You are on page 1of 2

Daniel Petrariu = I completely agree but the thing that I find circular there right is that we are

saying is that definitely then there is more data in the text.

Than in just the data right we've added something there in there to form which means that it's not
processable and now we're trying to get that extra detail.

We're trying to effectively try and figure out what it is that the person has said in the narrative
that interprets the water. I hate that phrase whatever the paper was based on.

I think that you've got a group of people who would argue that one of the things.

We may have better luck with is not is if we put that in there to form and also put it in some sort
of machine readable form rather than try and reverse engineering.

So if there's some way that we can start inventing more semantic information in the paper.

Two options to make that extra data that's being added evidence then that might also be an
approach.

Geoffrey Bilder = We're still in the format that represents what the paper today represents it
doesn't have to be with readable tax.

It should at least have on top of that write better machine readable semantic explanation of what
it actually says.

But I think the opportunities that something like this has goes far beyond what we can actually
imagine in terms of solutions that can be found.

I mean it's pretty much a tool that accelerates the work of scientists.

Because it can bring the right content to these people at the right time and really be a great tool to
make much bigger discoveries and what we talking today.

Daniel Petrariu = And I think this illustrates an aspect of a lot of the things.

At least I work on and that is an infrastructure.

And I consider the you know that the requirements are doing text and data mining.
On a large scale require fundamental changes in infrastructure infrastructure building
infrastructure generally faith-based activity.

And what I mean by that is that the problem with infrastructure is it's not useful until it's big
enough to be useful.

So everybody who's participating in it in the early days is doing so based on possibly scant
evidence but a firm belief that eventually will become useful.

So for instance the DOI system when there were two publishers implementing being wise was
not useful to anybody.

It only became useful when a critical mass of people were adopting the Amar's.

Same thing applies to physical infrastructure such as lighting her phones or whatever phones
aren't useful if only two people have them they only become useful.

If you have a lot of people text and data mining I think is probably under it has a similar property
which is that until we have enough data of text mind and enough data to analyze.

The true benefits of come from that.

So I you know I agree it's actually really hard to predict what will come of it largely because it's
sort of a generative process a network effect you.

You might also like