Created: 2025-09-25 06:40:02 - Touched: 2025-12-24 15:01:03 - Status: Seed

Lecture 1

You can do anything in a library if you understand how to organize information (I don't agree with this).

Information is not neutral

Knowledge is not objective, we all have our biases and experiences. Information is not neutral. Documentation is political. Classification systems reveal the political underpinnings. Organization is shaped by the organizers. Usage needs shape organization too. Information becomes meaningful when it interacts with context (and the purpose of the use).

Is information just its encoded form?

Does information exist if it doesn't have a concrete representation? Oral traditions for information would organize information differently than one that prioritizes written forms. Tools can only operate on a concrete representation of information, and therefore are inherently limited. "A document is any expression of human thought" -Buckland. The music isn't the sheet music document, every time you play a song it's different.

Related fields

Information theory is highly connected to communication theory. (cybernetic theory? sender->receiver electrical impulses? Cybernetics isn't sender, isn't receiver, it's the in-between. thought->encoding->transmission between sender and receiver->receive sensory info->decode->thought. Lots of opportunities to mess things up, lots of areas of responsibilities.)

Other

We're doing Collaborative Annotation this semester using hypothes.is, reread that maggie appleton article on silent reading sessions.

We don't document what kind of eggs and milk in our recipes, in thousands of years they may think they're coming from cats and dogs because they're our most common domestic animals currently -classmate Kit

Buckland "what is a document"

Buckland, M. K. (1997). What Is a "Document"? Journal of the American Society for Information Science (1986-1998), 48(9), 804-809.

What could be considered a document? Any expression of human thought or feeling. Documentation: the management of documents, "collecting, preserving, organizing (arranging), representing (describing), selecting (retrieving), reproducing (copying), and disseminating". Not just bibliography (describing documents). Needed documentation (the process of managing documents) because the number of documents was exploding (mostly printed text). Splitting responsibilities into specialists: librarians (collection development), bibliographers (describers), documentarians (access).

But once you have a definition for documentation (process), you can flip the question back around: a document is anything you can apply the documentation process to. Otlet (starting in 1934) argues anything which can inform you by observing it should count as a document, which would include sculptures, trees, animals, archaeological finds, educational games, etc. Conceptualist, not materialist.

Document : Any source of information, in material form, capable of being used for reference or study or as an authority. Examples : manuscripts, printed matter, illustrations, diagrams, museum specimens, etc....

Briet says documents are about evidence of a physical or conceptual phenomenon. Her examples:

Object --- Document? Star in sky -- No Photo of star -- Yes Stone in river -- No Stone in museum -- Yes Animal in wild -- No Animal in zoo -- Yes

The antelope in the wild isn't a document, but captured for study it becomes a document, and anything written about the antelope is by necessity a secondary source since the antelope itself is the primary source. It becomes a document by being placed in an organized meaningful relationship with other evidence. Materialist.

Ranganathan, Shores, Shera limit documents to paper and paper-related activities. So, not everyone agrees with a broad definition of what's a document. But anthropology and archaeology both require artifacts that aren't paper-related as documents which they can study. "there is always a meaning which overflows the object's use" Barthes says; objects signify more than they are.

Contemporary thoughts: social construction of meaning, perception of the significance and evidential character of documents. Relevance is situational. Signs can't be natural objects because they can'be searched for and found in nature, but rather have to have meaning assigned to them (birds flying south is a sign of winter coming, but "winter coming" is meaning assigned). Digital documents have even less physical representation than ever, so a conceptual understanding rather than material is ever-more important.

Information science road trip video

INFIDEOS (Director). (2022, March 23). Information Science Road Trip (10 Big Ideas of Information Science) [Video recording]. https://www.youtube.com/watch?v=9p50rd0zBTk

The Discipline of Organizing Chapter 1

Glushko, R. J. (Ed.). (2013). "Chapter 1: Foundations for Organizing Systems" In The Discipline of Organizing. The MIT Press. https://berkeley.pressbooks.pub/tdo4p/chapter/the-discipline-of-organizing/

Glushko, R. J. (2020). The discipline of organizing: 4th professional edition. Robert J. Glushko. https://berkeley.pressbooks.pub/tdo4p/

"To organize is to create capabilities by intentionally imposing order and structure." An organizing system is an intentionally arranged collection of resources and the interactions they support.

We organize physical things, information about physical things, digital things, and information about digital things. We might group boots together in our shoe organization, but books are more likely to be fiction vs non-fiction than hardback vs paperback. The more thought and energy put into the organization of the system, the more effective the access to the information is (allegedly); tradeoff of upfront work to organize vs work to access at time of use.

A resource is something with informational value. Distinguishing between the information resource and a reference to the information resource (like a card catalog entry for a book).

A collection is a group of resources selected to fit a purpose. A collection is a type of resource, but are unique in their use in organizing so they're talked about separately.

Intentional arrangement, human or human-caused or on-behalf-of-human organizing efforts. "self-organizing" don't count as intentional, since they only change with intention (ie a human intentionally adding an item to the collection). Conversely, the web has no central authority or repository, but creating a webpage is ostensibly a human action of adding to the web, and therefore the web is an intentional arrangement. "Organizing principles" are the principles utilized to create an intentional arrangement without assuming a specific implementation; "Making things easier to find" is a vague one, "alphabetical ordering" is a specific one. Co-location is perhaps the most basic one. An agent is one who can exert intention.

An interaction is "an action, function, service or capability" that makes use of the collection or its resources. "the marginal cost of allowing another access is nearly zero" cf Stallman https://www.gnu.org/philosophy/shouldbefree.en.html.

What is being organized? Why is it being organized? How much is it being organized? When is it being organized? How or by whom or by what process is it being organized? Organize by resource type (coins), or by purpose (memory institutions), or by audience size (personal to global), or by the tech used to implement them.

Custodian of information vs custodian of things. Authors are increasing classifying their own works into collections/generate their own metadata. Community content applications generate their own folksonomies and tag convergence.

Lecture 2 - 2025-10-02

Writing center and professional development center, certifications look good on resume. Thingifying something, turning it into a document, can be a violent act. A book is about something, but is her umbrella? Microfiche is a form, if it's damaged that might be part of the description. Metadata like tagging Mi'kmaw books to show up in a digital carousel on the library website. Cover color for those patrons who try to describe books that way (brief discussion of rainbow bookshelves and visual memory). How do we track things like name and gender changes if the metadata we store is only attached to the thingified document? How much does a document have to change before it's considered a separate edition? Some documents are "live" and continually updated by multiple authors (wp pages). Glushko's capabilities, do you want the capability to track publications by author across name changes, filter by gender or nation of origin, you have to impose an organizational structure to produce those capabilities. Physical placement within the collection can create capabilities and context. Grocery store organization isn't just about ease of finding an item, it's about promoting sales. Grocery stores change their organization intentionally to make people walk around more and possibly buy more stuff. Don't embed content in brightspace, instead link to library, and libraries get more engagement numbers.

Finally moving on, what is a document? Animate vs inanimate (Littlebear article). Buffalo child stone, geological info, indigenous stories of the stone as a living being. Gov wanted to flood the area where the stone was to make a hydroelectric dam. Someone blew up the rock. Fuck that's grim. They went diving after it was flooded to engage with it. It's not just a fossil, just a rock, just a necklace. "Can't just go in and flip on the lights and look at these objects, they're living beings and disturbing them would be rude." Describing indigenous things in English is shallow, loses a lot of meaning and context. Is an antelope a document? Thomas: distinction between document and record. Putting the antelope in a zoo to observe it to turn it into a document is a colonialist idea, observing it in the wild in its native context should also be valid. A swimmer.. who likes muffins? How does something go viral? Tagging on the tweets and tiktoks and whatever. A folksonomy that produces a meme, algorithms promote them. The organizational system that changes based on user interaction to tailor access to them, a moving target. Now we're talking about twitch and sea shanties, and in discord we're talking about pro-social media.

Group discussion about a specific book. Depending on context, how would you describe it? What metadata would you attach to it to make it accessible within users in that context? We talked a little bit about MARC and DC records. Thomas basically lists off the MARC-type records. Others talk about where in a library or other context it might fit into a broader collection, intended age range, themes and tags.

What is a knowledge representation?

Davis, R., Shrobe, H., & Szolovits, P. (1993). What is a knowledge representation? AI Magazine, 14(1), 17–17. https://doi.org/10.1609/aimag.v14i1.1029

  1. Surrogate. A substitute that allows for thinking about a thing rather than going out in the world and taking action in it. (wheels are outside the mind, but reasoning about how to attach them goes on inside the mind, what does this surrogate represent and how faithfully?)
  2. An answer to the question "In what terms should I think about the world?" (deciding on a representation forces us to reason about it in a specific way)
  3. The surrogate's idea of what intelligent reasoning means and what role it plays in that process, what you can infer from this surrogate, and what you should infer from this surrogate.
  4. It's a format that makes thinking/computing about it easy.
  5. It's a medium of human expression (how easily can we express this or that?)

Implications:

  1. Useful for defining properties we want representations to have
  2. Useful for characterizing representations
  3. Useful for settling previous arguments in the field
  4. Useful definition for research, broad idea of what's "important" about a representation, focus on representation of the natural world.

Contextualizing a scandal

Niergarth, K. (2024, April 29). Contextualizing a Scandal: A Brief History of Library and Archives Canada. Active History. https://activehistory.ca/blog/2024/04/29/contextualizing-a-scandal-a-brief-history-of-library-and-archives-canada/

Greer says Library and Archive Canada service has gone to crap since library and archives merged. Doucet (Director general, services) says we need to abandon RDA because "customers" don't understand it, which seems pretty dumb. Relentless funding cuts of "fat", putting things online doesn't make them useful, silos still exist, outdated practices mandated by law, researchers are left to fend for themselves.

The National Library of Canada, 1950-1968

Lunn, J. (1982). The National Library of Canada, 1950–1968. Archivaria, 86–95.

Started as the canadian bibliographic centre, lots of resistance to a national library lead to a dependence on academic and provincial libraries and on the US. Finally a national collection started in 1953, but nowhere to put anything. A warehouse in a pasture for 11 years, no one knew they were there. Finally got their own building in 1967. Difficult to hire given low prestige and pay scale; I'm reminded of the president who retired to instead be governor of NY I think. Making a unified card index took a very long time and involved cameras and an elaborate process of printing and cutting and stamping, had to send people to each library to get their inventory (but couldn't just take their existing cards, obviously). Cards came to central too fast to sort them all, so there were parallel sorted subcollections and you had to check all of them. Lots of duplication.

National bibliography, books written in canada, about canada, or by canadians. Lots of publishers forgot to send things, and finding other materials was difficult. Couldn't be published fast enough. Also published canadian theses/dissertations.

Fire, boxing of collection, warehousing. Lots of donations, wide variety of material types.

Eventually we get to 1968, an established library, unified bibliography, collection, and canadiana. 18 years of scrambling and setbacks and underfunding. But they got there.

Lecture 3 2025-10-09

Examination of several information profession job descriptions online.

Assignment 1 discussion. Non-digital is fine. Reference the textbook or other readings if you can/is appropriate. Privacy, collection, metadata. Within last 3 years. 1-20 sources.

Colonialism enforces a single worldview by violence of law. The goals of the people organizing information are directly expressed in its efficiency towards those goals. Nothing is neutral, even alphabetical ordering. Lots of discussions about names. Organization is a product of choices made by the organizer.

Thinking about your own collections. What are your organizational guiding principles? How does your system operate? Functional? Temporal? Produce in the fridge door so it's visible.

https://iportal.usask.ca/ Indigenous studies portal. Representation of a thing vs the thing itself; surrogate.

Guest speaker Caitlin Horrall LAC (she/elle) Director General. Caitlin Horrall is the Director of Description in the Published Heritage Branch at Library and Archives Canada. She was formerly the Corporate Secretary to the National Heritage Digitization Strategy. Gotta be self-directed, organize your own work. And probably some other stuff but it was so meandering I couldn't follow it.

Introduction: to classify is human

Introduction: To classify is human. (1999). https://doi.org/10.7551/mitpress/6352.003.0002 //FIXME put a real citation here

Classification is inescapable, whether it's organizing meds by expiration date and type or classifying areas into countries. Standards are similar but not the same as classifications.

Classifications:

  1. There are consistent, unique classificatory principles in operation (everything is classified the same way)
  2. The categories are mutually exclusive (items can't belong to two categories)
  3. The system is complete (no item is uncategorized)

Standards:

  1. Agreed-upon rules for the production of objects (texts, materials)
  2. Spans more than one community of practice or site of activity, and persists over time
  3. Allow for things to work together over distance and in heterogenous environments
  4. Often enforced by professional organizations or the state
  5. No expectation that the best standard will "win"
  6. Difficult and expensive to change

Large-scale infrastructures like ICD for medicine, used by a lot of different stakeholders with conflicting needs. Ethical concerns: how you define tuberculosis determines how long you're confined to the sanatorium; DMS definition of gay as mental illness. Work practice, making invisible work visible.

A Case Study in Indigenous Classification: Revisiting and Reviving the Brian Deer Scheme

Cherry, A., & Mukunda, K. (2015). A case study in Indigenous classification: Revisiting and reviving the Brian Deer scheme. Cataloging & Classification Quarterly, 53(5–6), 548–567. https://doi.org/10.1080/01639374.2015.1008717

"It is a challenge to classify Indigenous knowledge using conventional library classification systems." We're off to a great start there /s

LCC and DDC have inherent bias especially towards marginalized groups. Using a different way of knowing as the basis for a classification system is less likely to produce bias against that way of knowing and the people who employ it. No amount of changing the words on the LCC is going to change that the fundamental underlying assumptions it's structured on are eurocentric. Indigenous ways of knowing aren't exceptions to be classified. Some argue Indigenous materials shouldn't be cataloged in something like LCC at all, but be in a completely different classification system. Alphabetical or by province is not representative when the alphabet is Western and the provinces are colonial. Tiny slices of LCC or DDC ranges, often in the history section.

BDC developed across several libraries aiming to create classifications useful to the Indigenous folks using them, organized in ways that made sense to them (eg by tribe or region rather than alphabetical, by use rather than by Western genres). UBCIC started off all BDC but then some was reclassified in LCC and now it's a weird hybrid. Couldn't keep up with growing collection, so they abandoned LCC and drafted new revision of BDC.

UBCIC BDC added granularity and picked best concepts from other Indigenous organizing practices on Turtle Island. Classifying the LCC works into new BDC resulted in those materials being widely separated that were once close together. Refine classification based on who where when and what. Giant spreadsheets with former classification and new classification to create mapping. Bunch of volunteers etc to redo all the labeling. Even font selection proved challenging (1 vs l).

Revised BDC intended to be a "living classification system" that evolves with the needs of the community it serves.

The Taxonomic Urge (chapter 6)

Gartner, R. (2016). Metadata: Shaping knowledge from antiquity to the semantic web. Springer. https://web-p-ebscohost-com.ezproxy.library.dal.ca/ehost/ebookviewer/ebook?sid=7b3bde0c-2e49-4c8c-8e77-aefc52fbf65d%40redis&ppid=pp_65&vid=0&format=EB

Do I want to read this book? It seems cool.

Taxonomy: principled ordering. Putting things in categories is human nature. Things that cannot be categorized or are out of place often considered dirty or even revulsive. Taxonomies are hierarchical and composed of layers at equal height, those closer to the root are "higher" or superior to those below them. Language example, hierarchical structure of a sentence "colorless green ideas sleep furiously". DDC 27 digits for Canadian tractor industry highly precise (but uneven distribution). "Pre-coordinate indexing" adding classifications and disparate subject headings at the time it's added to the collection in the hopes they're useful to searchers in the future. At any point in the hierarchy, can get broader context by looking up, find more specifics by looking down. Position in the hierarchy can provide context for what a specific work is about. Roget's thesaurus tries to create a hierarchy of words so that you can find ones "close" to get synonyms and find just the right word for what you're trying to communicate, applies taxonomy to vocabulary. Is that what lexicography is? Modern thesauri can be subject-specific and create controlled vocabularies of "preferred" terms along with synonyms so that searchers can more quickly arrive at the controlled term to use in their search. Attaching these terms to specific items in the collection would allow a single item to appear in more than one spot in the hierarchy, eg pig in both cloven-hoofed and cud-chewing. Higher lower and adjacent in hierarchy doesn't tell us a whole lot about the specific relationships between terms; semantic web metadata can help by classifying relationships (and of course the classifications are a hierarchy, right?) Controlled vocabulary allows for tagging of resources and allowing the searcher to combine their own terms, less restriction by the way the cataloger was conceptualizing the search process, post-coordinate. User defines which terms are close by in the resulting hierarchy for searching? Allows for boolean searching. Hierarchies are now guides rather than the only way to find something (categories in online stores supplement search).

Lecture 2025-10-16

She's sick. Other people are out sick. Online is an option if you're sick but she encourages people to come anyway. International travel last week, she's sick this week, what a curious correlation.

When creating a classification system, you have to define what fields you're going to record and what values are possible for each. Database field datatype and enums. Another Svenonius quote, maybe she should have made that the class textbook. Taxonomy becomes more formal, rigid, systematic. https://xwi7xwa.library.ubc.ca/collections/indigenous-knowledge-organization/ first nations classification. Cutter numbers (and letters) for getting enough specificity. What is an accession number?

The house archives built

Berry, D. (2022, June 22). The house archives built. Up//Root. https://www.uproot.space/features/the-house-archives-built

Discusses the systemic racism embedded in archival principles and classifications, the appropriation of history by the institutions that collect it, how being inside the institution with credentials and citations doesn't solve the problems. How frustrating it can be to speak up day after day.

What metadata is and why it matters (chapter 1)

Gartner, R. (2016). Metadata: Shaping knowledge from antiquity to the semantic web. Springer. https://web-p-ebscohost-com.ezproxy.library.dal.ca/ehost/ebookviewer/ebook?sid=7b3bde0c-2e49-4c8c-8e77-aefc52fbf65d%40redis&ppid=pp_65&vid=0&format=EB

Data about data. What is chosen to be captured and what is ignored and lost? Not found in nature. Not objective or neutral. Google map is geospacial, subway map is just lines and stations. Three types of metadata:

Can't just write information down, you have to be able to find it again. Metadata is the glue that holds together data into information, informations into understanding, understandings into knowledge, knowledges into wisdom.

Enquire within upon everything

Carlson, S., Lampert, C., Washington, A., & Melvin, D. (2020). Linked data for the perplexed librarian. American Library Association. http://ebookcentral.proquest.com/lib/dal/detail.action?docID=6167564

Tim Berners-Lee ENQUIRE used bidirectional types relations (describes/is described by, made/was made by). Out-of-sequence reading, block insertion/footnotes. Critical threshold of information beyond which the system's usefulness would be enough to continue contributions to it. Paragraph tag in HTML both descriptive (indicates human writing) and procedural (tells browser how to render it). I'd call that semantic and presentation.

Semantic web: structured data about concepts people places events dates etc, meaningful relationships between nodes, capture the connections to provide meaning instead of just strings. Practical limitations: end up with highly detailed silos or broad shallow stuff. "Imperfect but published." Also-known-as records explicitly state this Charlie Chapman is the same as the one in this wikipedia article and that openmoviedb record.

Lecture 5 2025-10-23

"Metadata is a love note to the future." Creating capabilities by human-initiated categorization and structure imposition. I feel like we already had this class. I hope eventually she gets to the part about linked data and the semantic web, which is allegedly the topic of this week's class.

RAD: archival description (RDF?). What is a fond? Respect the order provided by the creator and also the ways that those who want it will try to find it. Libraries are based on individual items, but archives are based on bins. LGBT Seniors Archive. Community driven collection/archive bring their expertise about how that information should be organized and described. Key subject areas and featured collections. Surrogates, stand-in, metadata, bibliography item, index, card in the catalog. Describe it with enough detail to allow the user to decide if they want to access the item itself, not so much that they drown. RDA resource description and access (libraries), RAD rules for archival description (archives). Standards that guide decisions on what to include, what sources should be used for each field, how they should be formatted, etc. Not intended to be specific enough for machines to parse directly, more about what humans should record. Content guide, not computer-level specifications. Standards become invisible, taken for granted, the only way we can do this we don't think about alternatives, biases become hard to see and harder to change. Dublin Core Metadata Element Set, MARC machine readable cataloging (older), Resource Description Framework (linked data), much more rigid and specific (intended for machine consumption). DC defines the elements but doesn't specify their content (eg date format, controlled vocabulary for subject headings). Sarah MacKeil classmate notes similarities with organizational style guides. Metadata Standards had some notes about some of these standards.

Standards are necessary for cross-organization coordination, eg publishers send MARC records to the library that matches the book they ordered (to classify is human). https://jennriley.com/seeing-standards-a-visualization-of-the-metadata-universe/

Indexing (like the back of the book) vs indexing (like a bibliography). Include keywords, not? hierarchical? Possible application of LLMs. What's the difference between an index and a bibliography, is it access method? Index is list of keywords with items under each (or page numbers), bibliography is list of works on a similar topic. So maybe an index is a flat keyword-grouped collection of bibliographies? Classmate Maeghan: "The way that I understand it is an index is signified by abstracts and keywords, whereas a bibliography is source information that sometimes uses subject headings for categorization. Bibliography is less concerned about the content and more with the location and source."

Is there value in data about data about data? Maybe a record in our system could have a MARC record and an RDF record and each one would be tagged with its standard and the last time it was updated? If the cereal box is metadata about the product (name, manufacturer, manufacturer contact info, etc), the nutritional label is a subset of that with its own standard. Are metadata standards composable? Sugar mg references SI, date in an RDA record is (maybe?) iso8601, latin character set, lots of composition. Open Food Facts database that lists food item information, an index? Metadata is structured information that allows us to identify, discover, assess, and manage the entities (in a collection). Nutritional info label as metadata for a food item.

More discussion about categorization as a reflection of the values and uses of the community served. Audio metadata instead of textual. Capture of cultural context within metadata (Indigenous? North of the arctic circle)

LLMs trained on a Western corpus will perpetuate Western ways of knowing and biases, including when "generating" metadata like descriptions and abstracts. Just like we've been talking about in class today it becomes invisible, harder to identify and correct biases.

Descriptive (finding, about the object), administrative (pixel dimensions, rights (legal, Indigenous, access (getting it) is different from rights (what you can do with it once you have it)), preservation (providence, bitrot/digital file formats, format conversion history)), structural types of metadata (and standards, DoCO document components ontology, make sure it renders correctly, how it's put together, footnote navigation, chapter ordering, subtitles, language identifiers for audio tracks on a movie, much of it autogenerated by tools). Most standards focus on one type of metadata/encoding, but sometimes there's crossover.

Linked data and semantic connections, express meaning. Allows us to answer "how" questions. Linking books to an author allows us to navigate from the book to other books by the same author. Not just a connection, but a type of connection, connecting specific fields between items. "Things not strings." Could look at the size of the card catalog to get a sense of the scale.

Stars of goodness:

  1. Data is openly available in any format
  2. Data is in a structured format
  3. data is in non-proprietary structured format
  4. follows open standards like w3c's rdf and using uris. (DC alleged example since 1997?)
  5. includes links to other open linked data sources.

Subject->predicate->object. Predicate is the semantic meaning of the relationship. Triplets. LINCS Project linked data in cultural heritage. With enough links you can start asking some really complex questions. MARC isn't used outside of libraries so it's hard to connect that with other systems. BIBFRAME a successor to MARC allegedly (adds URIs).

Data Feminism - The Power Chapter

D’Ignazio, C., & Klein, L. (2020). 1. The Power Chapter. Data Feminism. https://data-feminism.mitpress.mit.edu/pub/vi8obxh7/release/4

I should read this book.

Serena Williams pregnancy instagram post about pregnancy complications, Black women 3x more likely to have pregnancy and childbirth related deaths even accounting for education and income. No US federal tracking of this data. Intersectionality of famous tennis player vs Black + woman. Data feminism requires identifying and examining systematic oppression underlying our daily lives, harder to see the more privilege you have.

Power and the matrix of domination

Power: "the current configuration of structural privilege and structural oppression, in which some groups experience unearned advantages" while others do not because of system design. Intersectionality and axes of privilege/oppression, "few pure victims and oppressors" (Collins 1990). Matrix of domination is from Collins, structural, disciplinary, hegemonic, and interpersonal are areas where privilege/oppression can be observed (not just with gender and race but other dimensions of identity too). Structural is laws and institutions. Disciplinary is administrators and managers who enforce the laws and policies via beurocracy and hierarchy. Hegemonic is culture and media, what is possible to be known, representation and access to information (sometimes memes). Interpersonal, individual experience; how does it feel for a male relative to give me an anti-suffragette pamphlet? How it feels to know these systems are designed against you.

Minoritized (in opposition to a dominant social group that holds power) vs minorities (fewer total person count).

Data science by whom?

Data science practitioners are rarely men, rarely BIPOC, and those numbers are getting worse over time. Algorithms used in hiring and university application processes are trained on prior approved applicants, reinforcing bias. Biases in Google search results. Ignorance of one's experienced privilege is a "privilege hazard." The burden to identify the harms of privilege consistently fall on the unprivileged group. Facial recognition better detecting Black and brown faces isn't necessarily good when those techs are used for oppression. Addressing that takes more than just making a representative training set for facial recognition AI. Needs to be addressed in all four domains Collins identified.

Data science for whom?

Missing datasets. Identify them, characterize them, advocate for creating them. Sometimes you have to do the data collection yourself. Counterdata, agnostic data collection, data activism, statactivism, citizen science. Those in power often see lack of data as a desired outcome because it protects their power, eg femicide in Mexico.

Data science with whose interests and goals?

Opposite problem, over-surveillance of minoritized groups. Inverse of facial recognition for university testing, training "detect suspicious behavior" on minoritized groups. Flawed assumptions: more data is better, all data points are neutral. Scarcity bias: "the idea that there are not enough resources for everyone so we should think small and allow technology to fill the gaps." Target pregnancy algorithm scandal, teenager loses the right to privacy around her own body. Less hesitance around exploiting women's data, more focus on prioritizing profits. Data science is expensive, so it usually is undertaken by the wealthy in service of the wealthy/powerful: universities, governments, and corporations. This applies to areas other than data science too. Only megacorps can afford to build data centers. Exploitation of data is similar to exploitation of oil in who it harms and who it disadvantages. Data is extracted from people and sold back to them. "there is no profit to be made collecting data on the women who are dying in childbirth, but there is significant profit in knowing whether women are pregnant."

The Power to Name

Olson, H. A. (2001). The power to name: Representation in library catalogs. Signs, 26(3), 639–668.

A "universal language" meaning a controlled vocabulary hides the exclusions under "neutrality." This is "both a limited system for the representation of information and a universality/diversity binary opposition." Examine 3 texts in two modes, subject headings (verbal representations of topics) and classification (topics for physical/electronic organization for browsing). Cutter's rules 1094, DDC 1932, LCC and LCSH (now).

Cutter: singular public whose convenience should be prioritized, the tyranny of the majority. Artificial binary between chaos and order, where order is hierarchical with leaf nodes more specific/narrow. LCSH is the modern implementation of his rules, Women being the authoritative term while woman and human females etc are unprivileged synonyms. Broader, narrower, and related terms. Entries also include suggested LCC numbers. Comparison, "gifted women" present but "gifted men" absent. LCSH is based on what writers write about, and they write books about gifted women but not gifted men and that's reflected in the headings, but this presents a clear bias in the collection and authorship. Male prostitute but no female prostitute. Prostitution has variants "female prostitution" and "prostitution, female", no male. Prostitutes is the same, female prostitutes are considered a variant but male prostitutes is a narrower term. Thus, LCSH presents as "universal" what's actually just patriarchy (positioning of the male privilege as the default).

Dewey: Universal language necessary to prevent "confusion", create uniformity which is necessary for finding items by grouping similar ones together. Classification is necessary to create meaning out of chaos, mind over matter, reason over emotion. Where Cutter had logical/absurd, Dewey has army/mob, with the army being disciplined and orderly; but those on the wrong side of the army certainly aren't treated well. Decision to use arabic numerals and base-10 excludes other numbers and everything else. How do you divide all of human knowledge into 10 categories cleanly? "Procrustean: aiming or tending to produce uniformity by violent and arbitrary method" Oxford English Dictionary. Railroads are efficient but procrustean, cars can go right to your door, sucks if you live far away from the train line. Focus on the categories rather than what goes in them. Subject classification as epistemic cartography: a map for where you find the information you want. Equality implies difference, otherwise we wouldn't need to advocate for it, but "equality might wel be defined as deliberate indifference to specific differences." If you split first by gender then by race, all the women are together but all the Hispanics are not. Division decisions privilege some and oppress others.

Talking Back: Thinking Feminist, Thinking Black (bell hooks) gets classified as Afro-American women, Feminism - United States, and Hooks, Bell. Nothing about "voice" talking back to authority the way hooks uses the term. No LCSH heading exists for that concept of self-expression even though it's "not a new topic." Also the rules require capitalizing the first letter of each word even though hooks chooses explicitly not to do that, not giving hooks the authority to name herself within this system, so it's not just failure to have the term but it overtly silences some voices. The more marginalized the idea, the more obscure its placement in the catalog. Cataloging standards are living documents to varying degrees, and yet here we are.

So what do we do about it? Find ethical ways to engage with those the system doesn't accommodate. Relinquish some power as information professionals to the people whose works (identities) we're cataloging. Three techniques: "to apply technology in innovative and subversive ways, to stretch standards such as LCSH and DDC, and to adopt an active stance by creating spaces in our boundaries for the voices of those who have been excluded." Bibliographic text search turns up a lot of items for "voice" but many irrelevant and won't turn up hooks's book, and full text search will turn up even more irrelevant results. Could use multilingual support to avoid some "variant terms." DDC has "options" to classify or shelve things differently, eg "put all of Canada at the top of 810 instead of splitting by language" could be used to put all queer authors together, or reorder the precedence tables. Author created mapping from feminist concepts to DDC sections so at least you can "browse" from a feminist perspective even if you have to use DDC to find it on the shelf. We can do these things, but we must also advocate for changes in the original systems.

Imagining: creating spaces for indigenous ontologies

Duarte, M. E., & Belarde-Lewis, M. (2015). Imagining: Creating spaces for indigenous ontologies. Cataloging & Classification Quarterly, 53(5–6), 677–702. https://doi.org/10.1080/01639374.2015.1018396

LCSH is highly subjective and often problematic. There aren't extant examples of hundreds of different Indigenous knowledge organization and categorization systems, but we can (with effort) imagine what such a system would be like, and let that imagining influence our work. Non-Indigenous need to set aside expectations about documents by for and about Indigenous peoples, and instead "acknowledge the reasons why Indigenous peoples might prefer to develop their own approaches." And then Indigenous do the developing of organizational systems and forge partnerships. Follow the lead of Indigenous knowledge organization specialists. Is this going to be the answer to the "okay, I need to identify all of the problems caused by my privilege, but how can I do that if I can't see them" question?

Indigenous experiences of colonialism are different from the experiences of ethnic minorities, though Native American sounds a lot like African American and Asian American and so on. "Native American" is a colonial aggregate term for hundreds of Indigenous nations.

When you consider texts to be the primary/only form of knowledge artifact, the majority of Indigenous knowledge documentation disappears.

  1. misnaming, or using Western-centric terms to describe Indigenous phenomena
  2. using parts to describe a more holistic phenomena, or the reduction, removal, and de-linking of a piece of a knowledge system from a greater ontology
  3. emphasis on modern nationalist periodization, inclusive of the notion that history as it is written by the colonizers cannot be changed
  4. emphasis on prohibiting changes to practices that would upset the efficiency of the existing standardized schema

Yeh asked for Native American topics to be addressed more appropriately within LCSH and got a "no" response, citing "historical remnants", literary warrant, and efficiency. Obviously that's a bad answer, but the request had its own problems: it's still trying to force Indigenous knowledges and experiences and history etc into the colonial organization system, effectively a form of assimilation. Even if you got approval, the oral and tradition history etc is never going to fit in LC anyway.

Imagining:

  1. Be aware of how colonization subjugates
  2. Identify tools and institutions that decolonialize
  3. Formal acknowledgement of IK in context
  4. Identify Indigenous knowledge specialists to work with
  5. Let Indigenous thinkers experiment

Lecture 2025-10-30

Apparently I didn't care enough to note anything.

Notes on the remote attendance policy. As an immunocompromised person, I have a lot of misgivings about attending class in person. Being in a confined space with poor ventilation with 60 other people is a high risk environment for me. I'd like to thank Dr. Allison-Cassin for her generosity in implementing the policy to allow ill students to attend class remotely. I would also like to extend my gratitude to my fellow students who have used this policy, taking on the technological and social difficulties that entails, to keep me and our other classmates healthy as Halifax sees high community transmission of respiratory illness going into the end of the year. Thanks for keeping me safe.

Lecture 2025-11-06: Archives

Dal's catalog uses RDA+LCSH+LCC (presumably encoded in MARC). Archives use RAD. Standards are useful to people organizing collections but rarely easy enough for laypeople to use directly. Metadata Standards

Dspace/archivematica/other tools for hosting your own archive, open source.

Brief introduction to Dublin Core. Don't be too worried if you have trouble with some fields during the assignment, just note it in your reflection. Stacy to host DC workshop during reading week.

Archives! Archives focus on the people/organizations that generate the materials, preserved because of continuing value. Lots of types, physical items or digital or paper documents, medical or gov etc. Archives are records, but archives are also the organizations that collect preserve and organize those records. Archivists are the ones who decide if there's continuing value. Archives are mostly unpublished materials.

Fonds: the entire body of material generated by a person, or family, or organization. Pronounced "fonh" because it's French.

Indigenous knowledge doesn't fit very well into the definitions of archives on the slides. Archives are focused by their mandate (what they acquire and who they serve). Interesting that archives can opt to not accept materials when they're presented, whereas libraries weed materials that are in the collection but wouldn't acquire them in the first place if they didn't want it. Sometimes archives will duplicate each other's collections or there will be overlap because of overlapping mandates. Archives do choose to discard things during reevaluation but it's not part of the frequent collection development process. Libraries don't tend to have preservation as part of their primary purpose. Both keep things, both discard things, libraries are more focused on individual items and use, archives more focused on fonds and preservation. Sometimes documents may end up in both places, but they're described differently and organized differently and different standards. There are cross-institute archive indexes but they're not comprehensive; sometimes they're at item level, or fonds level, or collection level. Libraries are very keen on exchanging records to save labor, and archives typically aren't because their contents are unique.

https://link.springer.com/article/10.1007/BF02435628 describes the power dynamics and critical analysis of archives. Archives aren't neutral. Noting a cross-reference between two records or people or collections is a political act and so is not noting one.

Lecture 2025-11-20 Ethics and presentations

Lecture for an hour and then do some presentations?

How does ethics relate to justice?

Professional ethics give us the guidelines for how to behave in our professional practice. Applied ethics is how to resolve dilemmas. Do we have an ethical responsibility to stand up against the implementation of biased standards etc?

Ethical citation practices, how do I choose an ethical journal to publish in? Had a hard time with parts of that conversation, esp the part about border crossings, censorship, safety, the expectation that we must stand up for what's right vs the actual cost that privileged people don't see.

Local Contexts (different from copyright). Provenance labels.

Mesh has tutorial modules, substances and publication type are separate fields. is pubmed not public access? pubmed vs ovid? AtoM (access to Memory) tool for RAD record creation and storage. emTree vs mesh

2025-11-27 Lecture and presentations

Learning about DDC and having tech issues as usual. DDC replaced fixed location. OCLC bought DDC in 1993. 2000 first web version, 2013 last published edition and now we just do rolling deployments. Most widely used? Browseability/adjacency/discovery/proximity. Flexible/easy to use/efficient bc numeric. Lots of biases, weird choices. Bigger usage means bigger resistance to change. Limitations of a linear organization system why aren't things grouped together the way I expect them to be? "No consumption, no juice" what does this mean.

LCC wasn't based on DDC but has similar hierarchical nesting structure. The "call number" includes letters. Cutter numbers are created to increase specificity, can have a second cutter for additional specificity. Plus date, plus unique numbers if required. Uniquely assigned to each work, standardized, well funded (for now). Difficult to use if you're not familiar with it, lots of bias and offensive terminology and choices. What does ecumenical mean?

For both DDC and LCC groups: If I have a specific book in mind, how would I find out what DDC or LCC call number was already applied to that book? I don't want to try to assign my own and possibly end up in conflict with a more authoritative catalogue. It may come with it from the publisher, or be on the front matter, or you can search LOC or worldcat, or some other way to import it automatically from someone else.

For Nomenclature, could this system be applied to a tool library, or other libraries of things? It sounds like very yes.

Nomenclature for museum cataloguing. 15,000 preferred terms available on the web. Small enough that requesting changes to terminology is done by email (or a form?). Supports interchange despite the unique nature of the objects contained in museums. 10 major categories (noticing a theme, DDC had 10, LCC had 21). What is a hossier? Amputation knife?? I wonder if this would be useful for library of things?

Intellectual property rights, open access, creative commons. rightsstatements.org. Cory Doctorow information wants to be free considered harmful (toaster? bread?).

2025-12-04 Lecture and presentations

CIDOC CRM supports relational model for museum artifacts but it includes things like historical events and people that relate to the object's history. Look into this more! No controlled vocab, gotta use another standard to fill that gap. CIDOC became a proper noun, the acronym'd term it referred to is no longer used. Explicit on formatting like fonts and colors? Recommends draw.io for creating the diagrams, imported diagram formatting from a github of a member organization but the standard contains sufficient detail to do it yourself.

Homosaurus is adorable. International linked data vocab of lgbtqia2s+ terms, intended to supplement other vocabs. Includes terms that may only be found online, medical terms, and controversial terms. Latest version include Spanish translation. Community built and peer reviewed, open to public submissions. Wonder if they'd be interested in plural queer terms? Some duplication, short entries, not always alphabetical, search function is limited, missed some nationalities. Missed opportunity, invite classmates to bring dinosaur-related stuff to the presentation. URI standards for controlled vocab REST to find parent terms and subterms, json or xml standardized response of term elements like description, alternatives, etc? Princeton university library uses HomoIT to supplement LCSH. Homosaurus includes machine-readable downloads (n-triples, json-ld) under cc-by-nc-nd. 650/655 in MARC are subject headings, OCLC has cataloguing tools and they'll make available alternative vocabs. Biden admin funded the Spanish translation.

Brian Deer. "To class if human" Bowker and Star. "Classification, however, although necessary in systems of language, contributes to hierarchies of power" Vaughan 2018. His organization changed/evolved/specialized for each organization he implemented it in. The specialization is a strength but also a weakness, requires ongoing training, unfamiliar to those used to DDC or LCC. Xwi7xwa "whei wha" only academic Indigenous library branch in Canada; based on Brian Deer but specialized and modified. Geographic instead of alphabetical. Pronunciation is hard, it seems like they're trying. ACCI adapted Xwi7xwa adapted to Quebec not just British Columbia (Indigenous and also French). ICRC at Simon Fraser University, based on a bunch of previous adaptations of Brian Deer, updated some terms to reflect values centering Indigenous peoples and communities. Seems more adaptable and non-standardized which makes exchange more difficult but better fit for served communities. Would the call numbers for Brian Deer/Wxi7xwa ever be written using Indigenous scripts instead of Latin characters and Arabic numerals? No, uses cutter numbers instead of author names. Can be used as a classification system (where it sits on the shelf) but also adapted to use as subject headings (aboutness).

CIDOC CRM vs Homosaurus, official organization vs grassroots community effort.

Break.

Discussion of "AI" and machine processing. Can a machine identify what is "meaningful"? No. You can't provide reference service if a collection isn't organized. Model poisoning, please stop externalizing your costs directly into my face. Lots of things might be relevant, should probably be a 4-year degree. Abstracts are copyrightable and are being moved behind the paywall.

Pages that link here: