<html>
<head>
<style><!--
.hmmessage P
{
margin:0px;
padding:0px
}
body.hmmessage
{
font-size: 10pt;
font-family:Tahoma
}
--></style>
</head>
<body class='hmmessage'>
<div><div><div><div><meta http-equiv="CONTENT-TYPE" content="text/html; charset=utf-8">
<title></title>
<meta name="GENERATOR" content="OpenOffice.org 3.2 (Linux)">
<style type="text/css">
<!--
@page { margin: 20mm }
P { margin-bottom: 2.12mm }
-->
</style>
<p style="margin-bottom: 0mm; font-style: normal; font-weight: normal; widows: 2; orphans: 2; ">
<font color="#000000"><font face="Tahoma"><font size="2">Yes Martijn there
are weak classifications, everything that is not mapped is not
represented in the database. For example the classification in five
groups of tags and in the subgroups in the map feature list is not
expressed in the database. OSM database is missing abstraction
levels.</font></font></font></p>
<p style="margin-bottom: 0mm; widows: 2; orphans: 2; "><font class="Apple-style-span" size="2">Moreover all
tags are treated seamlessly, even when they are properties or
subclassifications</font><font color="#000000"><font face="Tahoma"><font size="2"><span style="font-style: normal; "><span style="font-weight: normal; ">.</span></span></font></font></font></p>
<p style="margin-bottom: 0mm; widows: 2; orphans: 2; font-size: 10pt; "><br>
</p>
<p style="font-style: normal; font-weight: normal; widows: 2; orphans: 2; ">
<font color="#000000"><font face="Tahoma"><font size="2">Martin,
existing linguistic resources in the semantic web are able to
identify not only synonymy but almost all relation between concepts,
also some topological relation like partonomy (a thing is a part of
another).</font></font></font></p><p style="font-style: normal; font-weight: normal; widows: 2; orphans: 2; "><font color="#000000"><font face="Tahoma"><font size="2"><br></font></font></font></p>
<p style="margin-bottom: 0mm; font-style: normal; font-weight: normal; widows: 2; orphans: 2; ">
<font color="#000000"><font face="Tahoma"><font size="2">Gianfra</font></font></font></p><font class="Apple-style-span" size="2"> </font><br><br><br><br><br><div id="SkyDrivePlaceholder" style="font-size: 10pt; "></div><font class="Apple-style-span" size="2">> Date: Mon, 10 Jan 2011 20:27:01 +0100</font><br><font class="Apple-style-span" size="2">> From: dieterdreist@gmail.com</font><br><font class="Apple-style-span" size="2">> To: m@rtijn.org</font><br><font class="Apple-style-span" size="2">> CC: talk@openstreetmap.org; eda@waniasset.com</font><br><font class="Apple-style-span" size="2">> Subject: Re: [OSM-talk] Semantics layer for tags</font><br><font class="Apple-style-span" size="2">> </font><br><font class="Apple-style-span" size="2">> 2011/1/10 Martijn van Exel <m@rtijn.org>:</font><br><font class="Apple-style-span" size="2">> > (forgot to copy to talk)</font><br><font class="Apple-style-span" size="2">> ></font><br><font class="Apple-style-span" size="2">> > On Mon, Jan 10, 2011 at 6:35 PM, Ed Avis <eda@waniasset.com> wrote:</font><br><font class="Apple-style-span" size="2">> >> Martijn van Exel <m <at> rtijn.org> writes:</font><br><font class="Apple-style-span" size="2">> > The latter. The user would be able to tag a feature with "chemist",</font><br><font class="Apple-style-span" size="2">> > "pharmacy", "farmacia or "apotheek" and that would result in the same</font><br><font class="Apple-style-span" size="2">> > coding in the OSM database (currently: shop=chemist).</font><br><font class="Apple-style-span" size="2">> </font><br><font class="Apple-style-span" size="2">> </font><br><font class="Apple-style-span" size="2">> amenity=pharmacy, dispensing=yes/no</font><br><font class="Apple-style-span" size="2">> </font><br><font class="Apple-style-span" size="2">> When consuming</font><br><font class="Apple-style-span" size="2">> > OSM data, the process could be reversed; based on the locale, a</font><br><font class="Apple-style-span" size="2">> > feature tagged "shop=chemist" could (would) be output as being one of</font><br><font class="Apple-style-span" size="2">> > these culturally determined Things. Note that a "chemist", a</font><br><font class="Apple-style-span" size="2">> > "pharmacy", a "farmacia" and an "apotheek" are names for something</font><br><font class="Apple-style-span" size="2">> > that is similar across cultures and languages, but not literally the</font><br><font class="Apple-style-span" size="2">> > same.</font><br><font class="Apple-style-span" size="2">> ></font><br><font class="Apple-style-span" size="2">> > The idea is to *avoid* having different classifications on the</font><br><font class="Apple-style-span" size="2">> > database level, even though one concept could be represented by two</font><br><font class="Apple-style-span" size="2">> > different names in one language (consider freeway / highway). Any</font><br><font class="Apple-style-span" size="2">> > ambiguity arising from that would have to be handled by additional</font><br><font class="Apple-style-span" size="2">> > attributes.</font><br><font class="Apple-style-span" size="2">> </font><br><font class="Apple-style-span" size="2">> </font><br><font class="Apple-style-span" size="2">> I fear that a system like that will soon become utterly complex, thus</font><br><font class="Apple-style-span" size="2">> disabling most of the mappers of taking part in the</font><br><font class="Apple-style-span" size="2">> "tag-development-process". It would shift the discussions away from</font><br><font class="Apple-style-span" size="2">> the ML and wiki to defining the semantic rule set. And still we would</font><br><font class="Apple-style-span" size="2">> have to have definitions in natural language to define what a feature</font><br><font class="Apple-style-span" size="2">> is about, so there is no guarantee that there won't be contradictions</font><br><font class="Apple-style-span" size="2">> or different tags with the same meaning.</font><br><font class="Apple-style-span" size="2">> </font><br><font class="Apple-style-span" size="2">> I agree that it is a good idea to develop such a ruleset (or extend an</font><br><font class="Apple-style-span" size="2">> existing one like linked geodata) to make the usage of our dataset</font><br><font class="Apple-style-span" size="2">> easier (for developers), but I agree with you: it is not a magic wand.</font><br><font class="Apple-style-span" size="2">> </font><br><font class="Apple-style-span" size="2">> cheers,</font><br><font class="Apple-style-span" size="2">> Martin</font><br><font class="Apple-style-span" size="2">> </font><br><font class="Apple-style-span" size="2">> _______________________________________________</font><br><font class="Apple-style-span" size="2">> talk mailing list</font><br><font class="Apple-style-span" size="2">> talk@openstreetmap.org</font><br><font class="Apple-style-span" size="2">> http://lists.openstreetmap.org/listinfo/talk</font><br></div></div></div></div>
</body>
</html>