[00:00:00] Speaker 03:
23-1957 Xerox versus SNEP.

[00:00:02] Speaker 03:
Ms.

[00:00:02] Speaker 03:
Easley, please proceed.

[00:00:07] Speaker 03:
Good morning, and may it please the court.

[00:00:09] Speaker 03:
Alexandra Easley from McCool SNEP for Xerox.

[00:00:12] Speaker 03:
The board aired in finding claims of the 439 pen obvious, and there are three specific issues that I would like to discuss today.

[00:00:20] Speaker 03:
So first, the board misconstrued the term context graph by reading the word graph out of the claim term.

[00:00:26] Speaker 03:
Second, the board erred in analyzing limitations 1B and 1C because it did not identify where in the prior art each element is taught, and it relied on theories not in SNAP's petition.

[00:00:40] Speaker 03:
And third, the board relied on a conclusory and generic efficiency rationale for motivation to combine.

[00:00:46] Speaker 03:
Reversal is appropriate for these three reasons, and I would like to begin by discussing claim construction.

[00:00:52] Speaker 04:
Well, before you even get to claim construction, this is a case in which the board made an alternative finding, right?

[00:00:59] Speaker 04:
So even if you prevail on claim construction, the board seems to have addressed that as well, hasn't it?

[00:01:07] Speaker 03:
The board did make an alternate finding for obviousness, yes.

[00:01:10] Speaker 03:
But the problem here is that the board did not give a reasoned analysis for its obviousness finding.

[00:01:15] Speaker 03:
And so therefore, reversal is still appropriate.

[00:01:18] Speaker 01:
But the answer to the chief's question is that, yes, winding only on one would be sufficient.

[00:01:26] Speaker 01:
If we were to disagree with you but agree with the board on the second part, we don't need to reach the claim construction issue.

[00:01:34] Speaker 01:
Yes, Your Honor.

[00:01:34] Speaker 01:
But on that point, is there district court litigation pending now that implicates the claim construction question?

[00:01:42] Speaker 03:
There is a district court case, but it has been stayed until all the IPRs and the related appeals are resolved.

[00:01:51] Speaker 01:
And it involves these claims that would require a claim construction of graph?

[00:01:55] Speaker 04:
Yes, Your Honor.

[00:01:56] Speaker 04:
Are there other claims in this family of patents that would require that construction?

[00:02:02] Speaker 03:
I will have to check and get back with you on rebuttal if that's OK.

[00:02:09] Speaker 01:
So just to bring you back to claim construction, I see at least the issue for me seems to focus on the language in column seven of the patent and to the extent to which that gave the board a basis to broaden its construction of graph, to say it doesn't need a graph.

[00:02:31] Speaker 01:
So could you focus on that language and why that presumably does not do it for you?

[00:02:35] Speaker 03:
Yes, your honor so that what the board relied on is So first the board relied on the statement from the overview which is in column three and It stated that particular statement Reads a context graph as an in-memory model that stores facts and assertions about a user's behavior and interest but describing that particular

[00:03:01] Speaker 03:
that particular disclosure does not mean that it is not a graph.

[00:03:06] Speaker 03:
And we see throughout the specification and in the claims themselves that the word context graph is used, and there's nothing in the specification that indicates that the patentee was trying to render the word graph meaningless.

[00:03:23] Speaker 04:
Well, you have to answer Judge Prost's direct question, though.

[00:03:25] Speaker 04:
Her direct question was, what about the language at column seven?

[00:03:28] Speaker 04:
which says using a typeless approach to data storage.

[00:03:32] Speaker 04:
Why isn't that indicating that a context graph could be a typeless approach to data storage, which suggests it doesn't have to be displayed in a graphical format?

[00:03:45] Speaker 03:
My mistake, I thought you were referring to a little bit earlier in that, in column seven.

[00:03:49] Speaker 03:
So the disclosure at column seven that discusses typeless approach and entity relationship data is merely describing how the underlying data in the context graph is stored.

[00:04:03] Speaker 03:
So Dr. Martin, who was a retained expert in this case, stated that

[00:04:07] Speaker 03:
The context graph absolutely must store its graph-based model data using some type of lower-level storage.

[00:04:18] Speaker 03:
So this passage is merely giving examples of how the underlying data can be stored and identifies three types, entity relationship, unstructured, and typeless.

[00:04:30] Speaker 03:
And if I may just address the database argument that SNAP and that the board rely on.

[00:04:37] Speaker 03:
The board contends in its final written decision that it's not reading the word graph out because it says a person of ordinary skill in the art would understand that the data entries in NHTSA's relational database are conceptually nodes and that the relationship between the entries are conceptually edges.

[00:04:56] Speaker 03:
But there's two problems with that particular argument.

[00:04:59] Speaker 03:
First, that argument was waived.

[00:05:01] Speaker 03:
There's no mention of NHTSA's relational databases until SNAP's reply.

[00:05:06] Speaker 03:
And the prejudice here is real, because SNAP could have made the argument about NHTSA's relational databases in the petition, but it didn't.

[00:05:13] Speaker 03:
And Xerox had no opportunity to offer expert testimony on that particular issue.

[00:05:19] Speaker 03:
And second, as I just mentioned, the reference to databases in column seven is only offered to discuss how the underlying data in the context graph could be stored.

[00:05:31] Speaker 03:
And how the contents of the context graph can be stored does not change

[00:05:36] Speaker 03:
the contents of the context graph itself.

[00:05:39] Speaker 03:
So that particular argument goes to the storage, not the structure.

[00:05:44] Speaker 03:
And I would just again emphasize that we see the word context graph used in the claims themselves.

[00:05:49] Speaker 03:
And in dependent claim 20, we see that it refers to the context graph, which tells us that it's talking about the context graph recited in claim 1.

[00:06:00] Speaker 03:
And a dependent claim 20 specifically includes a built-in assumption that that context graph includes nodes and edges.

[00:06:08] Speaker 02:
Can I take you back to column three?

[00:06:10] Speaker 02:
Because it seems like the forward viewed the statement column three as a definition, but somehow not a helpful definition.

[00:06:20] Speaker 02:
What is your view, and that's the statement around lines 20, a context graph is an in-memory model that stores facts and assertions about a user's behaviors and interests.

[00:06:32] Speaker 02:
Why is that not a definition?

[00:06:33] Speaker 02:
And also very helpful, particularly indicating that

[00:06:37] Speaker 02:
a context graph is what that says, but doesn't necessarily have to be in a graph format.

[00:06:44] Speaker 03:
So first, I'll just start by going back to the word graph.

[00:06:48] Speaker 03:
Again, we see that the word graph is used, and there's nothing in the spec to indicate that the patentee was trying to render that word meaningless.

[00:06:56] Speaker 03:
Second, this is not the only time that we see the spec talking about a context graph is something.

[00:07:01] Speaker 03:
The language in column seven at 26

[00:07:06] Speaker 03:
through 28 states that context-graph 406 is a per-user in-memory graph-based model.

[00:07:21] Speaker 03:
Yes, Your Honor, I agree that context graph 406 is an embodiment.

[00:07:25] Speaker 03:
It is the only depiction we see in the 439 of a context graph, and it shows that graph with nodes and edges.

[00:07:33] Speaker 03:
But I will just note that the particular statement that you pointed out in column three is also discussing embodiments.

[00:07:42] Speaker 03:
If you look at the overview

[00:07:45] Speaker 03:
Column 2, line 48, it begins with, embodiments of the present invention solve the problem of efficiently developing context-aware systems.

[00:07:54] Speaker 03:
And then it goes on to describe the embodiments.

[00:07:57] Speaker 03:
So again, we see nothing in the specification that indicates that the word graph should be read out of the term context graph.

[00:08:05] Speaker 03:
And in fact, everything in the specification, both the statements and the references to the graph itself and the depictions and the figures,

[00:08:15] Speaker 03:
as well as a claim language confirmed that context graph is, in fact, a graph-based model.

[00:08:20] Speaker 04:
Now, if we agree with you that context base is graph-based, what do we do with the PTO's alternative finding that, nonetheless, it's disclosed by the, I'm just going to call it DG reference, because you can't pronounce the long name.

[00:08:34] Speaker 04:
So the DG reference, which is incorporated into the other piece of prior art that is cited.

[00:08:40] Speaker 03:
Yes, Your Honor.

[00:08:41] Speaker 03:
So there's two reasons that the alternate finding based on the DG reference is insufficient.

[00:08:48] Speaker 03:
So first, NIST cites the DG reference as an example of a method for extracting pertinent information from unstructured data.

[00:08:57] Speaker 03:
So the DG reference is not cited as an example of a method for using a graph-based model to store contextual data.

[00:09:05] Speaker 03:
And DG does not disclose

[00:09:07] Speaker 03:
a graph, and there's no mention of graph anywhere in that reference.

[00:09:10] Speaker 04:
Well, but the problem is, I think, and I don't know if I remember it right, off the top of my head, it's figure eight, right?

[00:09:16] Speaker 04:
They point to figure eight, and they say, oh my goodness, that looks exactly like 406 in your patent, and it pretty much does, and that is a disclosure to a skilled artisan of a graph.

[00:09:28] Speaker 03:
Yes, Your Honor.

[00:09:29] Speaker 03:
SNAP does point to Figure 8, and they include that in their brief as well.

[00:09:36] Speaker 03:
But the problem with Figure 8 is that it is not depicting a context graph.

[00:09:42] Speaker 03:
It is depicting an illustration that's used to show that the word Apple can refer to either a company

[00:09:48] Speaker 03:
or a person's name or a fruit.

[00:09:51] Speaker 03:
So the DG reference.

[00:09:54] Speaker 04:
It doesn't look like it has the nodes and the edges and all the other stuff that is in Disha of the context graph as you described it in the past.

[00:10:02] Speaker 03:
I agree that it does look very similar to context graph 406.

[00:10:06] Speaker 03:
But I think you have to look at the reference and what it's actually stated about.

[00:10:11] Speaker 04:
But I can't look at the reference.

[00:10:12] Speaker 04:
Because you know what?

[00:10:13] Speaker 04:
I mean, I can.

[00:10:13] Speaker 04:
But that doesn't help you.

[00:10:15] Speaker 04:
Because the standard of review for me on what a piece of prior art discloses is substantial evidence.

[00:10:21] Speaker 04:
And if you honestly think that I can look at a figure that looks exactly the same as your figure, which you say is a context graph, and if you think I'm going to say there was no substantial evidence for the PTO to conclude that the figure in the prior art is also a context graph, you can't win that argument.

[00:10:41] Speaker 04:
Do you see?

[00:10:41] Speaker 04:
You can't win that.

[00:10:42] Speaker 04:
That's not winnable.

[00:10:44] Speaker 03:
Well, I think the issue here with the DG reference is that the board has to provide a reasoned analysis once again.

[00:10:51] Speaker 03:
And it can't cherry pick figure eight and ignore what the reference actually says about it.

[00:10:58] Speaker 03:
And the board also never explains why or how a person of ordinary skill and art

[00:11:02] Speaker 03:
would be motivated to look to DG's machine-implemented data extractor to modify NITS's storage.

[00:11:10] Speaker 03:
In fact, all it says is NITS and the DG reference could be combined in a manner claimed.

[00:11:17] Speaker 03:
And it provides no explanation about how one would actually combine the references.

[00:11:21] Speaker 03:
It just says that a person of ordinary skill and art would know implementation details that could be used to combine the two references.

[00:11:30] Speaker 03:
So that type of generic and conclusory analysis for motivation to combine is insufficient.

[00:11:36] Speaker 03:
And I think that's established in the personal web case, which states that the fact that two references can be combined is just not sufficient.

[00:11:46] Speaker 01:
Unless there's any more questions about that.

[00:11:48] Speaker 01:
Just to follow up on that point, you say, I think in your brief, that the simple fact that

[00:11:53] Speaker 01:
a person skilled in the art would want a system that was more efficient is not sufficient.

[00:11:59] Speaker 01:
Is that really what our law says?

[00:12:02] Speaker 01:
That this is going to be more efficient?

[00:12:04] Speaker 01:
Is that not a motivation to combine?

[00:12:07] Speaker 03:
Well, I think there are several issues with the motivation to combine.

[00:12:11] Speaker 03:
So SNAP says that these two references, that NITS and NICANIN would be combined to preserve resources.

[00:12:18] Speaker 03:
And the board adopts that reasoning, which is error

[00:12:22] Speaker 03:
for several reasons.

[00:12:23] Speaker 03:
And first, it relies on this conclusory efficiency rationale.

[00:12:28] Speaker 03:
But all it states is that a person with ordinary skill in the art would have recognized that NITS was inefficient because it pushes all the data from the context graph to the recommender, and therefore would have looked at Nikenen, which only pushes the change data.

[00:12:44] Speaker 03:
But what the board doesn't address is the fact that Nikenen is all about

[00:12:51] Speaker 03:
security and ensuring that context data is secure and remains private.

[00:12:57] Speaker 03:
And so Nikenen would actually decrease efficiency.

[00:13:01] Speaker 03:
But because the board doesn't provide any more reasoned analysis, other than just saying that it would be more efficient, that's error.

[00:13:10] Speaker 03:
Because the purpose of Nikenen's registration is to prevent unauthorized access to users' private data.

[00:13:18] Speaker 03:
APPX 1128.

[00:13:21] Speaker 03:
And Nikonin includes an additional authorization layer, which is used to protect that data.

[00:13:27] Speaker 03:
And it would decrease efficiency due to an outlay of resources because of that particular layer.

[00:13:34] Speaker 03:
So the efficiency argument doesn't work because the reference that they're turning to, Nikonin, would actually decrease efficiency.

[00:13:42] Speaker 03:
And it's also a conclusory reasoning.

[00:13:46] Speaker 03:
There's no more explanation offered by the board, other than merely stating that a person of ordinary skill and art would want a more efficient system.

[00:13:53] Speaker 03:
And we don't believe that's appropriate or sufficient for motivation to combine under the active video case and personal web.

[00:14:04] Speaker 03:
Would you like to save some time for rebuttal?

[00:14:07] Speaker 04:
Yes, sure.

[00:14:25] Speaker 00:
Thank you, Your Honors.

[00:14:27] Speaker 00:
Jarczajkowski on behalf of SAHMP.

[00:14:29] Speaker 00:
I think as we just heard from counsel with respect to the patent and the context graph claim construction, as was rightly noted by Your Honors, the embodiment 406 includes an entity data relationship

[00:14:47] Speaker 00:
it's not limited to a graph-like structure, as we just heard from counsel in terms of graph being something that is in one figure and not even described in that figure.

[00:14:57] Speaker 04:
You don't think 406 is limited to a graph-like structure?

[00:14:59] Speaker 04:
What about column 7, line 26?

[00:15:03] Speaker 04:
Context graph is a per-user, in-memory, graph-based model.

[00:15:09] Speaker 00:
It is clearly a graph.

[00:15:10] Speaker 00:
We can't remove the word graph from context graph.

[00:15:13] Speaker 00:
does not mean that the word context graph, which does not.

[00:15:16] Speaker 00:
There's no evidence in the record that there's a plain, nor do you mean a context graph.

[00:15:20] Speaker 00:
So when we look at the words context graph in the court's cases, that we don't disembodied one word from a combined word here, we look to the fact that the context graph in this patent is broadly disclosed.

[00:15:33] Speaker 00:
It's not disclosed as having nodes and edges.

[00:15:35] Speaker 00:
And frankly, when it was intended to be claimed, it was claimed in claim 20.

[00:15:37] Speaker 04:
I don't understand.

[00:15:38] Speaker 04:
Context graph is a per user in memory graph based model.

[00:15:43] Speaker 04:
That means it has to have a graph.

[00:15:46] Speaker 04:
You've read out the graph.

[00:15:48] Speaker 04:
Why does it not have to have a graph?

[00:15:51] Speaker 00:
John, we do not disagree with the fact that it has to have a graph.

[00:15:54] Speaker 00:
But as you have in column seven, the context graph 406 may store data according to different data models.

[00:16:02] Speaker 00:
It's not just a node tree architecture, including data models for entity relationship data and unstructured data.

[00:16:09] Speaker 00:
If it was simply one structure, as is posited by Xerox, then it would not be able to have, as it's disclosed in column 7, this in all lines 42 through 46, AAD1, it would not be able to have these different data models.

[00:16:26] Speaker 02:
I'm confused by what your position is.

[00:16:28] Speaker 02:
I thought 406 is an embodiment that you agree is in a graph form.

[00:16:35] Speaker 02:
but that context graph as a clean term is used more broadly, and it doesn't have to be a graph.

[00:16:41] Speaker 02:
Did I misunderstand your position?

[00:16:42] Speaker 00:
No, you did not misunderstand your position.

[00:16:47] Speaker 02:
OK, so then how can we read graph out of the term context graph?

[00:16:53] Speaker 00:
I don't think graph is limited to, I guess, the dispute or the crux of that clank construction issue, which again is that

[00:17:02] Speaker 00:
graph is not limited to having nodes and trees.

[00:17:05] Speaker 00:
It would be like saying one figure is definitional when the patent is replete with repeating consistent usage.

[00:17:12] Speaker 00:
That context graph is a very, let's call it unstructured and broad-based storage mechanism.

[00:17:20] Speaker 00:
The data structure is broad, as shown in column three and repeated in column seven.

[00:17:25] Speaker 04:
I'm not following your argument.

[00:17:28] Speaker 04:
I think I follow it better in your briefs than I do right here.

[00:17:32] Speaker 04:
I mean, when it says a context graph, if nothing else, if it said nothing else of the spets, it was just the words context graph, I would think that has to be a graph.

[00:17:43] Speaker 04:
Like suppose it doesn't give you any specific details about what a context graph is.

[00:17:48] Speaker 04:
It just says, and then there's a context graph.

[00:17:50] Speaker 04:
I think that all normal humans would think that has to be a graph.

[00:17:55] Speaker 04:
And so you're starting from a proposition that you've

[00:17:58] Speaker 04:
There's a word used, and that word has a meaning that people would understand.

[00:18:03] Speaker 04:
And so what you really need is to show that the word isn't limited to what the normal meaning of that word would be somehow.

[00:18:11] Speaker 04:
How have you shown that?

[00:18:14] Speaker 00:
Your Honor, respectfully, I would say the first time the patent talks about it, as it does in column three, is it tells you what a context graph is.

[00:18:22] Speaker 00:
It's an in-memory model that stores facts and assertions about a user's behavior and interests.

[00:18:25] Speaker 04:
Yes, but it uses the word graph.

[00:18:27] Speaker 04:
So why is the word graph not subsumed as one of the limitations to what a context graph is?

[00:18:35] Speaker 00:
I believe, Your Honors, that there's no evidence that says a graph must be limited to nodes and trees.

[00:18:44] Speaker 00:
And as I said, there's even claim 20 that added a limitation after rejection that actually added

[00:18:50] Speaker 04:
The elements of... It didn't add the nodes.

[00:18:53] Speaker 04:
It doesn't render it superfluous.

[00:18:56] Speaker 00:
Claim 20 added... Yeah, nodes and edges.

[00:18:59] Speaker 04:
No, it didn't just add nodes and edges.

[00:19:01] Speaker 04:
It added something else.

[00:19:02] Speaker 00:
It did.

[00:19:03] Speaker 00:
You're right.

[00:19:03] Speaker 00:
A probological change.

[00:19:04] Speaker 04:
Claim 20 would not be superfluous if we understood nodes and edges to be part of a graph.

[00:19:10] Speaker 00:
Could you repeat that, Your Honor?

[00:19:11] Speaker 04:
Claim 20 would not be superfluous if we understood the term context graph to already include nodes and edges.

[00:19:18] Speaker 00:
not suggesting the claim 20 superfluous.

[00:19:20] Speaker 00:
I'm just saying when the applicant wanted to call out, because in particular, that's disjunctive in claim 20.

[00:19:25] Speaker 00:
When the applicant wanted to call out nodes and edges, the applicant called out nodes and edges in the claims.

[00:19:31] Speaker 00:
And in particular, I mean,

[00:19:32] Speaker 00:
you know, what we face in this litigation.

[00:19:34] Speaker 00:
I mean, if you looked at, you know, when originally Xerox's view of this asset, the patent, when we first got into litigation in the infringement contentions, as we cited in our brief, they had the same view we did as to the breadth of this patent and the breadth of the context graph.

[00:19:50] Speaker 00:
And it's only now in hindsight, with respect to the PTAB, that they've also all of a sudden reverted that the word graph has a different meaning and that it's somehow limited.

[00:19:58] Speaker 00:
I mean, that's kind of putting us in a different position.

[00:20:00] Speaker 04:
I don't understand that they said the word graph.

[00:20:02] Speaker 04:
The word graph means a graph.

[00:20:04] Speaker 04:
I don't understand your argument at all, because a graph is a graph.

[00:20:11] Speaker 04:
I'm just very confused.

[00:20:12] Speaker 04:
I'm sorry.

[00:20:13] Speaker 00:
Well, Your Honors, even if you, again, if there's disagreement with respect to graph, as was mentioned during argument,

[00:20:21] Speaker 04:
So here, this sentence you keep harping on in claim three.

[00:20:25] Speaker 04:
A context graph is an in-memory model that stores facts and assertions.

[00:20:30] Speaker 04:
You want me to read graph out of that, because you think everything that came afterwards with the in-memory model somehow eliminates the word graph from the sentence, because you think it's broader.

[00:20:41] Speaker 04:
It's defining it as something broader than just a graph-based architecture.

[00:20:45] Speaker 04:
But imagine I say, instead of this, I say a bicycle is a vehicle

[00:20:50] Speaker 04:
that transports a person across the street or down the road.

[00:20:54] Speaker 04:
The word bicycle is not suddenly any vehicle that transports the person.

[00:20:59] Speaker 04:
The bicycle actually has an implied meaning.

[00:21:03] Speaker 04:
The rest is a sense of purpose.

[00:21:05] Speaker 04:
It's not even a structure that causes you to think the word graph is no longer imported into that sentence.

[00:21:11] Speaker 04:
Just like my word bicycle would still conjure a particular image, the word graph conjures a particular image.

[00:21:18] Speaker 04:
It's got to be graph based.

[00:21:20] Speaker 04:
A non-graph-based thing would not be a graph.

[00:21:23] Speaker 00:
I would agree that in general English usage, graph may have some connotation in particular in mathematics.

[00:21:30] Speaker 00:
I think the one of ordinary skill in the art is the board found.

[00:21:33] Speaker 00:
And I think also in particular with Column 7, because then you're looking for data structures.

[00:21:37] Speaker 00:
As a computer science, one of ordinary skill in the art, which is someone with a computer science degree with two years and wasn't objected to by Xerox,

[00:21:45] Speaker 00:
looking at this, and then they see that the disclosure in Column 7 says it can store data according to different data models.

[00:21:51] Speaker 00:
It doesn't say I'm going to be storing data in nodes or storing data in edges.

[00:21:56] Speaker 00:
It doesn't say that.

[00:21:57] Speaker 00:
It says we're going to store entity relationship data and unstructured data.

[00:22:01] Speaker 00:
That sentence alone would tell someone that I'm not storing things with a specific methodology.

[00:22:10] Speaker 00:
That's what they have as a disclosure.

[00:22:13] Speaker 02:
The board did say, as you just alluded to, at A15, that a person of ordinary skill in the art would not have understood the 439 patent to be limiting in the sense that it requires a graph to store information about a user in the form of notes and edges.

[00:22:28] Speaker 02:
The board clearly understood the knowledge of a person of ordinary skill that way.

[00:22:33] Speaker 02:
What is the board's report for that?

[00:22:35] Speaker 02:
Is it just their reading of the patent, or are they relying on something else you can see in the record?

[00:22:40] Speaker 00:
No, I think as you see, Your Honor, as you cited to in 815, they point to, and they say see for example, but they point to that same citation that I just read on column 7, 36 through 40.

[00:22:55] Speaker 00:
I mean, that sentence, there's not many ways to read that sentence and change it to say the data structure is a broader data structure.

[00:23:04] Speaker 00:
It's not limited to a specific data structure.

[00:23:09] Speaker 00:
And they go further at the end of the page, and they do cite the word graph is not superfluous because of the relational database aspect.

[00:23:19] Speaker 00:
Once we have a relational database, a relational database does not necessarily mean or doesn't mean that I'm going to have nodes and edges.

[00:23:29] Speaker 02:
We asked your colleague about whether the patent scheme is serving in district court litigation.

[00:23:34] Speaker 02:
Is there some need for us to determine what the correct construction here is if, just for the sake of argument, if we were persuaded by your alternative argument about the application of it here?

[00:23:47] Speaker 00:
Yes, there is, Your Honor.

[00:23:49] Speaker 00:
In fact, on Rule 12, Judge Barat denied our 101 motion.

[00:23:55] Speaker 00:
Actually, on Step 1, although perhaps it was a Step 2 analysis by Judge Barat, he denied on Step 1 saying there are factual issues.

[00:24:04] Speaker 00:
As I said, it's probably more Step 2, but in Step 1 he did it.

[00:24:07] Speaker 02:
If we were just for the sake of argument, if we were to affirm the board finding that the patent is unpatentable,

[00:24:13] Speaker 02:
Isn't that the end?

[00:24:15] Speaker 00:
That is the end.

[00:24:16] Speaker 00:
Context graph was an issue for Judge Barat.

[00:24:18] Speaker 00:
That's all I'm saying from a client infrastructure.

[00:24:20] Speaker 00:
But you're right.

[00:24:21] Speaker 00:
If we affirm, even on the alternative grounds, that's the end of the story.

[00:24:25] Speaker 04:
I guess I'm going back and forth with my law clerks.

[00:24:27] Speaker 04:
I'm trying to understand something.

[00:24:29] Speaker 04:
Are you saying that you can have a graph without nodes and edges?

[00:24:34] Speaker 04:
I don't understand that.

[00:24:37] Speaker 04:
Are you saying that there can be a graph-based model

[00:24:40] Speaker 04:
that is still a graph, as an ordinarily skilled artisan would understand what a graph is, but it doesn't have nodes and edges.

[00:24:49] Speaker 00:
Yes, I guess there's two ways of looking at it, Your Honor, for the computers.

[00:24:54] Speaker 00:
I guess maybe I'm burdened from having a computer science degree.

[00:24:56] Speaker 00:
There's two ways of looking at it.

[00:24:58] Speaker 04:
I hate to tell you this, but I also have one, but keep going.

[00:25:00] Speaker 00:
No, is that the context graph, first of all, is a phrase together.

[00:25:05] Speaker 00:
We're not isolating out the word graph, right?

[00:25:08] Speaker 00:
And so when it describes a context graph... Just answer my question.

[00:25:11] Speaker 04:
Is there another form of graph in this technology that doesn't have nodes and edges?

[00:25:15] Speaker 04:
Because you're skirting all over the place, and I'm trying to figure out

[00:25:18] Speaker 04:
your construction, and I don't understand your argument.

[00:25:21] Speaker 04:
But I want to.

[00:25:22] Speaker 00:
I understood your honor.

[00:25:24] Speaker 00:
I apologize.

[00:25:26] Speaker 00:
It's, again, just a context graph in this patent, how it's described.

[00:25:32] Speaker 04:
I'm not asking about this patent.

[00:25:34] Speaker 04:
You're saying, oh, the PTO found it doesn't have to have nodes and edges.

[00:25:38] Speaker 04:
And Judge Stark pointed to the sentence.

[00:25:41] Speaker 04:
And I'm trying to understand, well, does that mean they say it doesn't have to be a graph-based architecture?

[00:25:48] Speaker 04:
because they've concluded it doesn't have nodes and edges.

[00:25:51] Speaker 04:
And then I'm thinking to myself, or are they just saying it doesn't have nodes and edges, but it's still a graph-based architecture?

[00:25:57] Speaker 04:
And then I'm thinking, well, then what does it look like?

[00:25:59] Speaker 04:
I don't understand.

[00:26:00] Speaker 04:
I'm trying to understand this technology.

[00:26:02] Speaker 04:
So what is your argument that it doesn't require a graph-based architecture when it says context graph?

[00:26:08] Speaker 04:
Or is your argument that it requires a graph-based architecture, but a graph-based architecture could have many forms, forms that are still graph,

[00:26:18] Speaker 04:
but not requiring nodes and edges.

[00:26:20] Speaker 00:
I think the argument is obviously the latter.

[00:26:25] Speaker 00:
It would fit more in the description of the latter in the sense of it's still obviously a context graph, which is a graph.

[00:26:32] Speaker 04:
I didn't ask if it was a context graph.

[00:26:33] Speaker 04:
I said it's a graph-based architecture.

[00:26:35] Speaker 04:
I'm being very clear.

[00:26:36] Speaker 04:
Stop skirting me.

[00:26:37] Speaker 00:
I apologize.

[00:26:38] Speaker 00:
I'm not trying to skirt.

[00:26:41] Speaker 00:
To extent, we're looking at a graph-based architecture in the context of this patent.

[00:26:44] Speaker 00:
The graph-based architecture does not require nodes and edges.

[00:26:48] Speaker 04:
Does that mean it doesn't require a graph?

[00:26:50] Speaker 04:
Is there another graph-based architecture, graph-based architecture, that doesn't require nodes and edges in this field?

[00:26:59] Speaker 00:
A relational database, for example, when you look at it too, a relational database doesn't require nodes and edges, or if you want to look at a relational database... A relational database is not a graph.

[00:27:08] Speaker 04:
It's not graph-based architecture.

[00:27:10] Speaker 04:
That's not graph.

[00:27:11] Speaker 04:
If it doesn't require the nodes and integers, it's not graph.

[00:27:13] Speaker 00:
Well, this patent's telling you that this graph is a relational entity of relationships.

[00:27:20] Speaker 00:
That that's included.

[00:27:21] Speaker 04:
OK, well, relational database is a broader label.

[00:27:27] Speaker 04:
A graph-based architecture is one of many forms of a relational database, correct?

[00:27:36] Speaker 04:
Subset.

[00:27:37] Speaker 00:
It could be, Your Honor, it could be a subset.

[00:27:40] Speaker 00:
I don't know that it is in all instances a subset.

[00:27:52] Speaker 04:
OK, do you have anything further?

[00:27:54] Speaker 00:
I mean, unless you have further questions, again, even if you were to reverse on the context graph, as you said, the alternative findings by the board are supported by substantial evidence with the DG reference, let alone the combination of Nitz and Eichanan.

[00:28:11] Speaker 00:
And so in either direction, even if you were to find that context graph would have a different construction,

[00:28:18] Speaker 00:
We don't think it's of any import.

[00:28:20] Speaker 02:
Your office's counsel says that you waived the argument that MITS's relational database are the nodes and edges.

[00:28:28] Speaker 02:
Did you waive it?

[00:28:29] Speaker 02:
And if not, how could you be sure that you did not waive that argument?

[00:28:33] Speaker 00:
No, we did not.

[00:28:35] Speaker 00:
It was disclosed.

[00:28:36] Speaker 00:
It was disclosed in particular, for example, in Dr. Almorov's reply declaration.

[00:28:42] Speaker 00:
They didn't depose Almorov.

[00:28:43] Speaker 00:
They didn't move to strike.

[00:28:45] Speaker 02:
Isn't the reply too late?

[00:28:47] Speaker 00:
No, because it's consistent with what is in the petition.

[00:28:51] Speaker 00:
And all the way, again, there was no motion to strike.

[00:28:54] Speaker 00:
They moved to strike many items.

[00:28:56] Speaker 00:
That was not Dr. Almorov, and they didn't depose Dr. Almorov.

[00:28:59] Speaker 00:
They decided not to depose him.

[00:29:01] Speaker 00:
And so note that this is not a new ground, no deposing.

[00:29:05] Speaker 00:
It's just providing another reason why the art discloses context graph.

[00:29:11] Speaker 00:
And so, no, it's not a way of drama.

[00:29:17] Speaker 04:
Thank you very much.

[00:29:19] Speaker 00:
Thank you.

[00:29:30] Speaker 04:
Ms.

[00:29:30] Speaker 04:
Easley, are you burdened by a computer science degree?

[00:29:33] Speaker 03:
I am not, Your Honor.

[00:29:34] Speaker 04:
Oh, well, then you're maybe not smart enough like the rest of us to understand this case.

[00:29:38] Speaker 03:
That might be the case.

[00:29:40] Speaker 03:
But I don't think so.

[00:29:41] Speaker 03:
And I just want to start by first addressing your question about the 439 family members.

[00:29:47] Speaker 03:
There are no family members of the 439 that were asserted in the district court case.

[00:29:52] Speaker 03:
So it's just the 439 pattern.

[00:29:55] Speaker 03:
regarding the DG reference.

[00:29:57] Speaker 04:
I hope you know that wasn't a slam on you.

[00:29:59] Speaker 03:
Yeah, no, I got you.

[00:30:01] Speaker 03:
Thank you for that.

[00:30:03] Speaker 03:
They point to figure eight in the DG reference, but that's not enough.

[00:30:07] Speaker 03:
It doesn't have nodes and edges.

[00:30:09] Speaker 03:
And nodes and edges aren't about storage.

[00:30:11] Speaker 03:
They go to structure.

[00:30:12] Speaker 03:
And so they have not pointed to a context graph.

[00:30:16] Speaker 03:
And instead, they just rely on

[00:30:18] Speaker 03:
Conclusory reasoning, and that's insufficient.

[00:30:21] Speaker 04:
Can you answer my question?

[00:30:22] Speaker 04:
If the word context graph doesn't require nodes and edges, is there anything else in this field that's been articulated by anyone that would constitute a graph in this sense?

[00:30:37] Speaker 03:
No, Your Honor.

[00:30:39] Speaker 03:
And in fact, Dr. Martin said that a person of ordinary skill and art would understand a graph-based model to have

[00:30:47] Speaker 03:
Nodes and edges and face fix expert also said that he understood context graph to be a graph based model.

[00:31:00] Speaker 02:
would have understood that data entries in a database conceptually are the nodes, and the relationships between the table entries conceptually, such as pointers and keys, are conceptually the edges.

[00:31:14] Speaker 02:
Isn't there substantial evidence to support that that is the way a person with a skill in AR would understand what is meant by context graph in this pattern?

[00:31:24] Speaker 03:
Your Honor, I think the Board, that's the extent of their analysis, and they don't explain why or identify their support.

[00:31:30] Speaker 03:
They merely cite to SNAP's reply, which is why I mentioned earlier that this argument has been waived.

[00:31:39] Speaker 03:
And if we look at the petition, we can see that the SNAP said, if context graph is

[00:31:47] Speaker 03:
construed to require graph-based qualities, then the board would need to go beyond NITS.

[00:31:52] Speaker 03:
So it agreed that NITS didn't disclose graph-based properties, which is where we get the relational databases.

[00:32:02] Speaker 03:
And then in the institution decision, nothing was said about the databases.

[00:32:07] Speaker 03:
is YAROC did not address the databases and the patent owner's response, since it hadn't yet been raised.

[00:32:13] Speaker 03:
And it did not come up till SNAP's reply.

[00:32:16] Speaker 03:
So again, we would argue that that has been waived, and that the mention of databases M439 merely goes to storage of the underlying data.

[00:32:25] Speaker 03:
And finally, I will just say that we will rest on our briefs for the arguments about limitation 1C.

[00:32:32] Speaker 04:
All right.

[00:32:32] Speaker 04:
I thank both counsel.

[00:32:33] Speaker 04:
The case is taken under submission.