Course:LIBR509/Notes/2022WT2

Important Links

Canvas Site

Class Notes by Week

Week 2: Classification Principles

- Classification Principles: systematic categorization of objects.
- Categorization: The process of dividing the environment in to equivalence groups. Members of a category are "similar" in that they share some perceived quality or attribute
- Categories are fundamental to the way we learn and the way we learn experience.
- Categories are the building blocks or cognition. The apprehension of similarity brings knowledge. By grouping entities according to observable similarities, we form concepts about the environment. This allows us to make generalizations about past experiences.
- Ways to define a category:
  - Intension- a definition or description of a concept
  - Extension- the member or examples of an object
- Classical theory of categories: Defining features can be specified, in order to be a member of the category, an entity must have all the defining features (all entities that have the defining features are in, this group is called the extension, all other entities are out), there's no such thing as one member being a better fit for the category than another. Hierarchical inheritance.
  - Not always possible to list all properties and properties don't strictly determine membership (i.e. the mother category)
  - Some items fit better in a category than others (i.e. pb&j is a better fit for sandwiches than a Big Mac)
  - The classic theory doesn't allow for shifting and complex nature of social categories (i.e. Gender)
  - Can be politically charged or biased (i.e. Apartheid)
- Prototype theory (Rosch) and radial categories (Lakoff) are alternatives to classical theory
- Faceted classification: developed for library classification to develop shelf order but has been adapted for various online retailers. They help limit search results to the facets you specify (i.e. price, colour, etc.). The categories are exclusive and don't have any overlap.
  - An individual value is an isolate, a group of isolates is a facet (red, blue, green---> colour)
  - Each isolate is a type of the facet to which it belongs
  - Some isolates can be grouped together into sub-facets (blue for navy blue and sky blue)
  - Principles of isolates: principle of homogeneity (all isolates must be the same kind of thing), principle of mutual exclusivity (no two isolates can overlap in meaning), principle of joint exhaustivity (together, the isolates can cover all cases) -Ranganathan, S. R. (1945)
- Fundamental categories from Ranganathan: personality, matter, energy, space, time
- How to create faceted classification:
  1. Identify attributes
  2. Group attributes into facets
  3. Sequence facets (citation order)
  4. Sequence isolates within facets (order in array)
  5. Create notation
  6. Create specific classes

Create Assignment: Faceted Classification

Select a type of thing you want to sort, organize them in an order
Step one: choose a set of resources: what are these things? Who is searching/browsing through them? How many are there? What are the important or salient differences between them?
Step two: create facets and isolates.
- Aim for at least 4 facets and 12 isolates (doesn't have to be 4x3. for example one may be binary (yes/no) which another has 6 categories
- Make sure each facet is describing one and only one thing
- Make sure all isolates are mutually exclusive of each other
Step three: decide order of facets and isolates: facets should go in order of most important to least. Isolates should have some sort of intuitive order (best to worst, smallest to biggest)
Facets you create should work when additional items are added to the collection
Additional challenge: try to have facets cover the 5 areas Ranganathan put in his original approach to faceted classification (personality, matter, energy, space and time)
- Make your resources information bearing objects.
SUBMIT: a sentence describing the set of items and the context. Then a list of facets and their isolates (as numbered list or diagram). Then a few example items with their classification.
Do not feel like you have to submit perfect work; submit by the deadline and allow for peer review

Week 3: Thesaurus Principles

- Thesauri exist as a type of controlled vocabulary. These exist in many contexts, anywhere you are limiting the kinds of words that can be used in a certain context.
- One place you encounter controlled vocabularies is drop down menus (ex. Language lists where you're asked to "select your language") which have a fixed list of words based on a particular context.
- Thesauri: a specific type of controlled vocabulary which are instituted in information bearing objects (i.e. libraries)
  - Not only are we dealing with longer lists of contents, but they are also not all homogenous or identical
  - Want people to choose the term that is most specific w=for what they have in mind
- Example: detailed view of a scholarly article. Terms under the "subjects" heading is the thesaurus (not to be confused with "author provided key words" which are different)
  - With the thesaurus the same limited number of terms is being applied over and over again in the database
  - Usually supplied by information professionals (indexers) after reviewing the publications in a set, going through the articles, review the thesauri and apply the appropriate terms
  - This makes the list valuable when it comes to information retrieval.
- Hierarchical classification vs faceted classification vs thesauri:
  - Hierarchical classification creates more and more layers of increasingly detailed levels
  - Faceted classification lists out one asset per facet
  - With classification you are getting more and more specific at different levels in order to get only one location to embody an information resource
- Thesauri are not about finding only one location. It is a network of terms that have varying relationship with one another. We are also selecting as many terms as seem appropriate for a given resource.
- Syndetic Structure: The three principles for organization in thesaurus are
  - Equivalence: USE/used for
  - Hierarchy: Broader term/narrower term
  - Other relationships: Associative-related term
- Example: the ERIC thesaurus. The list of possible search terms is incredibly long and is used for research as well as indexing. Once you select one of the terms (i.e. Art History) you are provided with related terms and possible better alternative terms.
  - These terms are selected once an indexer has chosen the term that will appear in ERIC in the list of descriptors representing that article
- Relationships used in thesauri:
  - BT (Broader term): the larger umbrella/parent term (ex: intellectual history as the BT of art history)
  - RT (Related term): similar terms that are related to the selected term (ex: artists is a related term with art history)
  - NT (narrower term): the more specific term related to the original term (ex: art history is the narrower term to intellectual history)
- Equivalence relationships:
  - Equivalent phrases (things that are just minor differences in the way we express things differences in dialect or spelling, etc. Ex: aeroplane vs airplane)
  - Inverted forms (bilingual education vs Education, bilingual)
  - Acronyms and abbreviations (WHO vs World Health Organisation)
  - Antonyms (treated like synonyms as the resources about both terms will be the same ex: student retention vs dropouts)
- Lead-in terms: people often search subject heading systems and thesauri using queries. Therefore, there are terms that "lead" a person to an authorized heading.
- Example of equivalence relationships written out:
  - Aeroplanes USE Airplanes
    - Airplanes
      - UF Aeroplanes
      - BT Vehicles
      - NT Passenger Planes
- Upward posting: sometimes a specific term is "equivalent" to a more general term (Cats USE Animals), this can happen when users are interested in cats, but there aren't enough documents about cats to make a subject heading called "Cats". So instead "Cats" becomes a lead-in term that is upward posted to "Animals"
- Hierarchical relationships: the relationship between a concept and a more specific concept. It is trying to drive people to the most appropriate term for their query. 3 types of hierarchical relationships:
  - Generic: a link to a more specific type
    - Teachers NT School Teachers
    - Thinking NT Reasoning
  - Instance: a link to a particular example
    - Seas NT Baltic Sea
  - Partitive: a link to a part
    - Canada NT British Columbia
  - Through this successive linking of broader term-narrower term, you can get incredible depth. You can also have a lot of breadth with terms that have a lot of narrower terms
- Associative relationships: related terms that don't have another formal relationship type. Hierarchical relationships are transitive (the NT has to be a type of the BT). Associative relationships don't have to be transitive.
  - In order to avoid having too many associative relationships stick to these types:
    - Operations and instruments ((hairdressing RT hair driers)
    - Actions and products (roadmaking RT roads)
    - Causal relationships (accidents RT injury)
    - Field of study and objects studied (Xenobiology RT aliens)
  - Don't typically make these links between terms that relate to the same family (ex. Flies and bees wouldn't have an RT relationship)
- Scope Notes (SN): a brief note that describes how the heading is to be used. Many headings have no scope notes.
  - Ex: Flashback. SN: use only when a scene is included which occurred before the current narrative. When characters travel to prior events and experience them in real time, use Time Travel.

Assignment #2 - February 2 (Thesauri)

How should you approach choosing your 12 terms? Consider...
- What are the terms important to the topic?
- What concepts are the most important that are embodied in these things?
- Essentially, you are trying to think of ways people would be trying to find your topic.
What constitutes a "term"?
- The 12 terms you will select will be the "preferred" terms
  - e.g. Canines USE Dogs, "dogs" will be considered one of your 12 terms.
How many relationships should we have for each preferred term?
- Try one of every term type: i.e. broader/narrower term, related terms, equivalence relationship.
Tips
- Your terms do not have to form one cohesive network. They don't all have to relate to one another.
- Pick something that you know well/are familiar with. You can narrow it down, like to a sub-collection of your books or favourite T.V. shows, so that there is some recognizable overlap.
- When considering your antonyms, think in terms of concepts or phenomena, not in terms of tangible things (e.g. "pet health" and "animal neglect", rather than trying to determine the antonym of "dog").
- We likely all have experience with thesauri, but as users rather than in the set-up process. This is the assignment that we're most likely to feel lost/get wrong. That's okay - it's fine to hand in something imperfect.
Indexing
- Optional portion of assignment.
- At the bottom of the page (after you have listed your terms and term relationships), you can select an example and list the terms that match your example.
- For example, you choose a specific book titled "Everything Cats". You can then write Cats, Humour, etc. (whatever other terms that apply).
- The number of objects you index is options, but 2 or 3 is sufficient. Select things that are at two different levels of specificity, or things with some differences and some overlap.

Week 4: Content Standard Principles

Key questions:
- Which aspects of a resource do we end to represent?
- What are the constrains placed on those representations?
- How much time and expertise does the person doing the description have?
Central Concepts to include:
- Title, creator (the person, group, or event who is responsible for the resource), and the version (a date, edition, etc.)
  - Minimum items needed for a "known item search"
- The other ways we describe the resource (categorization, labelling, classification through thesauri) are when someone doesn't necessarily know what they want but they know what it's about.
Other concepts to record to help make sure the version of the thing we have suits the person's need best:
- What the resource looks like
- What time of resource it is (text, sound, video, etc.)
- Who published it (and where)
- How to identify the resource (ISBN, ISSN, DOI, etc.)
- What the resource is about (thesaurus terms, class number)
Schema and standard are not exactly interchangeable terms
- Schemas become standards when they regulated, managed and shared among institutions
- At a minimum a schema is an outline for what a record should contain
Two things the description of the resource is meant to achieve:
- Language and consistency
  - Because people are going to do things like sort things numerically in a system, it's important we are concise and consistent in the language we use.
  - There will be a lot of differences in the resources themselves, it's the job of the system that gives you access to it to be more predictable to give you access to the item.
  - Want the record created at one institution to be the same at another
- Bibliographic relationships
Abstraction and specificity: through abstraction (things like categories) we can group similar things together, not so concerned with the minutia of every single item we have, but rather with what makes them similar.
- This is the reason we have thesauri and controlled vocab that provide a fixed list of terms that don't exactly match every item, but are a fair representation of the concepts of the items so we can group things together and present them to the user.
- Helps ensure the user the item presented to them is the best one for their information need
Access: "access points" carry over from card catalogue days. You may describe many different things about the resource but only some of those are considered access points (fields that would be appropriate to be turned into filters, good targets for key word searches), whereas other descriptions may be on the details page for the item but they don't really feed into the search and filtering processes.
IFLA's General Principles on how to describe an item: convenience of the user, common usage, representation, accuracy, sufficiency and necessity, significance, economy, consistency and standardization, integration, interoperability, openness, accessibility, rationality.
- Some of these are contradictory and can lead to redundancy
- Some collection swill choose some of these to be primary over others
Example of a content schema with minimal rules: Dublin Core
- Started at a metadata semantics workshop in Dublin, Ohio, March 1995
- Sets up a common core of metadata elements for web-based resources
  - What are the things we should want to describe about the item? Things we need to describe about nearly every web-based item?
- Main goal is easier search and retrieval mechanisms
- Settled on a main 15 elements across three categories:
  - General Information (title, date, creator, contributor, publisher, rights)
  - Physical description (identifier (ISBN, URI), type, format, language)
  - Intellectual content (description, subject, coverage (place, time), relation, source)
- Dublin Core as a schema has very minimal descriptions about what you would put for each element, meant to be applicable regardless of what item you are looking at. Meant to have local rules spell them out in more detail based on what you are describing.
- Popular because it is so simple but it has rules in it in order to get more specific to the needs of every context. This is through Dublin Core qualifiers where you give more specific versions of the elements
  - Instead of "date" you would have "date digitized" "date modified" "date available"
- There are more terms than the core 15 that have grown as Dublin Core has had more implementations.
Metadata and Cataloguing:
- In practice they describe different kinds of approaches but there are many parallels between the two.
- Arguably, catalogue records are a kind of metadata (and cataloguing standards are types of metadata standards)
- Where you will see metadata more is when we're talking about born digital resources. Most metadata is created specifically to describe digital content.
  - Digital content is harder to control and metadata is specifically required to provide and constrain access
Choosing/Creating a content standard: need to consider the balance between functionality and simplicity (i.e. how do you know if you have too many elements or too few?).
- Support both human and machine use (what is easily understandable to a human is not what is easily understandable to a machine and vice versa)
- Support interoperability (the idea that metadata from one institution should be readable by another (standardization) and that meta data from one schema can be readable in another) and extensibility (the idea that metadata schema is adaptable for local needs and that an extended schema can be simplified for global needs)
Every schema or standard has:
- A set of values
- Some instructions on which attributes are necessary
- Some instruction on how to modify or specify attributes
- Some instruction on how to fill out the values
  - Possibly a reference to controlled vocabularies for particular fields

Week 5: Classification in Libraries

- Hierarchical Classification: Every level of the hierarchy is creating distinctions that are really local or specific to that tree. Every level of the classification is only understandable if you read it upwards to the root.
  - Every difference there is at every level is really specific to that branch of the tree
- Faceted classification does create the same sort of tree structure, at every level we are applying a new facet. The branches have to be interchangeable
- Hierarchical Classification is also known as Enumerative Classification: It enumerates (i.e. lists) all possible subjects as mutually exclusive classes. Requires a "primary" (i.e. initial) principle of organization (organizes in a top down manner)
  - Faceted just give the components and tools whereas hierarchical lists everything something can be.
- Hierarchical requires a primary facet of classification, needs to be your starting point as everything is organized down from that.
  - The Library of Congress Classification (LCC)- everything is one of its 26 things or "categories"
  - In the Dewey Decimal System (DDC)- everything is one of 10 things or "categories"
- Note: DDC is hierarchical up until a certain point, after the decimal point the other determinant can be a faceted classification
- 4 basic assumptions of universal, hierarchical classification:
  1. The world has a universal order
  2. There is a unity of all knowledge available (so that it is possible to generalize further and further and there is a connection to all our intellectual practices)
  3. There must be an essential similarity of class members (mutually exclusive categories that can be defined)
  4. Every entity has some intrinsic essence that allows it to be classified in a set way
- Standards for classification:
  - It has to be inclusive (there's a place for every concept)
  - It has to be unique (each concept fits in only one class and there's no overlap between classes)
  - It is expressive (the number/label should represent the concept)
  - It is hospitable (there's room to add more classes)
  - It's linear (have some sort of inherent order)
- You can't have polyhierarchy in a hierarchical classification
- The arrangement of classes: the meaning of any class term within a given domain can only be comprehended, as part of the domain, and within the specific context of that domain's classification structure
  - What any class "means"' has to be understood as being inherited from the categories above it
- Examples of where things are placed in the hierarchical system:
  - Post offices in DDC are a sub class of "public buildings", in the Korean Decimal Classification, post offices as a subclass of "communication buildings"
  - There are cultural backgrounds to the classification systems that mean different things and concepts fall under different primary categories in different places.
- The Dewey Decimal System: this system does represent a universe of knowledge. It draws from a "theory' of organization based on Francis Bacon and 19th century white, Anglo-Saxon, Protestant academia.
  - Francis Bacon's idea that all of human knowledge is one of three kinds (philosophy, poetry, or history) this maps on to which sets of Dewey's system are next to each other.
  - The system breaks down categories by number (ex: 400s are language, so 420s are "English" and 430s are "German")
  - Strengths: It's inclusivity (room for any concept at least in theory), expressiveness (the class numbers express meaning), constant revision (means for input and critique at both the class level and the hierarchical structure)
  - Weaknesses: Bias (based on Dewey's world view, based on the historical proportions of publishing), inconsistencies both within classes and between them.
  - An inconsistence: Heterogeneous principles of division
    - 640 home economics
      - 641.52 Breakfasts
      - 641.53 Luncheons
      - 641.54 Dinners
      - 641.56 cooking for special situations
        641.561 cooking for one or two persons
    - These are different principles of decision.
  - Because of these inconsistencies there are now tables in the system to provide explanations.
  - The 200s section on religion reveals the issue of bias as there are a significantly larger portion of numbers for Christianity than the other religions.
  - There have been some fixes of the DDC 200s such as an optional arrangement that keeps the numbers but instructs how to order them by chronology and region. They also have recommended a solution to re-number the section if that is what people prefer.
- In understanding a classification system there are certain questions you need to ask:
  - What are the first principles of division? What a political and philosophical commitments are implied?
  - What set or scope of resource is it meant to organize? Is it specialist or university? Which institutions use it for shelf-order and access?
  - What is the maintenance and revision status of the system? Who is primarily responsible? What are the means to propose changes to the system?

Week 6: Check-in

No Lecture this week

Week 7: Thesauri in Libraries

- Thesauri: Typically in the "subjects" field of a library catalogue listing. The are designating the concepts of a resource by selecting terms from a controlled list.
- Aboutness and Evidence: concepts, the way our information resources are about certain concepts and how we know the appropriate way to designate them
- Aboutness: Is the relationship between a work and its subject matter. Sometimes expressed as a quality of words, how well they indicate a subject.
  - Controlled vocabularies are meant to designate a works aboutness
- Principles of Aboutness: how do we know how to connect a word to a concept?
  - Evidence, epistemology, obsolescence (you do this work for systems you think are going to stand the test of time, not for short-term ones, so we will have to prepare our terms for the eventual obsolescence)
- Evidence/Warrant: in information organization we use the term warrant to describe valid types of evidence. Warrant helps to prioritize which concepts need to be represented in the system.
  - Warrant can help us determine the right word for a concept (preferred term)
  - Warrant can help us determine which words are equivalent in meaning (lead-in terms)
  - Warrant can help us determine how words are related (broader and narrower terms)
- Common types of warrant: literary (most traditional), scientific/consensus, user, and ethical warrant (newer).
- Literary warrant: the idea of using the literature as a source of valid terminology. Could refer to either only current collection or the universe of books on the subject. Determines not only the terms used but also the level of specificity for different subjects. Ideally, it presents the collection in the collection's own voice
  - Not imposing new terminology on the collection, don't introduce a term if the concept isn't already represented in the collection
- Scientific/Consensus Warrant: uses experts to define the valid terminology. Might attempt to cerate a system that describes reality itself. Usually makes the collection conform to the agreed-upon terms within a community of experts. Ideally, creates a system that reflects fields of knowledge accurately.
  - Gets around the restrictions of literary warrant and makes the system less about representing the system itself and more about reality itself.
  - Would have to decide which discipline in the community has the authoritative position.
- User warrant: using the users as a source of valid terminology. Lists of possible terms could be derived from passive collection (search logs). Or could be assembled by direct user research (surveys, interviews, etc.). Is most powerful in creating exhaustive lead-in terms. Ideally creates an intuitive system of users' natural language.
  - Must decide whether the system should represent an average user, or a multiplicity of user types.
- Ethical Warrant: unlike previous forms of warrant less a source of terms than a test of their validity for a system. Considers the religious, racial, cultural, gender, and language representation of the system (among others). Often the basis for revising terms to remove slurs, or to correct for bias.
- Library of Congress Subject Headings: the most commonly applied system applied in Anglo-libraries today. It's very old, it's in nearly every English book you'll come across and it's not quite a thesaurus. It's got subdivisions and there's a set process for how to propose revisions to terminology.
  - Common LCSH subdivisions: geography, chronology, form, etc.
  - LCSH hierarchies- it an be difficult to organize subject headings in a consistent manner. They are extremely long and complex.
  - Growth and change in LCSH: it needs a method to grow. Proposing new terms happens through the colloguing in publication process. This happens at the Library of Congress. This process is down for every new work, when a new term is identified it is then proposed as a revision to be added to LCSH.

Week 8: Content Standards in Libraries

- Cataloguing and Metadata: Which aspects of a resource to represent? What are the constraints placed on those representation? Different elements are important to represent to ensure people know if it will help them meet their information need. For example, name of creator, age, publisher, etc.
  - Helps findability and verification
  - Content standards tend to contain a lot of constraints around how we represent information (i.e. surname before first name), and how we should interpret information from the source itself.
  - Central concepts when representing information bearing objects: title creator, revisions mat to the resource (important for digital resources)
- RDA: current cataloguing standard for libraries in US, Canada and Great Britain. It is managed by library associations from those three countries. The changes that it made from AACR2 have two broad types of changes. Changed the format that the information is delivered in and added some more types of information. These changes are largely because of the online environment and the shift to user focused catalogue searches.
  - Less reliance on jargon in RDA than in AACR2 (the previous standard). The thought is people other than librarians will read the records so they must be able to be understood, also since we don't use card catalogues anymore, there aren't as strict limits on characters used.
- RDA in a MARC world: RDA is still happening in the same data formatting environment (MARC) that the previous rules were based on.
  - FRBR is a model to think about cataloguing in a new way
  - RDA is a standard that was built from that conceptual work, to specify a new kind of cataloguing system.
  - MARC has made some adjustments for RDA, but it's not an RDA system.
- RDA Toolkit: Unlike previous editions of library cataloguing rules, RDA is only available as an online resource, the RDA Toolkit. The full text is available by subscription, typically multi-seat subscriptions to libraries. It undergoes constant revision, the current platform was launched as a beta version in 2018 and the text itself is edited multiple times a year.
  - Toolkit launched first as a beta version in 2018
  - Regular updates and townhalls on it.
  - Still anger about the paywall on access to it
- Open Rules for Cataloguing: movement started by a group of librarians as an alterative to RDA. It is open source.

FRBR

Functional rules for bibliographic relationships (FRBR): conceptual model that greatly informed the development of RDA
- Tries to capture all the relationships that exist inn the bibliographic universe
- Among bibliographic entities: primary relationships among the abstract intellectual work, its various versions and their physical manifestations in items.
  - And equivalence, derivation and part-whole relationships
- Responsibility relationships
- Subject relationships
Among bibliographic entities is FRBR Group 1, responsibility relationships is FRBR Group 2, and Subject relationships is FRBR Group 3
Group 1: 4 entities are work, expression, manifestation, item. Work is realized through expression which is embodied in a manifestation, which is exemplified by the item.
- Work is a distinct intellectual or artistic creation (tend to have creators, uniform titles and subject headings)
- Expression is the realization of the work (have illustrators, translators, performers, language, edition statement, content type, word count, duration, etc.)
- Manifestation is the physical embodiment of the intellectual and artistic content ( had title proper, place and date of publication, publisher, physical characteristics like media type)
- Item is a n exemplar of a manifestation (will have call items, condition, provenance, location, etc.)
Superwork or work set: connects works with common origin (MCU, Romeo and Juliet, Alice in Wonderland, etc.)
Group 2: person, corporate body, family (owned by, produced by, realized by, created by)
Group 3: concept, object, event, place (works can be about…)

Week 9: Data Formats in Libraries

- Example: Dublin Core in Data Formats such as CSV (like excel), XML, in MARC (most common in contemporary libraries), a physical index card.
- MARC21: machine readable cataloguing for the 21st century. MARC records are classified by three digit codes that mean different attributes (title, author, etc.). MARC formats are already laying out what kind of data can go in each space, so if there is additional information you want to add, MARC itself has to change to make room for that information.
  - The 336, 337, 338, 344 fields are the quickest way to determine if MARC is using RDA, as those are the RDA fields.
  - Indicators: provide additional information about a field
  - Subfields smallest unit of information in a field
  - Main entry: 100 (personal name), 110 (corporation), 111 (Meeting) can only be used once across a work. Indicated the primary creator of the work.
  - Added Entry: 700 (personal name), 710 (corporation), 711 (Meeting), for other authors that are not the primary.
  - 245 field is the title statement
  - 650 field is the subject added entry-topical term. The second entry after 650 tells you which thesaurus it is from (i.e. 0 is Library of Congress Subject Headings, 2 is Medical Subject headings, etc.)
- Each content schema or standard has: a set of values, some instructions on which elements are necessary, some instructions on how to modify elements, and how to twill out the values. It isn't tangible until it has a data format
- Data formats determine: how to express connection between attributes and values, how to express connections between attributes, what character you can use, how many characters you can use, how records can relate to, overlap, add to each other, how you can use the records,

Week 10: Systems Integration in Libraries

Add notes here!

Week 11: Systems Integration Principles

Add notes here!

Week 12: Wrap-Up and Review!

Add notes here!