Gimme! A Mobile Book Recommendation Engine

These are my notes about the Gimme App presentation by Aimee Fifarek in San Francisco on April 25th at IUG 2013.

2010 decision: “we need to be where our customers are …” – the palm of their hand. It had to be a web app so as not to exclude based on technology.
How do we know what our customers want and need? Market research (they hired out)! They not only surveyed existing patrons but they surveyed non-users – because we want to recruit!
Users want to renew books, see what’s available, see content, see what’s new, get book reviews, get book recommendations.
Aha! A what to read app. It had to be easy to use, fun, and read-alike categories: if you like this, then you might like this.
Gimme!

Usability and Design

Make it simple
Make it easy
Device Agnostic
Fun
They decided on a light and clean design.

Data

Where does the information come from? Syndetics and Goodreads
Staff reviews in Goodreads
RSS Feeds from the Catalog (Scottsdale has feed builder).
How do you keep it going? No API for the catalog so no good way to get stuff out of the catalog.
How to pick up specific titles from RSS feed builder? Catalogers created a series statement in 830; how do you know that only reviewed titles are picked up? The series statement like “Library Staff Review_Comedy” then that can be queried! You can use punctuation in Create Lists. Nobody has to know about the underscore because Encore doesn’t care about punctuation.
To create a feed: create a saved search and a simple configuration file (query=RSScomedy, max_items=50, item_title=:Vit|:Vbt, item_desc=:Vbn520|:Vrn520
Gimme has an administrative tool that allows for adding and deleted new feeds with a feed name and a feed URL.
Not enough reviews? Goodreads is readers advisory for the 21st century. This process needed to be formalized. Catalogers were brought into this process and for any new Marc records for which their were new reviews they encoded with the controlled vocabulary.
ISBN is key to this process. My note: this is how we would plan to grab both the review and the book cover from goodreads.

Marketing Plan

They bought tees, buttons, floor decals.
Create buzz but don’t give too much information so people would ask.
Easy peal-off stickers with a guerrilla campaign.
Life-sized recommendations (card board cutouts.
A training plan!!! Part of the grant money went to iPod touches to replicate the experience.

Measurement Plan

Does this increase circulation? Um.
They compared turnover rate of general collection to turnover rate of Gimme items – success!
Create new categories once every quarter.

Q-and-A

Total grant was 18,000 – only 6,000 went to development.
Catalogers pasting reviews into the MARC record in the event of a future API so they could ditch Goodreads.
Yep! Open source :)

Mobile Magic: Repurposing Kids Online

These are my notes for “Mobile Magic” presented by Peggy–damn she changed the introduction slide–on April 25th in San Francisco at IUG 2013. My notes are light.

Mobile Strategy

They realized they really knew they had to get in on this action so they originally purchased Boopsie. Problem solved? Well, nope. Users didn’t really like that Boopsie wasn’t cross-device compatible. Lacked features.
Ultimately, Boopsie was a separate site – users want the full site!
What about AirPac? Well, air pad didn’t appear to provide a very advanced search. More importantly, their display was in three columns!
Goals: create a light weight mobile catalog, device agnostic, brand however we want – etc.
They had this KidsCat Millennium module that was just collecting dust. Well, hey!
Beware: mobile devices detect any 10 digit string as a phone number – even ISBNs! You can use meta tag “format-detection” to disable that.
Users WANT advanced search capabilities.
Ask Innovative to copy SSL from port :80 to port :90. Set the SSLPORT=90:442
Reality check: not all features didn’t necessarily work – like spellcheck. However, it is real-time integration with the III server. Notice that RIM browser doesn’t allow for any browsing outside of port 80.

Preparing for Sierra (Panel)

These are my notes from the “Preparing for Sierra” panel presented at IUG 2013 in San Francisco on April 25th. Hopefully these panelists will be candid about migrating to Sierra and, afterward, living with it.

Nicole Lawson and Mary Dolan from Sonoma State University

Preparing for Sierra: moved to Sierra in Spring 2012 but didn’t aim for a to live date until that following January. Preparation boiled down to about 5 hours of work for tech services staff, not including training and documentation for library staff. A week after the server was installed the seeding process began; a week after that Sierra was ready for preview.
Downtime was just over 4 hours. There were a few firewall issues caught during innovative’s QA testing and it was related to an affiliate library.
Public Services Experience: certainly anxious. The prolonged preview period was definitely helpful. Apparently a change-averse staff member was generally pleased. He recognized that there really wasn’t much difference between the two systems, rather it just required a little extra reorientation.
The Tech Services Experience: “the easiest migration ever,” they were just assured that all the functionality from Millennium was still intact.
Issues: printout your micros and templates ahead of time because some will need to be created. All your create lists come over, but their create dates are preserved.
Any display issues (WebOPAC, etc.): not aware of any.
Any problems with create-list functionality? Prior to migration an efficient load profile no longer worked afterward.
Languages other than English? Sonoma does very little work with other languages, but the internationalization should be fine.

Donnell Perkins dr Brooklyn Public Library

Planning required executive buy-in, because closing the library may inspire patrons to contact their congress people.
To context user or not to context user? They eventually decided to use it, if only because it made staff feel more familiar.
Hardware survey: if your hardware is not correct it won’t work correctly. BPL paid more to have a training server as well as a production server. Worth it!
Everyone has to be informed throughout the process or the migration isn’t going to work.
Surprises: seeding database went undetected. BPL stayed on schedule (surprise!). Innovative was on-site – and that was a huge plus. It spoiled BPL. Having them there was a comfort.
Most major items were corrected; some weren’t. But none were show stoppers.
What could have been done better? Quality Assurance. Tight deadline. More focus on backend processing: circulation was lovely, but admin functions were sometimes flawed.
Don’t be afraid of the early-adopters. Many of those issues are no longer problematic.
BPL is on version 65 today.
SQL: still waiting … – but only so they don’t break shit (because SQL is powerful).
Examples of some open issues: various create lists; sysadmin alerts; printer settings not retained.

User Initiated Change with Google Analytics

These are my notes for “User Initiated Change with Google Analytics” presented by Robert Sebek, webmaster for the Virginia Tech Libraries, in San Francisco on April 25th at IUG2013

OPAC URLs are weird and inconsistent. Most people don’t search for the same thing, so pull out data tell more about user behaviors that we can actually use.
The best place to start is to familiarize with the main limiting parameters in the search URL: searchscope, sort, Da and Db, etc. it’s easier to do this using regular expressions.
Consider dates, mat types, language, and publisher using GA advanced segments. Match with RegEx to see how many times the conditions are matched. A result would look like: “the language condition has been used 123,4567 times.
Another way to drill down is to use GA’s filter option and include terms that appear in the URL.
You can use regex in an advanced segment to make sense of scopes in the URL to see whether users actually limit – spoiler: they don’t.
Robert shows that ever since VTL implemented a discovery layer to actual use of the OPAC plummeted.
so what is the user-initiated change? They stopped adding most online resources to the catalog, only to discovery service. And guess what? Zero complaints outside of library staff.

And that, kids, is why we don’t pretend libstaff are actually users.

WebBridge Power Tools

These are my notes for WebBridge Power Tools presented by WebOPAC wizard Bob Duncan on April 25th in San Francisco for IUG 2013. This presentation has a pretty niche application.

The presentation opened with a big overview of regex. Regex overviews are pretty damn useful, but can you imagine typing those out with your thumbs and managing to keep pace? At this stage I’m just padding my bullet point.
Regex resources: Regular expression pocket reference, O’Reilly
Secondary Lookups: secondary to what? OpenURL is primary, coverage data is secondary. The mechanics involves looking up the openurl data value in the coverage to get back the coverage data value. (Huh?)
In order to do a secondary lookup you must have match data to compare against. A couple of gotchas: how is data stored in coverage; how is field selector configured?
External lookups – as in external to what? Well, WebBridge. Why? Necessity with the onset of Google Scholar and PubMed as origins that didn’t necessarily include ISSNs. Also, if you have an origin that doesn’t offer all the metadata you want you can use an external lookup to find additional.
Common lookups: PMID, DOI, Metadata (CrossRef)
Any source that returns XML can be used as an external lookup.
Values sent to external lookup source which returns XML. These elements are mapped to OpenURL keys, so if you have “surname” you map it to OpenURL “aulast.” aulast
Enter quick-moving examples that are tricky to report. The mapping files live in “Live Web Server Configuration” or if you’re FTPing just don’t specify the /screens directory and it will land you in the configuration root. Then you configure the external lookup definition and apply to origin.
Troubleshooting external lookups: is the lookup actually working? Is the lookup successful? Is the resolution correct?
Matthew Riedsma has a user friendly alternative to stock link resolver layout. This is my note, because the screenshot from Lafayette College is all text.

Usability Testing with Eye-Tracking Software

These are my notes for Usability Testing with Eye-tracking Software presented by Bentley Library at IUG 2013 in San Francisco on April 24th

A Little Background

Primary focus was to test the library homepage and the different paths users take to find information.
Evaluate users and how well persona types are able to complete tasks.
Bentley’s Usability goals: are users making use of research guides? Is the new tabbed search interface effective? Are the users using faceted search option in Encore? Does the library website meet the needs of personas?
Bentleys Methodology: the think aloud protocol is a one-on-one session with a user where testers keep their mouths shut but the user thinks aloud through their process. Additionally they had an eye tracking tool (TOBII). They introduced a product reaction card, which is a bunch if different words on a card from which users can choose descriptors for what they think of the library’s website.
Protip: 5 seconds is a common measure to determine where users attention goes 5 seconds from start. Ideally, the task should be completed within 30 seconds.
Session length: 60 minute sessions consisting of a welcome and briefing, a pre-session questionnaire, the task, and room for post-session comments.

A little about personas.

Typically companies spend six figures on creating personas. Bentley did this for free with a really informal process based off anecdotes and data.
A typical persona includes a name, a stock photo, a summary of behavior (“Kyle doesn’t do extra credit.”) and a run down of that persona’s assumed information literacy.
Bentley created personas for two student users, a tenured faculty member – and a bunch of others (these were the three they began with; I expected a full rundown and now I’m caught off guard with an incomplete bullet point!)

Participant Profiles

Bentley had 296 sessions but then went back through that data and selected participants that seemed to closely compare to the created personas.

Findings and Insight

When users leave the library website to a resource or database they don’t understand of care that the library can’t control the appearance or usability of a vendor.
Users move quickly. They are not necessarily careful typists. Can you control for mistypes in searches?
Not all users read instructions. If they do, great – but most rely on intuition.
Even when databases are listed as links in the sidebar, many users gravitate directly to the search box and bypass even what seem like obvious menu items.
Users don’t know what “research guides” or “library guides” are. A few users actually tried to google it.
Placement of links is an art.
Regarding disruptions in the building, most users will not report an incident but just move away.
In the event of a report, users gravitated to the word “librarian” (as in “chat with a librarian.”)
Questions about a due date on a DVD spurned some users to a link that read “films on demand” because presumptively DVD loan periods would be associated with “films.”
Users are not really interested in finding information about due dates or fines, rather they rely on targeted emails that remind of impending due dates.
Help with citations: most users looked at quick links available. 70% sought citation tools through the “research” menu item. Several others would just ask a librarian about how to cite.
Without a discovery layer, users don’t naturally associate interlibrary loan with books they want but the library does not own. They do not disassociate the two different services. Getting to ILL was not easy when already in the catalog.
Students were more inclined to use facets to narrow down search results; faculty did not.
A single bad experience in discovery will persist and influence whether users try that same method for web unrelated tasks.
The search box was the number one choice for looking up information.
Navigation bar is a goto for predominately research-related tasks.

Interesting asides during the presentation

Disproven assumption that users know what to do with a search box with multiple tabs. Users much preferred one box to rule them all – a discovery layer.

Gained in Translation: Using Data Maps in Load Profiles

These are my notes for “Gained in Translation: Using Data Maps in Load Profiles” presented by Richard V. Jackson from Huntington Library on April 23, 2013 in San Francisco. This is probably pretty niche and the presenter was moving fast enough that I had trouble keeping track on my phone (how I post these). There were a lot of examples, so unfortunately these notes will be pretty brief.

Maps (Translation Tables) in Load Profiles

The load profile controls how an incoming MARC record is loaded in Millenium / Sierra.
The load profile says that for an incoming field that can’t be loaded as is, the translation table changes that data to be uploaded.
If you don’t have to use a map, don’t use one. I.e., if you need to delete certain sub fields in a field – you don’t need a map for that.
Interesting aside: there is a special function %replace that changes one literal string to another; this does not work with regex and you can’t include subfield delimiters. So since it’s not all that flexible its use is limited.
in the load table call map using %map to send an incoming field. It’s always starting with “m2bmap”
Inside the map there are global triggers beginning with @. Examples are hard to tap out (he is moving quickly). The real heart of the map comes afterward, on lines. The logic says if the incoming field matches the comparison expression then replace it with the value following the delimiter.
The comparison expression is regex. So if you’re looking for a literal period it needs to be escaped, whereas the replacement expression right of the delimiter is a literal expression.
Customize your delimiter with “@delimeter=…” but b god don’t use something that will appear in any comparison or replacement expression. Tildes, colons, pipes, etc….
Example for modifying 856s for ebooks: @delimeter=! @stop_on_map=true (.*)$0|z[^|]*(.*)$1!1 This outputs something like 8564 | uhttp://www.link.com/api.asp?bookid=1234
Why would you use a map instead of a tool like MARCEdit? Well, MARCEdit will always require some sort of human intervention. If you get the regex and map correct, then much of the work can be automated.
There were a lot of excellent emails on a handout uploaded to the conference site. Unfortunately, these aren’t available to non-III members.

Implementing RDA: How to Do It, Why to Do It

These are my notes from the IUG 2013 Preconference on April 23, 2013 in San Francisco. Note: I did not include the additional session about RDA Innovative Readiness by Ken Wells.

RDA Concepts, Glen Wiley from University of Miami

Builds on existing cataloging traditions in order to create a new content standard that can be used internationally.
it is a set of guidelines focused on user tasks as well as a product where cataloging tools are oriented for an online environment.
What will look different to us? Transcription rules, general materials designators replaced with content in the 330x field. Rule of 3 gone, meaning there is no limitation to how many contributors exist for one piece of work. Less use of punctuation, which is an effort to harvest more from content providers that are not necessarily adhering to present cataloging standards.
RDA won’t kill you, but it will affect systems staff, technical services, reference staff, admin, circulation and ILL (I probably didn’t have to write all that. Basically it affects the whole library). The point is that when planning for your RDA training you need to plan to train outside of in-house cataloging.
Your choices when to implement: new cataloging only; new original and copy cataloging; accept full-level records as they are; convert existing records to RDA; skip the whole thing …? Probably not.
Catalogers Training for RDA: FRBR (functional requirements for bibliographic records); RDA Toolkit; Practical application: examples and workflows; special sessions on focused materials.
You will do yourself and your staff an injustice if you don’t teach them to use the RDA Toolkit. They need to be self sufficient and enabled to find the answers to recurring questions.
FRBR: a conceptual model; not a set of rules; uses an entity relationship model rather than descriptive analysis without a structural model: it defines entities, attributes, and characteristics.
FRBR: the entity is the thing. It is a much higher level. Consider works like Shakespeare’s Hamlet where you have the work, the expression (i.e., translation), manifestation (i.e., imprint), and item.
Potential benefits: these conceptual models address user tasks. So the user is able to identify the length of the film, or whether to ascribe all the contributors to a film recording because that denotes a point of access.

The RDA Toolkit

http://access.rdatoolkit.com
You really should avoid print editions. Anything two years old or more is already out of date.
Look at the physical resource. You really have to explain to your staff the organization of RDA, I is broken into 10 sections with daunting titles, but here is the translated version:
1) and 2) Basic identification of the object.
3) The content: is it a book format, video format, etc.
4) How do you access the item for instance the URL to an electronic resource.
The other chapters are much more specific for instance certain musical, legal, or religious works. Chapters 5 and 6 really look at the date and origin of the work.
7and 8) Nature and scope of the content.
9, 10, and 11) Identification of the persons, corporate bodies, or authorities.
There are a lot more appendices. You can have your staff designate the relationship of a person to a particular work, whether the person is the printer or director.

Catalogers Training for RDA

Library of Congress RDA Training Materials
RDA NACO Training
Wiley’s staff always wanted to see more and more and more examples. The LC and PCC have created loads of example records.
Make a policy for what Catalogers need to so when they encounter an RDA record. How do you want the Catalogers to approach and edit the record? Make a checklist.
Local policy decisions that you need to make: RDA gives you man options and alternatives, but for example you want to figure out what staff do when they encounter a record that is in all Caps – because that is acceptable in RDA. Do you want to list all of the contributors, performers, and actors in a video and if so do you want your staff to spend time detailing that person’s relationship to the work. One of the basic things you do first is determine what core MARC fields you want staff to address. The national libraries have adopted a core set which you can choose to follow, but you should figure out what your library’s core set are.
Wiley recommends training for serials first and the your library’s special formats.

RDA Implementation with Lisa Robinson from MSU (woo!)

Don’t be surprised if you find RDA records already in your catalog.
Initial display issues: relationship entries were splitting the catalog in half but only for corporate entities.
Questions: should you expect RDA records as-is or edit them in copy cataloging or vendor-supplied records. What do you so with RDA/A2 records? Should RDA records be in a special workflow? How much training is needed and by whom? When do we start original RDA cataloging? Who will answer our RDA questions?
MSU formed an RDA implementation task group. The official charge was to plan for training and implantation, assess workflows, provide training, and provide RDA awareness to all library staff.
For those who will lead training, who trains the trainers? They must learn RDA (and learn how to troubleshoot).
What did MSU’s training look like? 45 hours over 3 weeks / 3 hours a day. Used LC online materials. Inuded RDA in NACO training (1 week). Looked at RDA record examples. Created practice records. Still holding Q&A sessions once a week.
Training challenges: everyone not equally receptive to RDA; RDA instructions changed after the training material was created
Lessons learned: should you train the copy Catalogers separately? Initial focus on FRBR was a great idea! Small group discussions were a great idea! Three weeks of training = stress.

Your Patrons Welcome a Classic Version

One of the hardest things for me to swallow when first I started working with the web for libraries was that patrons and the sometimes many, many stakeholders always begrudge change. There is an aspect of this work that demands sensitivity to the digital divide, an understanding that users by and large don’t often think about their browser – or care; do you care about the frame when you look through a window? Old browsers persist, and for a certain prolonged period because luddites pay taxes that pay wages, weblibs owe them the disservice of supporting deprecated, insecure software. It’s part of the job.

What I didn’t expect was how thick-skinned I’d need to be if, daresay, I switch-up the font stack. Something that in my head was very behind-the-scenes which insured consistency across browsers ruffled a surprising number of feathers. On April 12th when I–long announced and available to staff for months–re-themed the library’s site to adhere more closely to the university’s new scheme (and, taking the advantage, to upgrade the framework), people were downright bitchy – especially because in the transition I dropped support entirely for Internet Explorer 7. That’s another story. Cough.

Before the rollover I made a decision. A good one. I forked the entire predecessor website and dropped it into a directory called “classic.” In an announcement in the new header I gave users the option to use the “classic” version.

We are the 3%

I just looked at our stats and 3% of visitors in a little more than a week clicked “Classic Version.” So what’s the deal?

It may be different for you, but our users tend to be very task oriented. Browsing the library’s website isn’t necessarily all that functionally possible to begin with, but traffic tends be driven by assignments and extraneous research. Task-oriented, time-constrained users don’t want to reorient. They know where things are. Synthesizing the newness can make them lose track of their purpose. Library websites already tend toward cumbersome. We don’t need additional bad will.
Even if the old website was functionally less sound, repeating users have become accustomed to it. In a way, the new site they weren’t expecting is broken; the first response to seeing that the old stuff is still available is–all rushed together–”ohthankgod!” – the second: relief.
Suddenly I’ve started a dialogue with a userbase who is accustomed to a cluttered library website that just is and will be regardless of their preference. I had three–just three–users before clicking “Classic Version” leave feedback explaining why this version was preferred.
There were old features, such as a quick-links dropdown that was part of the university’s old searchbox, that had since been abandoned – so I gladly abandoned it too (less is more …); except a colleague told me today that “the classic version is the one place where [she] can still easily bounce from the website to blackboard or Sharklink.” Okay, wow. I didn’t expect to learn a feature was useful after no one had ever mentioned it ever.

Not that we should ever be okay with an unfinished product, but folks have deadlines. Thankfully, because the Classic Version has been there to pick up slack, I’m less burnt by the appearance of bugs in the new iteration. This is anecdotal, but the patrons and staff that discovered these were cordial and forgiving. The language of the announcement, the insight that some users might not yet be ready to change, made the library website human and subsequently more morally difficult to punch in the face.

Thanks for that, classic.

Proactively Addressing Color Blindness in Web Design

This last Friday (April 12, 2013) I flipped the switch that rolled-over much of the content of the Sherman’s website–an otherwise static site on a server with a moratorium on server-side scripting–to a WordPress network, then swapping the color palette to adhere to the new university color scheme – and honestly just give it some pizzazz. When I mention this I’ve been stressing that this isn’t so much a redesign as just a new wardrobe. Given the all-departments-have-a-say-in-content bottleneck at our non-prof institution, I’m taking the redesign in strategic steps: this, step one, didn’t muck with any of the content (cluttered and overloaded) or jargon, but I dropped the guy on a new mobile-first stylesheet with all new efficient, conditionally loaded scripts. So I count this as a win.

Over the weekend I got some troubling feedback through our contact form:

I cannot see the text in the upper portion of the page. I thought the site was down. This is not a very good design for anyone, especially those of us with visual impairments such as myself will have a very difficult time reading this page. You have a light colored background with white text?

“Of course,” I thought. “The beta of this site has been live for a month. It has made its rounds through a staff of 80+. The Library Student Advisory Board seemed to like the change. Only now is the coloring an issue.” But, admittedly, that is absolutely the wrong feeling to have. I was honestly a little aggravated that the new digs couldn’t live for more than half a day before the first negative comment rolled in. This is where I had to make the distinction: this isn’t a negative comment raining on my parade; this is a legitimate accessibility concern.

I turned immediately to my colleague @hfmabry who provided upfront assurance that, to her, the contrast looked fine – she even liked it alright. She linked me to the color contrast tools at accesskeys.com that explains

To conform with Web Content Accessibility Guidelines 1.0, foreground and background color combinations should provide sufficient contrast when viewed by someone with low vision or colour blindness, or when viewed on a black and white screen.

The formula suggested by the World Wide Web Consortium (W3C) to determine the brightness of a colour is:

((Red value X 299) + (Green value X 587) + (Blue value X 114)) / 1000

Two colours provide good colour visibility if the brightness difference is greater than 125 and the color difference is greater than 500.

where the library’s site fell within the recommended standards.

Fatefully, this morning I saw that @carlispina on her blog Novel Technology shared a link to another color blindness simulator that processes screenshots and supposedly outputs images that simulate the look of a site through three different types of color blindness. I want to share my results.

Here is a screenshot of the recolored front page of the Sherman:

Mind the quality of the following images. The simulator outputs pretty small, low res grabs. Our attention should go to the contrast. This, supposedly, is what the site looks like to someone with protanopia:

A screenshot of the Sherman's website as it might appear to someone with protanopia

This is what the site might look like to someone with deuteranopia:

A screenshot of the Sherman's website as it might look to someone with deuteranopia

This is what the site might look like to someone with tritanopia:

A screenshot of the Sherman's website as it might look to someone with Tritanopia

My Thoughts So Far

I am a little comforted by the simulator and the recommended WCAG standards, but only insomuch as they let me–a person with normal color vision–emulate color blindness – which honestly is probably not that accurate. It is the same feeling I have for giving low priority to functional issues that happen within Internet Explorer 8. I can make the case that it is a decreasing and dated portion of our userbase, but I still hate knowing that something I made looks like shit. The above feels inadequate, especially because I would properly feel like an asshole if I responded to this person’s comment with “Hey, we fall within the standards!” Web design isn’t about being an asshole, it’s about creating a neat and usable experience regardless of demographic. It is a democratically binding discipline and I struggle with the evidence that I am coming up short.