Shelf-order: How important is that to you?

For Ranganathan, the arrangement of books on shelves is "a problem of mapping the multi-dimensional space of subjects along the one-dimensional space of a line." I can see this played out in our discussions for the Open Shelves Classification project. I'm curious about how we are dealing with this problem in our digital world. How would we use the OSC to move from highly manipulable digital spaces to shove or retrieve a book arranged in linear order on physical shelves? How can we use the intellectual spaces structured by classification schemes to better navigate our physical spaces rather than allowing the limitations of the shelf order to define the OSC? How would you use a classification scheme for shelf arrangement? What expressions of the scheme would you use?

On the level of personal collections, I think that I might use descriptors (or even notations) from the scheme to tag digital records of my books and I might use its broad categories to group and label my books on the shelves. But I will NEVER take the time to print out notations on labels, stick them on to the spines of my books and then shove into and retrieve them from their particular slots in a linear ordering on the shelves.

Of course, it's another story for bigger collections used by people other than me and my family...

My answer is: OSC is *only* a shelf-order system. I do not believe you can start with an abstract, multi-dimensional system and map it onto a a 1-d space without serious problems. You should design for what you're going to use it for, directly confronting and embracing the limitation as a poet a particular narrow form, not write the the Waste Land and try to fit it on a fortune cookie.

And the world doesn't need a new attempt to bottle reality in either trees or facets.

Hmm. This is going to require some emptying of the mind of past habits as someone else in this forum expressed.

There was a non-tree (but faceted), notation-free classification for fiction developed for public libraries in Denmark called Analysis and Mediation of Publications (AMP) and used in a system called BookHouse (http://www.db.dk/bh/lifeboat_ko/SPECIFIC%20SYSTEMS/book_house_system.htm) but I don't read anywhere how it was used as a shelf-order system. Any LT user who knows if this system has been used for this purpose or if it is even still being used?

Ok, remember, the classification system purposed if for public and school libraries and not intended for home use. Of course, those so inclined might indeed classify their personal/family collections however, it is not the intension!

Shelf list/order then is very important. Part of the usefulness of shelf order is browseability. Given the poor indexing of many PACs and search engines, a browsable collection is useful and necessary. If not, a classification system would not be needed. We could then put the books on the shelf in any order and barcode them. The barcodes when scanned would tell us where we put the book last. This however would help no one find anything other than the one book they knew the title of.

I am all for a new classification system but not for no order!

In designing the OSC, to what extent should we consider the following scenarios in locating items on the shelves?

1. Patron searches the catalog, identifies and selects item wanted, notes the call number, goes to shelves, locates the item, checks out the item.
2. Patron searches the catalog, identifies and selects item wanted, places hold on item, library staff go to shelves and retrieve item, patron picks up and checks out item.
3. Patron gets a map of the library, goes to section of subject area interested in, browses area, identifies and selects item wanted, checks out item.
4. Patron asks help from a reference librarian, reference librarian helps patron identify and select resources, reference librarian goes to shelves and retrieves item for patron.
5. Patron randomly wanders around the library, guided by signs and labels, sees something interesting, pulls it out from the shelves and checks it out.

I guess what I'm trying to understand is who browses the open shelves? Who retrieves items from the open shelves? Who shelves? Should we design both for patrons and for library staff?

Tim
Excellent comment. I am neither a librarian nor a cataloging guru, just a humble engineer, but I think the statement you make is the basic assumption which must be kept in front of you (ie the design group for your project) to guide the direction and activities of the effort. It is the clarifying objective which will maintain the focus on an attainable goal and not allow the program to grow out of hand. It is my experience in systems design that a concise real world objective, like the one you have stated, is of immeasurable value in being able to resolve the inevitable problems of scope that arise during a program and and allow you to reach a solution.

5: I think your scenario 3 is the ideal we are striving for, wherein we create a system that improves the ability to browse in a meaningful way. But I also think that what may be beneficial for patrons can help the staff just as much.

5>

I think this system is primarily designed for scenarios 3 & 5. Librarians are going to pretty much use the catalog when looking for information, especially in regards to any request that they haven't seen a hundred times before. So 2 and 4 will work even with really crummy shelving systems.

There is also some interest in scenario 1 for patron retrieval and it also is important if you replace the word patron with librarian. You want to make the numbering system relatively easy to follow. I don't think LC and Dewey really meet that basic level of usability, although I'm not sure any system would without a significant level of user training.

5:
I think 3 is the primary scenario, but as a shelf-order system it will have to encompass all the scenarios you listed. Remember that in a public library setting, librarians, clerks, shelvers and possibly volunteers will all need to be able to use this system to retrieve or replace items on the shelf.

My usual modis operandi at the library is to locate the kind of thing I'm looking for in the subject catalog write down the two or three totally different call number neighborhoods that that subject area seems to be shelved under and go browse those shelves.

I think 1, 3 and 5 are the scenarios. Whether a user consults the catalog or a map is merely an artifact of the need to get to a specific location vs a general location. The important fact is that the user gets to the location he needs and then, based on time available and needs, browses for related and scerendipitus (sp) connections. Scenario 5, a random walk guided by signs is actually the same trial as scenario 3 with a weaker starting objective in terms of a particular item of interest.

Scenario 2 and 4 both entail two major operating differences from the other scenarios. First they require the use of library staff to retrieve the desired item. This immediately establishes an operating cost level prohibitive in most small libraries, public or private.
Second the baring of the user from the retrieveal task eliminates the entire browsing experience. Thus the casual connections of ideas and subject areas and all the intellectual value thereof, is totally eliminated from the value of the library.

Re-shelving is a need that the classification system must support but I believe this is supported by what ever system of organization is agreed upon. Re-shelving is no different than the negative of the use of the catalog.

So, I'm a 'patron' and for me 1, 3 and 5 all occur. I would suggest that they are not mutually exclusive categories. For example, often when I'm selecting a particular book from the catalogue, I then browse around it and pick out some more books in similar, related but not identical areas (after all, I'm interested in a particular topic at that time ... it's why I checked out the first book!). Sometimes 3 is the scenario but that doesn't mean it won't suddenly occur to me to search out something specific based on browsing-led ideas. Any time I'm in a library, 5 happens .... it's beyond me to stop it. So a shelf order classification system would have to satisfy all of those conditions not just one of them to be truly useful ... I'm sure I'm not alone in that.

As an aside, I use my University academic library regularly and then I often use the librarians to help me find specifics. What I'm referring to above is how I browse my public library because that's what I figure OSC is aimed at.

When I visit my public library and I'm not after a very specific subject/book I first visit the DVD/Video shelves, then the new fiction, new nonfiction, new YA fiction in that order. If I'm still looking I go on to browse the main fiction section and after that, if I have time, the nonfiction shelves for favorite categories. I rarely 'browse' the catalog unless I'm at home (online) looking for books, etc. to put on my 'Hold' list. Of course, if I'm looking for a particular book or subject I will consult the catalog.

How do others use their libraries as patrons?

How do others use their libraries as patrons?

I rarely go to the library just to browse; I usually have a specific topic in mind, if not an exact title. My library trip actually starts at home where I can search the catalog. This is impossible to do at the library itself because there is no dedicated catalog terminal, just computers being used for internet purposes.

Once at the library, I grab what books I've researched and check out.

Years ago, when I went to the library often I would find the 900s locate 935 then browse to the left and right. Then I would go to 635 and browse. I never had problems with Dewey, once I learned the basic classification I knew where to look. Now, when I went to a new library I would ask “Where is the history section” not “Where are the 900s” but the librarians understood what I needed.

>5 glendac:

Scenarios #1, 3 and 4 are about retrieval, not browsing. Therefore, it wouldn't matter how the books are organized on the shelf, as long as they have a call number. They could be shelved by size, by acquisition date, etc. Or, they could have a DDC number or an LC number or any number, really, becuase as long as that number is the same in the catalog as it is on the book, that book can theoretically be retrived.

Therefore, I don't think #1, 3, and 4 are really pertinent to what this project is trying to accomplish, which is about browseability. If you have a classification schema that allows for browsing, it will also function for retrieval--but not vice versa.

I'm a patron, and I sometimes do 1 or 2, more often 3 and 5. Also, like jjwilson in post #10, I usually look up DD "areas" that I then search for books on my subjects. Of course, by now I know where the subjects I'm most interested in are located, so I usually just think "Oh, I want a cookbook" and head over to the cookbook area without ever having to consult a catalog or map or know the DD number.

Now when it comes to fiction, I either walk along looking for authors I know or keep an eye out for the stickers my library sometimes uses to code major genres: romance, western, mystery, scifi/fantasy. It makes finding books I might enjoy easier. (Or I head to the non-fiction side which is where they keep the genre-specific anthologies. Never made sense to me.)

I'm trying to understand here the connection about the designs of our bibliographic records and databases and how that is displayed and browsed on physical shelves. How does a classification scheme as expressed through notations carry and display that data for finding and discovery? To what extent are the designs of our familiar sources and forms of bibliographic data (MaRC records and the OPAC) determine the browseability of information resources on the shelves? How much data can an item represent and display on its own on the shelves (with the help of signs, labels and notations)?

I guess we don't have to dwell too much on these kinds of discussions and we can go right ahead in listing top level categories and sub-levels and deciding the notations to represent these.

>10 jjwilson61:, 17

I was going to suggest this as #6. I use this technique frequently when people ask me for the X area, as it is often split up. 3, 5, and 6 are essentially guided browsing employing handouts, signage, or a human guide respectively.

>16 pivy:

I think you mean 1, 2, and 4 are strictly retrieval and are not really impacted by a classification system. However, the system's potential granularity can affect retrieval. A very coarse system is no better than rounding the accession number to the nearest hundreds place. Perhaps we should ask what size library is being served. A system that is great for a collection of 10,000 volumes may not work for New York Public.

I suggest we focus on these three considerations:

A. Clustering - Any system will break some subjects into multiple locations. Situations 3, 5, and 6 are best served by a system that makes these divisions easy for a browser to navigate. I think this is where patterns come in. We should make sure the scattering of a subject corresponds to how narrow the browsing interest is likely to be.

B. Granularity - This addresses situations 1, 2, and 4. It is easier for a local implementation to reduce granularity than to add it. But too much granularity makes the system unnecessarily complex. A system that supports "close enough" shelving could improve shelving time with a relatively low impact on retrieval time. If we reduce shelving time by 9 seconds but increase retrieval time by 3 seconds, we are still saving 6 seconds. Multiply that by 5,000 books processed per day, and we are talking 8 hours of saved time. This gets back to shelf order. How much room is there for imprecision?

C. Potential Collection Size - Patterns in the classification become more important for larger collections. Larger collections also need higher granularity.

My library is far smaller than most at LT, by a long shot. I tend not to save books unless they are of particular interest to me, and I planto reread from them at some future date. So, there is almost no fiction in my library...and those mostly short stories which can be read independently, one or two at a time...as a consequence of giving it away after reading.

I tend to clump books by main subject. Religion books grouped, atlases grouped, language books grouped, etc.Beyond that, I do not place them in either author order or by title, and I don't require numeric or alpha-numeric classification to find books in my house. It's the KISS system, or, as I prefer, the KIRSS system.