February 16, 2010 at 22:13Structuring Aspects 2
Readers of Structuring Aspects 1 have a misunderstanding of what it was written for. I suspect that with the second series will come out similar. Therefore, I will explain that in the process of writing a coherent text explaining some presumptive start-up, I had to paint certain things in relatively detail, so that they occupy pages instead of a paragraph or two. But since Internet readers in general and Habr in particular do not like long texts, this also has to be taken into account. As a result, I chose from two evils and designed the overgrown places with two separate notes, which summarized my attempts to collect, comprehend and comment on or add to the existing ones regarding the topic of structuring.
Regarding the structuring of knowledge, the approach to Aristotle with hierarchical classifications,taxonomies are almost the only one. In any case, completely dominant. For example, folders are organized in Windows. This is convenient, especially coupled with the visualization of the hierarchy. The downside is that a hierarchy can be built only on any one criterion or feature, while its elements usually have a number of features by which they can be built into different hierarchies or combined into different sets. For example, the same music files can be systematized by genre, artist, creation period, and more. But if you want to have hierarchies according to all these criteria, you will have to stupidly copy files, taking up space in memory.
About an alternative method - facet classificationprobably not everyone heard. Not to mention the fact that it is far from being as intuitive as hierarchical classifications. However, a big plus of this method is undoubted, since the problem noted above is solved - the same objects can belong to different hierarchies or sets at the same time.
Judging by the Wikipedia article , this is a complete list of conceptual approaches. It would seem that these things should relate to epistemology / epistemology (the science of knowledge) or cognitologybut it is not spelled out clearly. In any case, I could not at once find (sub) a section of knowledge devoted in general to the problems of classifications and structuring of knowledge. As far as I understand, epistemology focuses on several other things, and cognitology does not focus on the problems of structuring in terms of applicability purely for people, human perception. People are studied in terms of constructing models of memory and thinking, but not with the aim of deriving general principles of structuring, on the basis of which people could most adequately and easily perceive knowledge or operate with it, but with the goal of using these models in artificial intelligence. In addition, the question can be raised not only about ease of perception. For example, the above-mentioned option of alternatives without duplication of content may be important not only in terms of saving memory, but also in terms of social interactions in a common wiki space. The lack of a specialized section on structuring is rather strange, given its universal role in society (as I wrote inthe first article ), as well as the practical value of such knowledge for Internet services.
At the junction of human and machine, in the aforementioned field of artificial intelligence there is a rather developed section of knowledge representation (see also knowledge engineering ), which is largely related to the problems of structuring. For example, in its framework there is the concept of semantic networks . The latter, as well as classification hierarchies, are representable in the form of a directed graph. Those. these things could be studied on a general basis, again, if this subject had been formulated as an independent section of knowledge.
Apart from all this, such a method of structuring information as supplying content objects with tag tags has become widespread on the Internet. In this case, the object actually belongs to as many sets at the same time as many different labels are assigned to it. A good way is until the structured space is too large and / or too heterogeneous. For example, it's hard to imagine a reasonable tag cloud for Wikipedia. However, on this occasion, even books are being written - Tagging: Peaple Powered Metadata for the Social Web (which I have not read. See also the folksonomy link ). Judging by the name, the author understands tags as metadata.
The latter slightly differs from what is most often called metadata, for example, in the concept of the Semantic Web. There it is data about data, brief standardized descriptions of web resources, moreover, standardized for the convenience of operating machines, not people. Although the topic of a generalized understanding of metadata is notedthat if we move away from strict standardization and be closer to human perception, the difference between data and metadata will become conditional and relative - the same content, depending on the context, can serve both data and metadata. It is not noted that the role of metadata can be played by relationships between objects. For example, in hierarchical (tree-like) structures, relations between objects go from general to particular by default (or vice versa, this is not important). It is important that such a relationship contains certain information about the objects being linked. If we take semantic structures, then the general totality of various relations-relations for a given object can describe it quite comprehensively. This is one of the main ideas of the startup that I plan to write about.
Regarding the structuring of knowledge, the approach to Aristotle with hierarchical classifications,taxonomies are almost the only one. In any case, completely dominant. For example, folders are organized in Windows. This is convenient, especially coupled with the visualization of the hierarchy. The downside is that a hierarchy can be built only on any one criterion or feature, while its elements usually have a number of features by which they can be built into different hierarchies or combined into different sets. For example, the same music files can be systematized by genre, artist, creation period, and more. But if you want to have hierarchies according to all these criteria, you will have to stupidly copy files, taking up space in memory.
About an alternative method - facet classificationprobably not everyone heard. Not to mention the fact that it is far from being as intuitive as hierarchical classifications. However, a big plus of this method is undoubted, since the problem noted above is solved - the same objects can belong to different hierarchies or sets at the same time.
Judging by the Wikipedia article , this is a complete list of conceptual approaches. It would seem that these things should relate to epistemology / epistemology (the science of knowledge) or cognitologybut it is not spelled out clearly. In any case, I could not at once find (sub) a section of knowledge devoted in general to the problems of classifications and structuring of knowledge. As far as I understand, epistemology focuses on several other things, and cognitology does not focus on the problems of structuring in terms of applicability purely for people, human perception. People are studied in terms of constructing models of memory and thinking, but not with the aim of deriving general principles of structuring, on the basis of which people could most adequately and easily perceive knowledge or operate with it, but with the goal of using these models in artificial intelligence. In addition, the question can be raised not only about ease of perception. For example, the above-mentioned option of alternatives without duplication of content may be important not only in terms of saving memory, but also in terms of social interactions in a common wiki space. The lack of a specialized section on structuring is rather strange, given its universal role in society (as I wrote inthe first article ), as well as the practical value of such knowledge for Internet services.
At the junction of human and machine, in the aforementioned field of artificial intelligence there is a rather developed section of knowledge representation (see also knowledge engineering ), which is largely related to the problems of structuring. For example, in its framework there is the concept of semantic networks . The latter, as well as classification hierarchies, are representable in the form of a directed graph. Those. these things could be studied on a general basis, again, if this subject had been formulated as an independent section of knowledge.
Apart from all this, such a method of structuring information as supplying content objects with tag tags has become widespread on the Internet. In this case, the object actually belongs to as many sets at the same time as many different labels are assigned to it. A good way is until the structured space is too large and / or too heterogeneous. For example, it's hard to imagine a reasonable tag cloud for Wikipedia. However, on this occasion, even books are being written - Tagging: Peaple Powered Metadata for the Social Web (which I have not read. See also the folksonomy link ). Judging by the name, the author understands tags as metadata.
The latter slightly differs from what is most often called metadata, for example, in the concept of the Semantic Web. There it is data about data, brief standardized descriptions of web resources, moreover, standardized for the convenience of operating machines, not people. Although the topic of a generalized understanding of metadata is notedthat if we move away from strict standardization and be closer to human perception, the difference between data and metadata will become conditional and relative - the same content, depending on the context, can serve both data and metadata. It is not noted that the role of metadata can be played by relationships between objects. For example, in hierarchical (tree-like) structures, relations between objects go from general to particular by default (or vice versa, this is not important). It is important that such a relationship contains certain information about the objects being linked. If we take semantic structures, then the general totality of various relations-relations for a given object can describe it quite comprehensively. This is one of the main ideas of the startup that I plan to write about.