Re: Two examples of semi structured data.

From: mAsterdam <mAsterdam_at_vrijdag.org>
Date: Wed, 25 Aug 2004 01:05:57 +0200
Message-ID: <412bc9db$0$78279$e4fe514c_at_news.xs4all.nl>


Jan Hidders wrote:
> mAsterdam wrote:

>>Jan Hidders wrote:
>>>mAsterdam wrote:
>>>>Jan Hidders wrote:
>>>>>mAsterdam wrote:
>>>>>>Jan Hidders wrote:

[...]
> You implied that Serge Abiteboul did by accusing him of speaking about
> "potentially meaningless data". But anwyay, then we agree. So remind me,
> what exactly was again the problem that you had with his paper?

I made no accusation. The paper is written from a perspective I do not find very interesting. There are many papers which do the same, and I don't have any problems with that.
We do not all walk the same road all the time. You do find it interesting. That's fine.

You invited me to to be more specific.
In trying I ranted about the widespread use of the word data as potentially meaningless.
This termininological habit obscures many - to me - interesting issues. I am very interested in the source of that habit. I suspect it demarcates a school boundary. That is why I asked you if you knew a good source. Maybe you missed that question.

Can shared data be meaningless?

>>>>Let's zoom in here:
>>>>The store/forward structure carries the signs. Can we or can we not
>>>>change that structure whithout affecting the stored/transported signs?
>>>>Can we change the signs without affecting the conveyed meaning?
>>>
>>>Not without agreeing first on what the new interpretation is of the
>>>structures and the signs.
>>
>>Yep.
>>Let's zoom in some more:
>>Not caring about meaning, we can change structure without affecting the
>>signs:
>>"The secret agent walked into the room." "The agent walked into the
>>secret room." and we can change both signs and structure, and yet convey
>>the same meaning:
>>"At 4:05 pm tomorrow the doors will open" "The doors will open tomorrow
>>at 16:05"

>
> Yes, but to be really orthogonal that should *always* be possible. And as
> you just showed, it isn't, so they are not.

I admit that it wasn't easy to come up with examples, and they are not even good ones. Would you (or maybe somebody else - Hey! Are we the only people who think this sub-thread is worthwile? I enjoy it, but if nobody joins I'll take it off-line.) be so kind as to try to provide an example where it is impossible to make a change in a structure while preserving meaning? It's not all that clearcut (to me).

> But perhaps you have a
> different idea of what "orthogonal" exactly means. Note that we were
> discussing your following claim:
>

>>>This structure does not determine meaning, neither is it determined by 
>>>meaning. Buzzword bingoish: it is orthogonal to meaning.

>
> In your example you yourself just showed that changing the structure
> changed the meaning. So what then do you mean when you claim the structure
> does *not* determine meaning?

Just what it says.

The structure does not determine the meaning. For this claim it is trivially easy to give an example.

This is a structured chunk of signs:

<name>Jantje</name><amount>5</amount>
<name>Pietje</name><amount>7</amount>
<name>Keessie</name><amount>2</amount>

What does this chunk need to get meaning? Well, I could tell you (don't need to really do that, I think).
This is more important:
This chunk could be *all* that is stored between programs or incarnations of a program - BTW, it is not unlike the registry. Now my argument:
Though the structure is complete for this purpose, the meaning is not determinable from it without extra means.

The structure does not determine the meaning.

>>>>>I could send a simple string with flat text or I could add structure
>>>>>in the form of XML mark-up and then send it to you. If we have agreed
>>>>>before on what this markup means then the added structure will add
>>>>>additional meaning.
>>>>
>>>>??? Are you suggesting we can add meaning without changing the
>>>>agreement?
>>>
>>>Of course we can. Why does that surprise you? If I first send you
>>>"Harry, 43" and then "<name>Harry</name><shoesize>43</shoesize>" then
>>>you will probably know more after the second message. Note that no
>>>change of agreement is necessary.
>>
>>Are you telling me the second message relates to same Harry and 43 as
>>the the first? It contains some of the signs which are also in the
>>second message. Do they mean the same? Their closeness suggests it. Your
>>use as an example suggests it. If I had to make a decision based on it
>>I'ld say they do.

>
> They might, and that is sufficient for the point we're discussing here.

Aside: Some of your deletions suggest to me that you are not discussing the same thing I am (to be clear: I am not, not even slightly, suspecting bad intentions here, I simply wouldn't have answered in that case). I am not overly bothered when somebody else appears to be not interested in what I think is important. We don't all walk the same road.

They might, and that is sufficient for being maybe related to facts. Proto-data.

> But what you should be focusing on is not the meaning of the symbols but
> the relationship between them that is communicated.

Both. But you can't at the same time. Hence the back and forth that is needed to get the cabbage, the goat, and the wolve across. Focus on the relationship blurs the symbolvision and vv. Received on Wed Aug 25 2004 - 01:05:57 CEST

Original text of this message