tag:blogger.com,1999:blog-9094731596169096992.post2766736034066309445..comments2024-03-19T07:38:13.661-04:00Comments on The Accidental Taxonomist: Avoiding Mistakes in Taxonomy Hierarchical RelationshipsHeather Heddenhttp://www.blogger.com/profile/16424216206886861070noreply@blogger.comBlogger3125tag:blogger.com,1999:blog-9094731596169096992.post-72345618423653338192017-04-09T10:50:43.880-04:002017-04-09T10:50:43.880-04:00Excellent article, Heather! This is absolutely one...Excellent article, Heather! This is absolutely one of the first errors students make when learning to create hierarchical taxonomies, and likely because we are so used to organizing various aspects of information in TOCs, vs. creating the specific types of relationships as defined in the standards. <br /><br />Thank you for clarifying not only what the issues are, but also how to rectify them. There's always more to learn!Deborah Fanslowhttps://www.blogger.com/profile/11265617577930324147noreply@blogger.comtag:blogger.com,1999:blog-9094731596169096992.post-53780755416785514892017-02-20T18:50:09.641-05:002017-02-20T18:50:09.641-05:00I haven't noticed this problem as much, but yo...I haven't noticed this problem as much, but you bring up a good point, Kurt. While the standards specify that a hierarchical relationship may be either of the generic, instance, or whole-part types, the standards do fully explain when it is acceptable and when it is not to mix the hierarchical relationship types for the same term. Doing so, such as for geographical place, is thus not a violation of current standards, but rather goes against best practices. It's not recommended.<br /><br />In re-reading the ANSI/NISO Z39.19 standard now, I see it gives the example of Bones NT Skull (generic/class type of hierarchical relationship), and Head NT Skull (whole-part type of hierarchical relationship), and it says "In some cases, polyhierarchical links can be based on logically different relationships." In my opinion, what is missing from the standards is something saying: The polyhierarchical link based on logically different relationships is OK with the combination of the generic and the whole-part types of hierarchical relationships, but polyhierarchical links should not be based on logically different relationships with the combination of the instance and the whole-part types of hierarchical relationships. This would be the example, of Canada NT British Columbia and States/Provinces NT British Columbia, within the same taxonomy. Heather Heddenhttps://www.blogger.com/profile/16424216206886861070noreply@blogger.comtag:blogger.com,1999:blog-9094731596169096992.post-43355652274298511752017-02-20T17:12:20.882-05:002017-02-20T17:12:20.882-05:00Nice work. I think one of the biggest problems tha...Nice work. I think one of the biggest problems that I've seen with many taxonomies is failure to distinguish between a class and an instance. The classic example of this is something like:<br /><br />Country<br />+ USA<br /> + Alabama<br /> + Alaska<br />+ Canada<br /> + Alberta<br /> + British Columbia<br /><br />These are hierarchies, but they are not topical hierarchies. Canada is an instance of Country, British Columbia is an instance of StateProvince. Yet British Columbia has a partOf relationship, to Canada, which is distinct from the Broader Term relationship of StateProvince to Country. Anonymoushttps://www.blogger.com/profile/11746604103523406806noreply@blogger.com