[Date Prev] | [Thread Prev] | [Thread Next] | [Date Next] -- [Date Index] | [Thread Index] | [Elist Home]
Subject: [xtm-wg] Re: TM Structure Template; and Request for Association Types
I have been working on the problems related to structural links versus semantic links where structural links can be considered as a higher level of abstraction of the semantic links. If you consider human to human communication the humans using them interpret abstract (structural) links as semantic links. In most cases this interpretation will be a "good" one i.e. it will correspond to the meaning the creator intended for it. When following a link chain (e.g. "is located in") at some places a meaning change will occur. You don't say that a continent "is located in" the Earth but that a continent "is located on" the Earth. Human users make that shift naturally. In fact they will often not even notice that there is a meaning shift. But what about software? There will be fewer instructions necessary on how to handle structured links then the ones needed for handling semantic links. So the structural links seem to be a good solution. But as our view on the world grows (semantic network) quite quickly we encounter problems with the associations/links between topics. A simple example might illustrate this. In classical animal and plant classification we use the Linnée classification (Order, Family, Genus, Species) but in modern classification we use the concept of clades (based upon genetic characteristics). Both tree structures would have a structural link "belongs to". But what happens if you mix them without making the semantic distinction? You will end up with and network of loops with a lot of "contradictory" information. The way I think we might solve this problem is to delegate the responsibility for "acting" at the level where it belongs. If you keep the information as pure as possible (using or abusing the semantic of an association) you can always tell the user (human or software) how to handle that information. On the other hand if you apply an abstraction level to the information itself you will never be able to get the detail at a later stage. In my opinion it is the usage of a link that defines its structural meaning. In some case even the structural meaning might be different as you change the context of usage. You can add usage "hints" to an association/link, but let the user decide which hint to apply. These hints might take to form of possible structural links. Some aspects of TheBrain as a possible interface for NS (and I think the almost the same applies to TM also) are described very shortly in a paper on my web site. Go to my notion and jump to the paper I wrote. The implications of the usage of structural links are mentioned. Friendly greetings Ronald Poell rapoell@notionsystem.com http://www.notionsystem.com --- In xtm-wg@egroups.com, Andrius Kulikauskas <ms@m...> wrote: > Two requests, which I write more about below: > ===> I would be very interested to know if anybody has done research > into the kinds of global structural constraints that are implicit within > association types (for example, that the associations of the given type > are expected to form a sequence, or an unordered tree, or a directed > network, etc.). > ===> Please let me know of any lists of association types that I might > study. > > > I am looking forward to participating in the XTM group. I am not sure > yet if I will be able to attend in Montreal, but would like to. > > Our laboratory, Minciu Sodas, www.ms.lt, is focusing on the development > of a standard that will allow individuals to import/export aggregates of > notes from software tools for organizing them, such as TheBrain, > www.thebrain.com, MindManager, www.mindmanager.com, Multicentrix, > www.multicentric.com, Thoughtstream, http://thoughtstream.org, Lucid, > www.memes.net Our laboratory's emphasis is on the needs of the users of > such tools, so that they can work with their accumulated thoughts > regardless of environment, or at least be able to switch every few > months or so. > > We have been working through the Infrared Data Association, but the IrDA > leadership has decided that we fall outside of their focus and that we > end our working group. Our work so far is at: > http://www.ms.lt/importexport.html > > I have to learn more about TopicMaps, but right now I think that they > are adequate for expressing our standard. My only reservation so far is > that I think it would be simpler for us if associations were topics. > But I presume there are work-arounds for this. I will be focusing on > our particular needs, but I hope this may offer insights into how a > segment of the population, especially individuals, may use topic maps. > > One thing that I have learned is that users of "thought organizers" need > is not so much an import/export format, but rather, a modeling > language. The modeling language is a construct for understanding the > consequences of transforming (to greater or lesser satisfaction) > structured aggregates from one tool (user interface) to another tool > (user interface). The modeling language should be expressed in more > than one format, for example, XML, CORBA, Excel... If it is expressed > in only one format, then the format will be confused with the modeling > language, which conceptually can be disastrous, confusing both the user > and any assisting programmer. > > I feel more comfortable with Topic Maps upon realizing that it, too, is > really both a modeling language (a way of structuring how we think about > our information) and a format (HyTime, XTM, ...) I think it is crucial > to separate the two aspects, and I get the impression that this is being > done. It may be very convenient for us to describe our modeling > language as a subset of, or template within, the Topic Maps "modeling > language". > > I would like to organize the creation of a Topic Maps Template that > would serve our needs, but might also have additional value. My main > goal is to identify, and make available, the dozen or so "structural" > association types that occur in practice. By structural association > type (perhaps you can suggest a better term - syntactic?), I mean that > when I make a structural link, it may presume some global structural > constraints. For example, an X "IS A PART OF" Y association may presume > to be structurally part of a forest of such associations, where the > branches are unordered. An X "IS IN" Y association may presume this > very same structural environment. In our case, semantically, we are not > interested in the difference, at least not for our purposes, which is to > make sure that these structural constraints are made clear to the > author, and if desired, can be preserved. In other words, there may be > thousands of types of semantic associations, but if we are only > interested in the structural implications, there may be as few as one or > two dozen structural associations to worry about (different kinds of > sequences, hierarchies, networks, etc.). Users of tools for organizing > thoughts care about the structural associations, rather than the > semantic associations, because: > - Users use the structural environment the tool offers, that is, the > structural constraints it imposes, to shape their writing and thinking. > - Users do not have the time, energy or interest to mark up or otherwise > tend to their robustly multiplying thoughts. > - The ideas most worth writing down are typically the ones that we have > the least ability to express what we mean. And it is very disruptive to > have to work with artificial categories which can destroy that meaning. > - Users want to have the right to abuse tools, that is, use them for > strange purposes that their creators may not have semantically intended. > > I will be making a list of such structural link types. > ===> I would be very interested to know if anybody has done such > research already. Although my impression is that this has never been > done, I suppose it is considered too "simple". However, I think that > before delving into the complications of semantic relationships it would > be a helpful first step to identify the kinds of structural > relationships that arise, which seem quite limited. This is especially > helpful for the kind of import/export of aggregates of thoughts that we > would like to enable. > > If the research has not yet been done, then I would like to do it very > simply. > ===> Please let me know of any lists of association types that I might > study. I will go through these lists and try to figure out the > structural constraints that come up. I will be glad to share my results > in the public domain. > > I will write more about the Topic Maps Template I have in mind. > > Yours, > > Andrius Kulikauskas > Director > Minciu Sodas > http://www.ms.lt/importexport.html > ms@m... > +370 (2) 60-67-38 ------------------------------------------------------------------------ Replace complicated scripts using 14 new HTML tags that work in current browsers. Form the Web today - visit: http://click.egroups.com/1/5769/4/_/337252/_/963039985/ ------------------------------------------------------------------------ To Post a message, send it to: xtm-wg@eGroups.com To Unsubscribe, send a blank message to: xtm-wg-unsubscribe@eGroups.com
[Date Prev] | [Thread Prev] | [Thread Next] | [Date Next] -- [Date Index] | [Thread Index] | [Elist Home]
Powered by eList eXpress LLC