that
`
`presented a search that was linear with respect to
`
`the size of the database, but it was suhlinear with
`
`respect to the size of the pattern.
`
`would that prior art demonstrate a
`
`suhlinear search as it's used in Claim 25?
`
`A
`
`People say it's a linear search, again,
`
`because it's in relation with the size of the
`
`database. And as you just said,
`
`that complexity is
`
`still linear; so people would say it is a linear
`
`search.
`
`Q
`
`A
`
`It is not a suhlinear search?
`
`It's only sublinear in terms of the size
`
`of the query, which is generally not
`
`the
`
`parameter of -- the relevant parameter.
`
`Q
`
`Is it the case that if we had a piece of
`
`prior art that was linear with respect to the size
`
`of the database but suhlinear with respect to the
`
`size of the query or the pattern,
`
`that that prior
`
`art would not
`
`teach a sublinear search as it's used
`
`in Claim 25?
`
`A Again, if one understands suhlinear to be
`
`in terms in relation to the size of the database,
`
`that would he a -— a linear search.
`
`Q
`
`My question wasn't "if one understands";
`
`I
`
`TSG Reporting - Worldwide - ST?-Tfl2—95Bfl
`Pag¢27 of3$4
`
`
`
`want
`
`to get your understanding of Claim 25.
`
`so you
`
`still haven't answered my question.
`
`I'll ask you
`
`again.
`
`3 I I I I I I - D C C C C - C - C C - D - - C
`
`TSG Reporting - Worldwide - ST?-Tfl2—95Bfl
`Pag¢2EI of3$4
`
`
`
`Q
`
`Let's assume we have a ~— withdrawn.
`
`what I've tried to write there -~ and see
`
`if I wrote it correctly -- is we've got a search,
`
`and the search time is Oifimln}.
`
`Does that —— do you understand what that
`
`refers to?
`
`A
`
`Yes.
`
`Q What would that indicate?
`
`A Well,
`
`that the search time, as a function
`
`of that quantity in parentheses, grows at most
`
`linearly.
`
`Q
`
`Is it the case that if N,
`
`in that example,
`
`refers to the size of the database, do we have,
`
`then, a suhlinear search?
`
`A Well, you have to tell me if M grows too,
`
`or if it's a constant like in your —- in your
`
`previous example?
`
`Q
`
`Let's assume that M is the size of the
`
`query.
`
`359.
`
`Q
`
`Uh-huh .
`
`M, as in Mary. And that N, as in Nancy,
`
`is the size of the database.
`
`A
`
`Okay.
`
`TSG Reporting - Worldwide - ET?“-Tfl2~95Eifl
`F'agc2E|of3$4
`
`
`
`Q
`
`In that instance, does this describe a
`
`search that has -— that is -- has sublinear
`
`execution time when compared to the size of the
`
`database?
`
`A
`
`Your question is incomplete. There's an
`
`assumption you need to specify- Does M grow, or
`
`does N grow, or is it both?
`
`Q Well, my question is if we're presented
`
`with this information, and we want to know is this
`
`search sublinear or linear with respect to the size
`
`of the database, where N is the size of the
`
`database, does that tell us?
`
`A
`
`No. Because you have not told me how
`
`either M or N grows.
`
`The product of f{m}n has to
`
`grow.
`
`For that to grow, either M or N has to grow,
`
`or both. But you need to tell me which one.
`
`Q
`
`If N refers to the size of the database,
`
`and we're trying to assess whether that describes a
`
`search that is sublinear with respect to the size of
`
`the database, can we —— what does that tell us?
`
`A
`
`This tells us that the search time grows
`
`at most linearly in terms of the product
`
`f{m}n.
`
`You
`
`have not told me whether it is M that grows or N
`
`that grows.
`
`The product has to grow, but there are
`
`different ways in which it can grow.
`
`N alone can
`
`TSG Reporting H Worldwide - 8T?—Tfl2-958$
`Page 3:11 of 334
`
`
`
`grow and M is fixed, or M can grow and N grows.
`
`So the answer depends on your assumption.
`
`You need to specify your assumptions.
`
`Q
`
`If we're trying to determine or assess
`
`whether or not this search is sublinear with respect
`
`to database size,
`
`and if N, as in Nancy,
`
`refers to
`
`database size,
`
`then would that indicate that we're
`
`going to be assessing this assuming that N grows and
`
`that M is fixed?
`
`A
`
`It depends on the problem.
`
`In some
`
`problems,
`
`indeed, M could be fixed and N grows.
`
`That's a —e a possibility.
`
`v|—*
`
`'-D
`
`1'\J (II
`
`Ix]
`
`|~—‘
`
`l\) |'\.'|I
`
`|'\) LIJ
`
`1\J uh.
`
`1'\J ILJ'|
`
`Q
`
`If that's the case when we're presented
`
`with this information,
`
`then is the assumption that M
`
`is fixed if we're trying to determine whether it's
`
`suhlinear with respect to the growth of N?
`
`a Different applications could have
`
`different M.
`
`Some applications could have fixed M;
`
`other applications could have M grow like M.
`
`It
`
`TSG Reporting - Worldwide - 3TT*'?fl2~E+5Eifl
`F'agc3Iof3$4
`
`
`
`depends on the applioation.
`
`Q
`
`If we're trying to -- suppose I present
`
`you with an algorithm, and I say to you,
`
`"I want to
`
`know,
`
`in your opinion, does this describe a linear
`
`or a sublinear algorithm with respect to the
`
`database time?"
`
`And I --
`
`A
`
`"Size," you mean.
`
`Right. With respect to the database size.
`
`And I present -- let's take Algorithm 1,
`
`the one that says "O{mn}.“ I present that
`
`Yes.
`
`Do you have an opinion?
`
`Can you repeat your question.
`
`Q
`
`Yeah.
`
`If —— so just to be olear,
`
`in fact,
`
`I'm going to take Exhibit EDUE, and just for
`
`clarity,
`
`I'm going to put a number next
`
`to each of
`
`these formulas we've written down. Starting at the
`
`top. it will be 1, 2, 3, 4. 5.
`
`Let‘s take a look at Formula Number 4.
`
`Assume you're presented that information about a
`
`given search.
`
`and the question is -- withdrawn.
`
`Assume you're given number —— Formula
`
`Number 4 and told that M, as in mother, refers to
`
`TSG Reporting H Worldwide - 8T?—Tfl2-95Bfl
`Page 32 of 33.4
`
`
`
`the size of the query; N, as in Nancy, refers to the
`
`size of the dataset that we're searching over.
`
`And you're asked the question,
`
`"Does that
`
`tell us anything about whether or not
`
`the search is
`
`sublinear with respect to the size of the database?"
`
`A
`
`I don't have enough information to answer
`
`this.
`
`It's really exactly like if you told me,
`
`"Ten
`
`is the sum of two numbers; tell me which ones."
`
`Q
`
`If we know that M refers to the size of
`
`the query and N refers to the size of the database,
`
`and we're interested in whether or not
`
`the —— a
`
`search is sublinear with respect to the size of the
`
`database, does that tell us that it —- with respect
`
`to the size of the database —— that it grows on the
`
`order of N, as in Nancy?
`
`A
`
`M and N oould be related.
`
`You have not
`
`defined the relationship between them.
`
`If M is
`
`fixed,
`
`there's one answer, and -- which I've given
`
`earlier -- it is:
`
`The growth is at most linear.
`
`But if M grows with N as well,
`
`then the
`
`answer is different.
`
`Q
`
`Let's -- I want you to assume that M
`
`refers to the size of the query, N refers to the
`
`size of the database, and that there is no
`
`relationship between them except that N is greater
`
`TSG Reporting H Worldwide - 8T?—Tfl2-958$
`Page 33 of 334
`
`
`
`than M.
`
`The size of the database is greater than M.
`
`A
`
`Okay.
`
`so when you say there's no
`
`relationship, do you assume that M is fixed?
`
`Q Well,
`
`if we're trying to answer this
`
`question,
`
`is that the assumption that should be
`
`made?
`
`A
`
`The way you have said it, mathematically,
`
`to suggest M is fixed.
`
`so I want
`
`to make sure we're
`
`on the same page.
`
`Q
`
`Okay. What about the way I've said it
`
`suggests that M is fixed?
`
`A
`
`Because you said there's no relation
`
`between them.
`
`So that suggests to me that if one --
`
`say N increases, M does not because it's not
`
`related. That's my assumption from what you said.
`
`Q
`
`Okay. Let's assume they're not related.
`
`In other words, were -— is that the
`
`typical problem we're facing,
`
`is that we've got a
`
`database of a certain size that may change, and
`
`we've got different queries of different lengths,
`
`whether the -- the query length is not going to
`
`affect the database size in a typical problem --
`
`A
`
`Q
`
`A
`
`There may be a relationship.
`
`It's possible?
`
`Depends on the problem.
`
`TSG Reporting H Worldwide - 8T?—Tfl2-958$
`Page 34 of 334
`
`
`
`Q
`
`In the typical cases,
`
`is it that they do
`
`not have a relationship?
`
`A
`
`I've encountered both types of problems.
`
`Q Well,
`
`the question isn't whether you've
`
`encountered them.
`
`What's typical?
`
`A Well, both are typical,
`
`I would say.
`
`It
`
`depends on the problem.
`
`Q
`
`Okay.
`
`Now, let's assume that we have a --
`
`a problem we're dealing with where the size of the
`
`query and the size of the dataset are unrelated.
`
`And you're presented with the information
`
`that's in Formula 4,
`
`that the search time is on the
`
`order of M, as in mother,
`
`times N, as in Nancy,
`
`where M refers to the size of the query, N refers to
`
`the size of the data set?
`
`what
`
`information does that tell you about
`
`whether it's linear or sublinear?
`
`A
`
`So based on your assumptions that you laid
`
`out.
`
`I treat M as a fixed number.
`
`Then the answer
`
`is the same I gave earlier:
`
`It would tell me that
`
`the search time is at most linear in N, as Nancy.
`
`Q
`
`Let's assume we've got a search that's
`
`D{f{m}n}, where M and N are unrelated. What does
`
`that -— where N refers to the size of the database.
`
`TSG Reporting H Worldwide - 8T?—Tfl2-958$
`Page 35 of 334
`
`
`
`What
`
`information does that tell us about
`
`whether it's linear or sublinear?
`
`A
`
`Q
`
`A
`
`Same answer.
`
`which is what?
`
`That, again, it would be -- the search
`
`time would be at most linear in M.
`
`Q
`
`Linear as to the size of the dataset that
`
`we're searching over?
`
`A
`
`Exactly.
`
`res.
`
`and that's because you
`
`have treated M as fixed.
`
`So whether it's {m} or
`
`f{m} that appears in the equation is irrelevant.
`
`Q
`
`I've written down Equation or Formula
`
`Number 6, which I'm going to read it as on the order
`
`of N, as in Nancy,
`
`times log of M;
`
`is that right?
`
`A
`
`0
`
`Yes.
`
`Let's aqain assume that N refers to the
`
`size of the database, M refers to the size of the
`
`query.
`
`Does that -- if we have a -- withdrawn.
`
`TSG Reporting - Worldwide - 3TT*'?fl2~E+5Eifl
`Pagc3¢:’:of3$4
`
`
`
`Q
`
`Let me ask you a different problem.
`
`Let's assume we have a search algorithm,
`
`and what it does, it is able to perform its
`
`search —— withdrawn.
`
`Let's assume we've got a search algorithm,
`
`and we're trying to take a pattern or query and see
`
`if it matches any portion of a dataset. And in that
`
`process, our particular algorithm operates such that
`
`it doesn't need to look at every single character in
`
`the dataset.
`
`are you familiar with algorithms of that
`
`i--" n-'|'-‘I
`
`I--" U'|
`
`|—" CT‘:
`
`In-F --J
`
`1-‘ CO
`
`v|—|
`
`‘-53
`
`1'\J (II
`
`Ix]
`
`|~—‘
`
`l\) |'\.'|I
`
`|'\) LIJ
`
`1\J in
`
`!'\J ILJ'|
`
`TSG Reporting - Worldwide - 3TT*'?fl2~E+5Eifl
`Pagc3'.u'of3$4
`
`
`
`A
`
`Q
`
`Yes.
`
`Knowing just that information and no more,
`
`does that tell you one way or the other whether it's
`
`linear or suhlinear?
`
`A
`
`Q
`
`A
`
`No, it does not.
`
`Why not?
`
`Because you have -- when you tell me it
`
`does not
`
`look at every oharaoter, you don't tell
`
`me -- you know, does it look at only half the
`
`characters? Does it look at the square root of
`
`number of characters?
`
`So many possibilities.
`
`You need to
`
`specify that.
`
`Q
`
`If it's the ease that it looks at a
`
`fraction of the characters, one —- one out of eight,
`
`one half, something like that —— does that tell you
`
`whether or not the algorithm is sublinear?
`
`A
`
`If it's a fixed fraotion -— for instance,
`
`you tell me one eighth -- so if the algorithm looks
`
`at one eighth of the number, if it looks at one
`
`eighth of the size of the dataset,
`
`that would be
`
`linear. Again, as the dataset -- as the data size
`
`goes to infinity, if you keep looking at one eighth,
`
`it would he a linear relationship.
`
`It's one eighth
`
`of M.
`
`TSG Reporting H Worldwide - 8T?—Tfl2-958$
`Page 35 of 334
`
`
`
`Q
`
`It would be something like Formula
`
`Number 2, Qikn}?
`
`A
`
`No- Because Formula Number 2 uses the
`
`order of notation:
`
`I would need to use the theta of
`
`N notation.
`
`Q
`
`So you would say that this one is not just
`
`Qinl, it would actually be --
`
`It behaves as a constant times N.
`
`and th