Monday, February 22, 2010

The Red Room: workflow photo tour

(part two in a short series)

In response to questions on the RFID_LIB list, I created a short photo tour of the red room, focussing on the staff side of things: the types of crate used, usability issues we encountered etc.

I've used the full range of Flickr metadata to describe the issues, unfortunately the slideshow doesn't show descriptions by default, and notes not at all. So best viewed as set: Flickr Red Room.

Alternatively, when watching the slideshow, in the options turn 'always show description' on, and watch it fullscreen (bottom right).

The red crates are made of sturdy plastic. When it became clear that custom crates were way too expensive, we settled for industry standard parts in standard sizes, and we adjusted our shelves accordingly. Same for silkscreening the numbers, so we used industrial strength plastic numbers, which turned out very well, in half a year I haven't even seen one beginning of peeling. The lesson learned: don't try to be special, and look outside the box, err, book world.

For staff determining when to add to an existing crate, and when to pick a new, we use these rules-of-a thumb:
  • The display shows a filling % of each existing crate and the # of items inside. This is enough for staff to figure out if there's still room. If not, new crate. If there is:
  • in peak periods, when the number of empty crates becomes small: always add.
  • otherwise, it depends on the day on which the items in the existing crates were added. If the same, we add; if in the past, pick a new.
This way, we have the flexibility to deal with peak periods with slightly more than 1000 boxes; and in less busy times, we can avoid crates with content from multiple days, which makes the workflow for processing of items not picked up more complicated, or forces us to leave the whole box until all items are expired, causing delays for other patrons.

Wednesday, February 17, 2010

The Red Room: self-service for a closed stack library

Recently, the Libraries of the University of Amsterdam (UvA, not to be confused with Virginia's UVa - yet another reason to avoid small caps for abbreviations!) and the Amsterdam Polytechnic (HvA) completed the introduction of RFID technology for security and selfservice. It was an interesting project in many of ways. And not just because it finished within budget!

European tendering was mandatory as the costs were well above the 200k€ limit. At first, I balked at this as a necessary bureaucratic evil. My personal opinion on this has completely reversed, however: with an unexpected outsider, Autocheck Systems, winning with a clear margin both in price and quality, this was a textbook case for the merit of the tendering process.

By clearly committing our demands to paper in a neutral way, prejudice is taken out of the equation, or at least reduced to a minor multiplier. The trick is writing good specifications.

Selfservice: for open and closed stacks

Public libraries have used RFID technology for over a decade now. This has created a mature market for open stacks. However, as an academic library, the vast majority of our circulation comes from closed stacks. Here, a different solution is needed, and when we embarked on this journey two years ago, turnkey products that are affordable for the amount of traffic did not exist.

We were hoping for a clever, high-tech solution, not limited to our own imagination. We wanted to tap the creativity of the vendors, bring on fresh ideas! But we most certainly also did not want to write a blank check.

The tender therefore was split up in lots. One for the mature technology, where the functional requirements were formulated clearly, and the scoring algorithm favoured price over extra features (to be precise, in a 7:3 ratio).

For the closed stack solution however, we described our situation, with detailed circulation figures. The nature of the solution - intelligent shelves, lockers, and so on - was left to the vendor. To judge functionality against cost, the vendor would have to supply a detailed description of number of staff still needed to run the closed stacks, and all the actions in their workflow.

Closed stack circulation: the old situation

In the old days, patrons would request materials in the online catalogue. The items would be picked up by the warehouse staff and brought to the backoffice to be checked and processed, and in piles on stacks behind the desk, sorter by patron name, accessible only by staff. A few hours or one day later, depending on the location of the items, the patron would come to the desk, and staff would retreive their material.

For this system, a large number of staff was needed. Not only because the patron was serviced, but also since in the absence of a proper tracking system, the piles had to be checked time and time again, to add new requests for patrons that had already more material waiting, to remove materials that had not been picked up, and to keep everything sorted on alphabet... There was clearly room for improvement. Self-service was only one aspect of the overall workflow.

However, there was one important restriction: privacy. A patron must only be able to borrow items that he or she requested, not items requested by others; and the name of the requesting patron must never be visible to others. In other words, the system must be fully anonymous. We've had run-ins in the past with professors that were spying on each others requested items...

To cut a long story short, we're very pleased with the end result of this project, for both the open and closed stack solutions. In the remainder of this post, I'll concentrate on the Red Room, the closed stack.

The red room

Autocheck Systems supplied the RFID technology and innovative workflow systems. The eye-catching design of the room is by Bureau Ira Koers and Roelof Mulder.

Red light district

Winner of the Great Indoors 2009 award - Trendhunter: showy red reading rooms - ArchDaily - Abitare (Italy) - ...

Some have called it the most beautiful circulation desk in the world... people love it or hate it, but it leaves hardly anyone untouched. Which is precisely what a library needs in these dark days, isn't it?

Patron's point of view
From the patron point of view, it works like this:
  1. Patron requests an item in the online catalogue;
  2. when the item is ready to be picked up in the Red Room, the patron receives an email with box number(s);
  3. the patron can also check the box numbers where requested items may lie by scanning the library card at an 'infostation' outside the Red Room;
  4. patron picks up the objects from the box, checks them out using selfservice machine inside the room and leaves (and when checking out is forgotten, the alarm at the entrance/exit of the room will go off).
The first weeks, we scheduled staff to stand by, but we stopped this when it became clear that patrons were just 'getting it' very well by themselves. Of course, staff is still available at the information desk nearby, which deals with various oddities that can come up, as well as patrons that need handhelding.

As for the design of the room that leaves nobody untouched, it fulfills its purpose well: it looks quite glamourous, which invites the patrons to treat it with care; and it does not invite to linger, which is good as patrons are supposed to get their items, check them out and then leave the room, either to the many study places in the building or outside.

On the whole, patrons have quickly adopted the new system, and reacted very positively. Requested items can now be picked up during opening hours, seven days a week, every day except Sunday til midnight. The original manned desk was open during office hours, two brief evening windows and Saturday morning.

To our surprise, what our patrons liked even more was the email service announcing the availability of the item. Because this email comes not from the ILS, but from the system that handles the boxes (more about that later), it is sent out the very moment the material is there. In the early days of the system, a glitch caused the mails to be sent out a few minutes early - which caused angry patrons at the information desk requesting why their box was empty...!

There is occasional dismay on the loss of face-to-face interaction. That was to be expected. There are however still plenty of opportunities for human interaction, both on-topic and off-topic. For the former, our information desk is doing brisk business. As for the latter... coffee can be had literally around the corner.

Also for special materials
The Red Room has security gates, forcing a patron to always check their materials with the selfservice machines inside. The security gates of the Red Room check the EAS bit, the regular gates at the main entrance check the AFI. This strategy with two separate zones enables us to also use the Red Room for materials that patrons can only use inside the building, but are not permitted to take elsewhere. For these materials, the selfservice machine leaves the security bit for the outer zone protected.

For this to work however, the ILS needs to send the object handling status to the machine. This is not a part of the SIP2 protocol, as implemented by Aleph! Luckily for us, for materials which may not be taken home, the return date in the SIP2 string is always set to today (it's implemented in the system as 1-day loan). By checking this date, we can work around this limitation.

As any professional knows, RFID cannot provide 100% security. The truly rare material - 100+ years old, collectors items and such - is therefore not handled by the Red Room, but sent to the Rare Books department, where the security measures borders on paranoia, and rightly so.

This also relieves us from nasty dilemmas, such as where on this Blaeu Atlas would I glue this tag... with an estimated lifespan of one or two decades?

Internal workflow
Behind the scenes, the system is, as per Einsteins quip, as simple as possible, but not simpler.

Tagging happens on per-need-basis - with 3+ mln items on closed stacks, it would be impractical and needlessly expensive to tag all items in advance.

Requested items come in from the stacks. If not yet tagged, staff add and program an RFID tag first. In rare occasions, they add a barcode if the item doesn't yet have it. Even though barcodes have been added to all requested materials since 1985, still relatively frequently items pass by that apparently have not been requested in 25 years... another reason to tag only when needed.

After tagging (as well as several other checks), incoming items are sorted by patron name. This is printed on the slip that starts the process inside the warehouse, no interaction with the ILS is needed.

Finally the items are paired with one or more boxes. This is an RFID-supported process that takes place at one of two specialized workstations with a touchscreen and large RFID antenna.
The staff member takes a pile of items and puts it on the RFID antenna of the station. The item ID's are read from the tags. From the item ID's, the user that has made the request is queried from the ILS (unfortunately this is not part of SIP2 or other standards, so a custom webservice needs to be set up. A big thank you to Leiden University for sharing their code!).
The station then first check whether all items on the pile are requested by one patron. Some names are common, and where people work, mistakes inevitably are made. If not, the staff needs to take away items until only items from one patron remain.
Then, the system checks whether this patron already has materials ready in the Red Room. In that case, the staff is presented with a list of boxes. Staff can choose to add items to an existing box, in which case a little slip is printed with book title and ID, and box number to add it to.
Finally, if the item(s) are to be put in a new box, the staff takes an empty box. The boxes are visibly numbered, but also have a tag, so the box only needs to be put on the RFID reader, and the link is made.
The box, filled with the items, is put on a trolley. The request slips are removed - important as the patron's name is printed on them.

After some time, or when it is full, it's driven to the Red Room and the boxes are put in place. On returning, the staff member sends off the patron emails.

In the old days, ~20% of the requested items were never picked up. Retreiving these overdue items was a labourious process that the new system has greatly simplified. At off-peak times, staff print a list of boxes that can be emptied.

Conclusion: happy with lo-tech
In practice, after some initial quirks, the system has been working remarkably well. It's fast, and does not get in the way; it is indeed simple. In the end, lo-tech proved the way to go.

Unfortunately, this summer also saw the migration from our old ILS to Aleph. This makes it hard to calculate the actual staff saving, since the entire workflow has changed in many ways. Current estimates are however that the business case is sound.

Our tender documents are available on request (and have already been used by one other institution), and I'll be happy to answer any further questions.