Loading...
 

Mapping to and from Dewey

Revisions of EDUG’s recommendations for mapping to DDC

Posted by Grete Seland on 2016-04-08 02:00
Dear EDUG colleagues:

On behalf of the project Mapping to Norwegian WebDewey, I would like to suggest a few revisions to EDUG’s recommendations for mapping to DDC, cf. the present version published 20151009 at http://edug.pansoft.de/tiki-download_file.php?fileId=113. In our team, we have experienced that the mappers relate to the sub-sections of chapter 5 in the recommendations as a checklist and the most actively used part of the recommendations. We therefore think it would be useful to integrate the general principles governing the selection of exact equivalence and inexact equivalence in this chapter. Furthermore, we have added a couple of clarifications that are returning matters in the training of mapping staff. Suggested revisions are *between square brachets and stars*:


First, we suggest a new paragraph at the beginning of part 5.1 (consequently pushing today’s 5.1.a and 5.1.b into 5.1.b and 5.1.c, respectively):

5.1. Exact equivalence (=EQ)


a. MATCHING CONCEPT
If the DDC class as a whole is considered identical in scope to the concept from the source vocabulary, this should be mapped as exact equivalence (=EQ), in SKOS skos:exactMatch.

Examples:
The concept “Religious organizations” =EQ to DDC 206.5 with caption Organizations and organization
The concept “Crises” =EQ to DDC 904 with caption Collected accounts of events


In today’s 5.1.a, which would become 5.1.b if the paragraph above is to be added, we would like to add a couple of sentences (bold) at the end of the present first passage (also cited below):

b. SINGLE CONCEPT
If the DDC caption represents a single concept matching the concept from the source vocabulary, this should be mapped as exact equivalence (=EQ), in SKOS skos:exactMatch. For single concepts listed both in the schedules and in an auxiliary table, both instances should be mapped as exact equivalence. * Remember to check the hierarchical context of the DDC class. Beware that concurrent source term and caption is not sufficient in itself – there should be matching concepts.*


In today’s 5.1.b, which would become 5.1.c if the new paragraph above is to be added, we would like to add a couple of sentences (bold) at the end of the present first passage (also cited below):

c. INTERDISCIPLINARY NUMBERS IN ONE-TO-MANY MAPPINGS
If a decision has been made to map to all disciplinary contexts (cf. subsection 2.2.c), one might decide to ascribe a special status for interdisciplinary numbers in connection with one-to-many mappings (cf. subsection 4.6), by mapping them as exact equivalence. * Beware that there might be several mappings with exact equivalence (=EQ) in a set of one-to-many mappings, justified by different paragraphs. Thus, the application of =EQ for an interdisciplinary number does not rule out the application of 5.1.a and/or 5.1.b for other mappings of the same source concept. *


We suggest a new paragraph at the beginning of part 5.2 (consequently pushing today’s 5.2.a through 5.2.d into 5.2.b through 5.2.e, respectively):

5.2 Inexact equivalence (~EQ)


a. CLOSELY MATCHING CONCEPT
If the DDC class as a whole is considered close in scope to, but not exactly the same as, the concept from the source vocabulary, this should be mapped as inexact equivalence (~EQ), in SKOS skos:closeMatch. Concepts may be closely matching in several respects, e.g., they may be equivalent in some contexts but not others, or the concepts may have overlapping scopes (cf. ISO 25964-2, 11.3 p. 27 for discussion).

Examples:
The concept “Visual impairment” ~EQ to DDC 617.7 with caption Ophthalmology
The concept “Symmetry groups” ~EQ to DDC 512.2 Groups and group theory


In today’s 5.3.a, we would like to add a couple of sentences (bold) at the end of the present first passage (also cited below):

5.3 Broader Mapping (BM)
a. BROADER CONCEPT
If the DDC caption represents a broader concept relative to the concept from the source vocabulary, this should be mapped as broader mapping (BM), in SKOS skos:broadMatch. * Each mapping team should decide whether the broader relationship should be interpreted as generic, partitive or instantial only, or whether the DDC practice for hierarchical structure should be used (cf. ISO 25964-2, 17.2.3 p. 54 on the different principles governing hierarchies in thesauri versus classifications). *


Please let me now if you have any remarks on these suggestions, by making a reply to this post. I will comment on this issue during the session called “Mappings – what has happened since last year’s workshop” in Göttingen. I am looking forward to seeing you all!

Regards from Grete Seland at the Norwegian mapping team, University of Oslo Library

response
Posted by Rebecca Green on 2016-05-06 22:39
New section 5.1.a: The text looks fine, but the examples don’t seem to match the text. The “and organization” part of 206.5 Organizations and organization keeps the concept “Religious organizations” from being identical in scope with the DDC class. And 904 Collected accounts of events covers both more and less than Crises (on the one hand, fortunately, not all events are crises; on the other hand, 904 is limited to Collected accounts of events and is further limited by the topics drawn off in the class-elsewhere notes).

Addition to (renumbered) 5.1.c: According to section 4.6, “Beware of the general rule for the selection of mapping relationship types for interdisciplinary or comprehensive concepts designated as such according to instructions in class-here or including notes: Interdisciplinary and comprehensive concepts should be treated as any topic listed in these notes; they should be mapped as inexact equivalence when listed in class-here notes (cf. subsection 5.2.c), or as broader mapping when listed in including notes (cf. subsection 5.3.b), respectively.” (Renumbered) section 5.1.c contradicts this statement by according “special status” to interdisciplinary numbers. The problem with (renumbered) section 5.1.c generally and with the examples in (new) section 5.1.a is that they would rob exact equivalence of its essence. (Note: So far this is mostly a comment about the recommendations as they currently stand, and not about the suggested addition.)

Exact equivalence in mapping should parallel mathematical identity. The only circumstances under which exact equivalence should be other than a 1:1 mapping is if either of the terminologies in the mapping includes synonyms (as the DDC Relative Index does). Having both Exact equivalence and Inexact equivalence means that Exact equivalence should be used for only the very closest matches. Can 5.1.a and (new) 5.1.b justify Exact equivalence between a source concept and two or more different classes? Highly unlikely. The utility of (new) 5.1.b is to take care of situations where captions do not represent single concepts.

New section 5.2.a: The text looks fine (except that we don’t have, and others may not have, access to ISO 25964-2; does it include anything that needs to become part of the EDUG recommendation?). However, again, the examples don’t seem to match the text. Why is the mapping of Visual impairment to 617.7 Ophthalmology not an associative mapping? Why is the mapping of Symmetry groups to 512.2 Groups and group theory not a broader mapping?

Addition to section 5.3: It would be helpful if the intent of the reference to ISO 25964-2, 17.2.3 were spelled out.

response
Posted by Grete Seland on 2016-06-10 12:50
Thank you so much, Rebecca, for your response and further suggestions. I agree with you that my examples did not quite match the text. I have replaced all the examples in the new sections 5.1.a and 5.2.a. Concerning your comments on the use of exact equivalence: I assume that you by “mathematical identity” mean that exact equivalence mappings should be reversible. When the recommendations were made last year, the option to attribute a special status for interdisciplinary numbers in multiple independent mappings (assigning them =EQ) was agreed upon. I have added a note in section 4.6 (second half of the citation to come) to make clear that this choice is an optional exception to the general rule: “Interdisciplinary and comprehensive concepts should be treated the same as any other topic; for example, they should be mapped as inexact equivalence when listed in class-here notes (cf. subsection 5.2.d), or as broader mapping when listed in including notes (cf. subsection 5.3.c). Note: This general rule can be set aside if a decision is made to map interdisciplinary numbers in multiple independent mappings as exact equivalence, applying the optional rule in subsection 5.1.c.” You point out that users of the recommendations might not have access to the ISO 25964-2. To simplify matters, I have added citations from the ISO as footnotes when the text gives direct reference to ISO paragraphs. For the rest, I have made revisions according to the presentation I gave in Göttingen this April. Today I have sent the revised recommendations to the EDUG board, and assume that they will soon be published here at the EDUG website.
Regards from Grete Seland, University of Oslo Library


Two more suggested revisions to the EDUG recommendations
Posted by Grete Seland on 2016-04-21 12:44
Dear EDUG colleagues:
Referring to my post on April 8th concerning suggested revisions to the EDUG recommendations, I would like to add two more suggestions:

First: Replace the seven occurrences of the expression “one-to-many mappings” with “multiple independent mappings” throughout the recommendations text.

Second: Make an addition at the end of paragraph 2.2.c which now reads:
COMPOUND MAPPINGS Decide whether or not to use compound mappings.
I would like to add the following formulation in parenthesis: “(cf. one-to-many mappings in which a single concept in one vocabulary is mapped to a combination of two or more concepts in another vocabulary)”.

The EDUG recommendations unfortunately does not differentiate between one-to-many mappings versus multiple independent mappings. My understanding is now: One-to-many mappings concern the concatenation of several concepts, which then work as one combined concept in the source or target end of a mapping. Multiple mappings concern several mappings between a Source:Target concept pair, but that (in case of mapping to DDC) the target concept is represented several times in different disciplinary contexts in the vocabulary.

Concerning the first suggestion: Thanks to my project colleague Dan Michael Heggø, I have realized that the way we use the expression “one-to-many mappings” in the recommendations (7 occurrences) does not match the definition in the ISO 25964-2, which states in paragraph 3.55 that a one-to-many mapping is a “mapping (3.41) in which a single concept (3.17) in one vocabulary is mapped (3.39) to a combination of two or more concepts (3.17) in another vocabulary”. (This I should preferably have realized last year when I participated in the editorial team for the EDUG recommendations!) In the recommendations, we have used “one-to-many mappings” to refer to the typical situation we face when mapping to DDC, that we establish several independent mappings from one source concept to several DDC classes. Therefore, we have overlooked the note below the ISO definition of one-to-many mappings, which states: “This situation is different from one in which a concept has two or more independent mappings to concepts in the other vocabulary”. I can find no standard expression or definition for this phenomenon, but I suggest that we from now on use “multiple independent mappings” to refer to several independent mappings from one source concept to several DDC classes. We have one occurrence of the expression “multiple mappings” with this meaning in the recommendations text on top of p. 11 (section 5) – this I suggest adjusted to “multiple independent mappings”.

Concerning the second suggestion:
I suggest the mentioning and definition of one-to-many mappings to be listed in the EDUG recommendations paragraph concerning compound mappings, as it will be useful for the readers of the recommendations to become aware of the distinction between one-to-many mappings and multiple independent mappings.

Regards from Grete Seland at the Norwegian mapping team, University of Oslo Library

response
Posted by Rebecca Green on 2016-05-06 22:28
The two additional suggested revisions make sense. We offer two further suggestions:

It’s not clear how “cf.” is meant to be interpreted in “(cf. one-to-many mappings in which a single concept in one vocabulary is mapped to a combination of two or more concepts in another vocabulary)”; is “i.e.” what is meant instead?
It would be useful somewhere to contrast “compound mappings” and “multiple independent mappings” in the recommendations themselves; this could be a good place to do so. Both could be characterized as “one-to-many mappings in which a single concept in one vocabulary is mapped . . . to . . . two or more concepts in another vocabulary.” In the case of compound mappings, that becomes “a combination of two or more concepts”; in the case of multiple independent mappings that becomes something like “independently to each of two or more concepts.”