Librarians all across the globe are being trained in RDA: a new way of cataloguing, and in order to understand RDA we must first understand FRBR. Functional Requirements for Bibliographic Records is a new (1996) way of thinking and talking about bibliographical records that derives from the way computer geeks (that's most of us) describe records in databases.
We tend to think of our library management system as a unique and shining child however a computerised catalogue is a relational database.much like any other type of company management system, and another way of thinking of it is that it contains bits of information (entities) with descriptive bits of information attached (attributes) The entities are joined to other entities bits of information by various relationships. Actually, rather like the difference between what they tell you at school and then what they tell you in University, when you have to forget everything you thought you knew, all computer information is just made up of bits in a string, but that is not the most helpful way to think for these purposes.
So, at the moment a library management system contains authority records, containing information about authors ( persons, corporate bodies, families, now called "group 2 entities") or subjects (persons, places, objects, concepts, etc. - now called "group 3 entities" and, yes, a person could be both an author or a subject so they come under both categories.) Each time we want to make a record for a new work we have to find an authority record and copy the information from that record into the 100 field or as appropriate, and the subject into the 600 field. If we type correctly, the system can recognise the words we have copied and use them as a hyperlink to take us to a collection of other records using the same words. But if we fail to type them correctly the record will not link through, hence if you are searching the library for an author's name, a results list may contain several different variations of the author's name, with a different group of books linked to each (i.e. not ideal.)
In a relational database containing six million bibliographical records in an idealised best practice you should not have to type out the same piece of information (the particular author or subject) six million times but should have only one record for such concepts, which all the bibliographic records linked to. Hence if you updated the one record (equivalent to an authority record) containing the author's name, then when you looked at any book by that author in a database you could see that the form of the name had changed for all of them at once in the twinkling of an eye.
It has been said that in an ideal world, there would only be one record in the world for each work (a work is a type of entity too: a group 1 entity), and everyone's library records could link to it on the web. How amazing that would be! (will it ever happen, I wonder.) Anyway, not much of this was mentioned in the FRBR training, and I digress a little.
In practice we spent a lot of time thinking not only about entities, relationships and attributes, but also about Works, Expressions, Manifestations and Items (all group 1 entities) and the relationships between them. We thought about Harry Potter, and how J.K. Rowling had the idea in her head (the Work) and then published it in many different formats (Expressions with different relationships such as equivalence relationships for those at the same time in different places) and then other people wrote films and books about Harry Potter which were new works entirely with a derivative or descriptive relationship to the original. The concepts range over a spectrum so it is often difficult to place a particular work, such as one containing a long critical introduction or an illustrated edition of the original. Relationships between different items in a series could be sequential prequel/sequel or sequential multivolume (isn't that non-sequential?) where books could be read in any order.
It was interesting to think how these concepts are dealt with presently: some of the relationship concepts related to what we think of as field titles or even field numbers : "is the publisher of" or "is the author of" is contained in coded form in our records and the library system display settings convey this part of the information. A Work might currently be expressed as a uniform title, and there are currently no authority records to tell you how to record the name of any particular publisher, so records vary wildly in their publisher fields and it is difficult to search the catalogue by publisher, because it is therefore difficult for the system to group records together. Moreover, the Internet can use hyperlinks to link records to other records or to search records: we need to make sure that our modern way of cataloguing is the most appropriate for those who are going to be searching for them online.
A truly relational database would reduce duplication of work and reduce inaccuracies and organise the data far more efficiently than our current records do. However we have inherited six million bibliographic records, and a current way of displaying and organising records, so we cannot re-invent the wheel but must work with the systems that we already have and try to ensure that they are adapted as best they can to the requirements of the modern world. RDA aims to achieve this, so it will be very interesting to see what it involves in practice.
The Cambridge sessions were led by Celine Carty and Vanessa Lacey (thanks and well done!) and the training materials we used were created by Esther Arens for CILIP CIG. Esther recommended FRBR: A Guide for the Perplexed by Robert Maxwell as the best reading material, and suggested we look at examples of FRBR in databases such as LibraryThing http://www.librarything.com and Explore the British Library http://explore.bl.uk
See: FRBR: A guide for the perplexed / Maxwell, Robert L.- Chicago : American Library Association, 2008
We tend to think of our library management system as a unique and shining child however a computerised catalogue is a relational database.much like any other type of company management system, and another way of thinking of it is that it contains bits of information (entities) with descriptive bits of information attached (attributes) The entities are joined to other entities bits of information by various relationships. Actually, rather like the difference between what they tell you at school and then what they tell you in University, when you have to forget everything you thought you knew, all computer information is just made up of bits in a string, but that is not the most helpful way to think for these purposes.
So, at the moment a library management system contains authority records, containing information about authors ( persons, corporate bodies, families, now called "group 2 entities") or subjects (persons, places, objects, concepts, etc. - now called "group 3 entities" and, yes, a person could be both an author or a subject so they come under both categories.) Each time we want to make a record for a new work we have to find an authority record and copy the information from that record into the 100 field or as appropriate, and the subject into the 600 field. If we type correctly, the system can recognise the words we have copied and use them as a hyperlink to take us to a collection of other records using the same words. But if we fail to type them correctly the record will not link through, hence if you are searching the library for an author's name, a results list may contain several different variations of the author's name, with a different group of books linked to each (i.e. not ideal.)
In a relational database containing six million bibliographical records in an idealised best practice you should not have to type out the same piece of information (the particular author or subject) six million times but should have only one record for such concepts, which all the bibliographic records linked to. Hence if you updated the one record (equivalent to an authority record) containing the author's name, then when you looked at any book by that author in a database you could see that the form of the name had changed for all of them at once in the twinkling of an eye.
It has been said that in an ideal world, there would only be one record in the world for each work (a work is a type of entity too: a group 1 entity), and everyone's library records could link to it on the web. How amazing that would be! (will it ever happen, I wonder.) Anyway, not much of this was mentioned in the FRBR training, and I digress a little.
In practice we spent a lot of time thinking not only about entities, relationships and attributes, but also about Works, Expressions, Manifestations and Items (all group 1 entities) and the relationships between them. We thought about Harry Potter, and how J.K. Rowling had the idea in her head (the Work) and then published it in many different formats (Expressions with different relationships such as equivalence relationships for those at the same time in different places) and then other people wrote films and books about Harry Potter which were new works entirely with a derivative or descriptive relationship to the original. The concepts range over a spectrum so it is often difficult to place a particular work, such as one containing a long critical introduction or an illustrated edition of the original. Relationships between different items in a series could be sequential prequel/sequel or sequential multivolume (isn't that non-sequential?) where books could be read in any order.
It was interesting to think how these concepts are dealt with presently: some of the relationship concepts related to what we think of as field titles or even field numbers : "is the publisher of" or "is the author of" is contained in coded form in our records and the library system display settings convey this part of the information. A Work might currently be expressed as a uniform title, and there are currently no authority records to tell you how to record the name of any particular publisher, so records vary wildly in their publisher fields and it is difficult to search the catalogue by publisher, because it is therefore difficult for the system to group records together. Moreover, the Internet can use hyperlinks to link records to other records or to search records: we need to make sure that our modern way of cataloguing is the most appropriate for those who are going to be searching for them online.
A truly relational database would reduce duplication of work and reduce inaccuracies and organise the data far more efficiently than our current records do. However we have inherited six million bibliographic records, and a current way of displaying and organising records, so we cannot re-invent the wheel but must work with the systems that we already have and try to ensure that they are adapted as best they can to the requirements of the modern world. RDA aims to achieve this, so it will be very interesting to see what it involves in practice.
The Cambridge sessions were led by Celine Carty and Vanessa Lacey (thanks and well done!) and the training materials we used were created by Esther Arens for CILIP CIG. Esther recommended FRBR: A Guide for the Perplexed by Robert Maxwell as the best reading material, and suggested we look at examples of FRBR in databases such as LibraryThing http://www.librarything.com and Explore the British Library http://explore.bl.uk
See: FRBR: A guide for the perplexed / Maxwell, Robert L.- Chicago : American Library Association, 2008
Comments