4. Entities

frbr-lrmEntities

p. 12-23

Advertisements

18 thoughts on “4. Entities

  1. Pingback: World-wide review of the FRBR-Library Reference Model, a consolidation of the FRBR, FRAD and FRSAD conceptual models | FRBR Open Comments

  2. I agree about Group 3. Subject is a relationship, and it seems to best not to limit what can and cannot be a subject. Basically, if you can name it or give it an identifier, it can be the subject of a cultural work.

    Like

  3. It seems that anything which is fictitious can only fall under Res in FRBR-LRM. This is explicitly stated for persons (p. 19) and places (p. 22), but I assume that this is also true for all entities apart from Res (there can also be fictitious Works, Time-spans a.s.o.).

    I find this very problematic for a number of reasons:

    1) It makes sense to use the same attributes for a fictitious person as for a real person, for a fictitious place as for a real place a.s.o. The same goes for relationships: Why should I not be allowed to use e.g. LRM-R15 “Res – has association with/is associated with – Place” to express that Gandalf has an association with Middle-Earth? But LRM makes us fall back on the most general entity (Res) and the most general relationship (“Res – Res”) for all fictitious entities. This is highly unsatisfactory. It’s also unnecessary, because we could easily add an attribute “Fictitiousness” to Res (perhaps with four values: real, fictitious, uncertain and unknown). In the German Integrated Authority File, we use the same fields for fictitious entities as for their real counterparts, but they are marked with special codes.

    2) According to LRM it depends on our knowledge of something whether this something is classed as a Res or as one of the subordinate entities. As long as we believe that somebody exists, he or she is a Person. But when we find out it’s a made-up person, then it changes its status and becomes a Res. I find this very odd.

    3) As a colleague from Germany pointed out: How are we supposed to decide which entities are real and which aren’t? There are many difficult cases, and sometimes it’s up to personal beliefs. Is Gilgamesh a Person or a Res? What about Jesus? And how are we supposed to treat novels about real persons, which treat their subject-matter in a highly fictitious way? Is such a novel about the real Person or about a fictitious version of this person, i.e. a Res?

    Like

  4. Entity “res”

    The LRM definition for “res” is “any entity in the universe of discourse” (p. 13). Now, what does “universe of discourse” mean? At first I thought it might be a stilted way of saying “everything we can talk about”. A colleague put it slightly differently: “everything which is mentioned anywhere”.

    The scope note says: “Everything considered relevant to the bibliographic universe, the universe of discourse in this case, is included.” So it seems that the “universe of discourse”, for FRBR-LRM, is identical with the “bibliographic universe”.

    After checking Wikipedia, I now know that “universe of discourse” is a set phrase from the formal sciences: “The term ‘universe of discourse’ generally refers to the collection of objects being discussed in a specific discourse.” https://en.wikipedia.org/wiki/Domain_of_discourse

    I conclude that “universe of discourse” by itself is not really meaningful. So I think it would be much better to change the definition to something like “any entity which is relevant to the bibliographic universe”.

    There was also some discussion here in Germany about the Latin name “res” for the entity. One colleague pointed out that it isn’t really necessary to use a specific name at all. He argued that we already have a perfectly good name for this, namely “entity”. This would also be in line with the approach of the CIDOC CRM, where the top level entity is simply called “E1 CRM Entity”:
    http://new.cidoc-crm.org/Entity/E1-CRM-Entity/Version-6.2

    (I note that CRM also uses “universe of discourse”, but non on its own like LRM does. They specify it like this: “This class comprises all things in the universe of discourse of the CIDOC Conceptual Reference Model.”)

    Like

    • As Karen pointed out: we already have “Thing”. Using “Entity” as the top level thing would only add to the confusion because we are dealing here with an Entity Relationship Model, where “Entity” means ‘Class”/”Type”. If we were to name the top level Entity “Entity”, well, you see what is going on there.

      Like

      • Regarding Res, is its inclusion necessary? My short argument is that Res is just “bibliographic thing” & a universal superclass is already assumed by the creation of the document in its intellectual context , but that is worth unpacking:

        Even highly structured technical documents have underlying semantic & ontological assumptions assumptions (and a note, I am using the philosophical and linguistic definitions of semantic, ontology, and metaphysics going forward). We can make the following assumptions before even reading a word of this document:

        1. Despite the under-definition and general uselessness of the concept “bibliographic universe,” we know that the document gives us some kind of thing for using to record bibliographic things (even if the document is just a conceptual model) — “requirements” for “bibliographic records.” As a result, the equivalent to owl:Thing in this model is already assumed. We are in fact already operating at a human ontological subclass below “thing,” at the “bibliographic thing.”

        2. Since the “bibliographic thing” (i.e. Res) is the domain of the document, a structured technical document should strictly define what the universe of discourse is. Leaving aside that the document doesn’t actually do this (again, what’s the bibliographic universe?), we already know we’re talking about the bibliographic universe. It’s assumed before we even get to the entity-relationship model. We shouldn’t be using the model if we don’t assume Res is the domain of the document.

        As a result, we can and should just start conceptually modeling at WEMIA-NPTS (now that’s a mouthful!). Good documentation &/or concept-forming will lay out the assumptions being made in the model–in this doc, should be in ch. 2-3.

        Like

      • I assume we will need (at times) to identify the bibliographic thing that appears to be now represented by Res. If nothing else, there will be needs to identify whole descriptions and portions of descriptions for the purposes of sharing cataloging data.

        Like

      • I think that’s a matter for implementation instances, though, not for this conceptual model. Using the example of MARC, it lumps WEMI together in a whole description but makes it very hard because of its structure to separate the information.

        Future implementations, should they be a faithful following of this model, will need some way to accurately delineate which entities have which relationships. In this manner, Res is simply “an aggreate of all descriptions available or possible” depending on what you’re looking to identify. It doesn’t seem to fit right. The descriptions will have a container and seems the container is what you mean to identify?

        Like

  5. This tweet showed up (uninvited) as a sidebar here:

    It is John MacManus asking: “Is every tweet a “work”? Cataloguing this at the moment: My Little Book of Tweets by Gerry Adams #frbr ”

    The entities WEMI are considered required in the LRM, but it does make sense to me to question whether everything has “workness.” I suspect that every tweet is a work, but what about a satellite image? Census data? Are there materials or types of materials for whom the separation into WEMI do not make sense? If so, what is the best course of action for those?

    Like

  6. “Constraints may operate between different entities. In general, other than those entities related by IsA hierarchies, the entities declared in the model are disjoint. Disjoint entities can have no instance that is simultaneously an instance of more than one of these entities. This means, for example, that an entity cannot be both a person and a collective agent.” p. 11

    The FRBR-LRM shows a lot of openness to varied developments of actual bibliographic data. This declaration of disjointness between entities is an exception. This is a rule that I perceive as having a large amount of hubris, because it declares that the entities as defined here are perfect and immutable. If some community finds that they cannot make use of separate entities for work and expression, and wish to combine those into a single entity, they are violating the model. More than just branding themselves as FRBR heretics, this could affect the ability of that community to share data with others for whom work and expression are treated separately.

    This issue, which seems to be inherited from FRBR, was addressed by Tom Baker, Sean Petiya and I in an article(*) on multi-entity bibliographic models. Treating the entities as disjoint makes it difficult — and in some technologies, impossible — to allow flexibility in the definition of attributes for the entities. Definitely in RDF, the association of attributes to entities loses its flexibility when the entities are considered disjoint. If work and expression are disjoint, one cannot, as is suggested in the LRM text, have one community determine that music key is an attribute of the work and another community make use of key as an attribute of expression.

    In RDF, the entities (as classes) do not determine the nature of the attributes, but it is the attributes (called properties) that define the class membership of the entity. So where we usually think of classes as being definers of their members, such as dog is a mammal because it is a subclass of “mammal”, in RDF one defines that a creature that is warm-blooded and has fur (for instance) is a mammal. So the “mammalness” is derived from the attributes. What that means is that if music key is defined in one bibliographic model as an indication of “workness” and another as an indication of “expression-ness” it becomes more difficult for those models to interact if you cannot mingle your work and expression. And it also makes it difficult to imagine mingling FRBR-based bibliographic data and data using other models, such as BIBFRAME.

    From my reading, no motivation is given for the decision that entities must be disjoint. Given the potential disadvantages of such a decision, it is best not to make this declaration in the model.

    (*) Baker, Thomas, Karen Coyle, and Sean Petiya. “Multi-entity models of resource description in the semantic web: A comparison of FRBR, RDA and BIBFRAME.” Library Hi Tech 32.4 (2014): 562-582.

    Like

  7. Like FRBR, the LRM does not create a general “bibliographic thing”. Res is the upper class for all things in the model, but that includes persons and places, which are not inherently bibliographic. It always seems to me that there is a missing level that would be “bibliographic entity” – the thing that is being described. So in my mind there would be:
    Res
    Agent
    –Person
    –Collective
    Place
    Time-Span
    Bibliographic “thing”
    –W
    –E
    –M
    –I

    I’m not saying that this is perfect, but I do think that there is a bibliographic “thing”, and that treating WEMI as separate entities related directly to Res doesn’t make sense because they are inseparable, and have a meaning as a unit.

    Like

    • Yes, the additional Bibliographic Thing would also solve some general modeling problems with super/sub classes/types, disjointness and inheritance. Currently, all entities are defined as subclasses of RES. The subclasses definition is either an “exclusive specialization” (implying disjointness, or something can only be of one of the subclasses) or an “overlapping specialization” (something can be of more than one subclass). However, it is explicitly stated here that “The entities Work, Expression, Manifestation, Item are disjoint”. But also (p. 11) “In general, other than those entities related by IsA hierarchies, the entities declared in the model are disjoint. Disjoint entities can have no instance that is simultaneously an instance of more than one of these entities.”. This is rather confusing.
      If a Bibliographic Thing entity is introduced, with subclasse Work, Expression, Manifestation, Item, as an exclusive specialization, that will solve not only the model infringements, but also the inheritance problems.
      There is no inheritance of attributes among entities linked by a relationship. But in the entities descriptions you find statements like: “an item exemplifying a manifestation normally reflects all the attributes that define the manifestation itself.” And in the attributes lists of the entities there are a lot of implicitly inherited ones. The same can be said for the relationships. In reality, Work, Expression, Manifestation and Item share a lot of properties.

      Like

      • No, and this is where I think FRBR causes us problems. WEMI are all separate but simultaneously dependent on each other, so the manifestation actually is a “manifestation of an expression of a work”. As defined in FRBR and FRBR-LRM, you cannot have a manifestation without an expression, nor an expression without a work. In LRM, it sounds like you also cannot have a manifestation without an item. I find this “daisy chain” of entities to be awkward because there is no way to refer to them as a unit even if you need to, and the IFLA committees have stated outright that there is to be no superclass that represents those four entities. So model-wise they are both entirely inter-dependent and at the same time independent as entities. This is where I have trouble understanding what is intended.

        Note that a manifestation has no creator, no topic, no genre, and you do not know if it is text or sound. That information is all associated with other entities.

        Like

  8. P. 18: “Item: A physical object carrying signs resulting from a production process and intended to convey intellectual or artistic content”. “An item is in many instances a single physical object, but in other cases an item may comprise multiple physical pieces or objects. An item may be a part of a larger physical object, for example, when a file is stored on a disc which also contains other files, the portion of the disc holding the file is the physical carrier or item.”
    If Items are always physical objects, then this limits the scope of the model drastically.
    But the most questionable statement is that “the portion of the disc holding the file is the physical carrier or item”. Looking at for instance ebooks, in my view a copy of an EPUB file would be considered an Item, similar to a physical printed copy of a physical book. The EPUB item obtained by the user is definitely not a piece of the disc in the library server.

    Like

  9. This is part 2 of my personal response to the FRBR Review Group, based on my comments on the RDA, FRBR, and PCC lists.

    2. Agent as an Entity

    Although it is true that any entity-relationship model is allowed to define its own entities, relationships, and attributes, it seems a mistake to me to define “agent” as one of the entities. Agency, to my thinking, denotes a relationship, not an entity. A person, family, or corporate body may have a relationship of agency with a work, expression, manifestation, or item; but a person, family, or corporate body may have a different kind of relationship with entities that are not relationships of agency. A person many have a relationship of “colleague” to another person (RDA K.2.1). This is not an agency relationship. A corporate body may have a “predecessor” relationship to another corporate body (RDA K.4.3). This is not an agency relationship. A family may have a “descendant family” relationship to another family (RDA K.3.2). This is not an agency relationship. In my opinion, turning “agent” into an entity when it’s (in my opinion) more clearly a relationship stretches the limits of ER principles. It seems to me to be a flaw in the model.

    I realize the definition of agent is given as “An entity capable of exercising responsibility relationships relating to works, expressions, manifestations or items” and supposedly all “real” persons are capable of exercising responsibility, so presumably there would be no “real” persons excluded from the entity “agent”, even those who never exercise that responsibility. First of all, this proposition is questionable. A one-day old baby is a person. Is this person capable of exercising responsibility? Arguably, no. A person in a vegetative state who has been so since birth is a person. Is this person capable of exercising responsibility? Arguably, no. So by the definition of “agent” (which applies to all the subentities beneath it) some persons are, strangely, excluded from the entity “person”. Second, I’m not sure of the utility of introducing the idea of “capable of” into an entity definition. There are lots of things entities are capable of doing but which individual instances of the entity don’t in fact do.

    I do see the utility of having an over-arching entity that comprises persons, families, and corporate bodies because it simplifies the model, though simplification can be taken to unnecessary extremes. I do not see, however, the appropriateness of calling that over-arching entity “agent” and squeezing all persons, families, and corporate bodies into that role. I don’t have a suggestion for what to name the over-arching entity; on the other hand, maybe it’s not necessary to simplify the model to this extent. I do recommend that the two narrower entities be called “individual” and “collective” or “group” (rather than “collective agent”). Or just go back to having the three FRBR entities “person” “family” “corporate body.”

    Like

  10. This is part 4 of my personal response to the FRBR Review Group, partially based on comments to the RDA, FRBR, and PCC lists. The comment is specifically about 4.1.3 and LRM-E3.

    4. Definition of Expression

    I was more than a bit startled by the new definition of “expression” in FRBR-LRM: “A distinct constellation of signs conveying intellectual or artistic content.” (LRM-E3, p. 15). What does this mean? I have a hard time imagining this phrase meaning anything to someone who asked the question “what is an expression?” or even to people who already know what an expression is. The original FRBR definition “the intellectual or artistic realization of a work in the form of alpha-numeric, musical, or choreographic notation, sound, image, object, movement, etc., or any combination of such forms” is not flawless and does require a bit of thinking, but at least it does make sense (the word “realization” is key, when a work “becomes real” it has an expression). I’m not sure “constellation of signs” conveys the idea in any way at all. I suggest that the new definition isn’t really an improvement. In fact it makes the concept, which is important to the model but already somewhat tricky though not impossible for people to understand, much more obscure. This is a shame since “expression” is a crucial entity in the model.

    Like

Leave a Reply

Fill in your details below or click an icon to log in:

WordPress.com Logo

You are commenting using your WordPress.com account. Log Out / Change )

Twitter picture

You are commenting using your Twitter account. Log Out / Change )

Facebook photo

You are commenting using your Facebook account. Log Out / Change )

Google+ photo

You are commenting using your Google+ account. Log Out / Change )

Connecting to %s