sharing serial prediction patterns

One of the core strengths of libraries is shared standards and sharing library data. Since we migrated to Evergreen in May I’ve been doing migration cleanup, implementing acquisitions and trying to figure out serials. Setting up serial prediction patterns is ugly in any ILS because prediction patterns are ugly.

There’s a great opportunity for the open source ILS world (both the Koha and Evergreen) communities to develop a standard so that libraries using these systems can save time and money by sharing serial prediction patterns. As more academic libraries are considering migrating to Evergreen, this would also help remove a barrier to selecting Evergreen. While it’s painful and annoying for me to manually set up all of our serial prediction patterns, I work in a small library, so it’s still possible. There’s only about 150. For a large university library it would not be possible set up a prediction pattern for each title.

Examples of serial prediction patterns

Can you guess what these prediction patterns describe?

  • Published Monday , Saturday, except for Christmas Day. Issues are identified by date. (daily newspaper in most cities)
  • Published weekly on Thursday, except for a double issue in the last two weeks of December. Issues are numbered continuously and four volumes are published annually, starting in Jan, Apr, Jul, Oct (The Economist)
  • Published twice monthly, except monthly in Jan, Jul, Aug, Dec. Issue numbers restart in each volume, which starts in Jan (Library Journal)

None of these are terribly complicated and yet they are still pretty messy. Thanks to David Fiander for letting me pinch these examples from his slides.

What’s a serial prediction pattern? Who cares?

Scholarly journals/magazines/periodicals/newspapers are published on different schedules. For example, some are published weekly, monthly, bimonthly, quarterly or yearly. There are also cataloguing codes for semiregularly, 3 times a year, biennial, triennial, and completely irregular.

In academic libraries it’s important to know if the library has a specific issue of a title, as users are most often looking for a specific article in a specific issue of a title. Generally, in public libraries this level of detail is not necessary. However, if libraries shared these prediction patterns perhaps more public libraries might use them.

Prediction patterns are also used to figure out which issues of a title should have arrived but haven’t. Libraries can then claim the missing issues with the vendor or directly with the publisher. (As an aside, I think journal claiming is a silly process that involves a lot of correspondence that doesn’t often end up in the issue being replaced. Some libraries are giving up on claiming for each issue.) Still, it’s important for both the user and the library to know which issues are missing in a run.

If serial prediction patterns interests you I highly recommend watching David’s webinar from 2009 on this topic.

What’s information is included in a serial prediction pattern?

There’s a bunch of information in a MFHD record, namely:


  • Hierarchy of enumeration, for example volume, issue, number, part (can have up to 6 levels in the hierarchy)
  • Does the numbering restart? If so, when?


  • How often does the title come? weekly? monthly? 4 times a year?
  • Are there exceptions to this pattern? If so, what are they?

Pattern (both publication and enumeration)

  • When is the journal published?
  • What publications will be omitted?
  • What issues will be combined?

Next steps

I’m not really sure what the next steps are. I think the open source ILS communities are best positioned to tackle this and figure out a standard way of sharing prediction patterns. We might want to talk to serials and cataloguing experts, like perhaps the folks at CONSER or NASIG. Perhaps it would be useful to talk to folks at OCLC or NISO. We might want to look outside the libraryland–what other industries are sharing information about odd, picky, sometimes irregular patterns? How are they doing things and what can we learn?

I’ll be presenting on this topic at the Evergreen conference next week and want to explore some next steps with people. I’ll be copresenting with Grace Dunbar and Mike Rylander from Equinox Software on Resource Sharing in Evergreen on Friday, April 27th from 3-4pm


7 thoughts on “sharing serial prediction patterns”

  1. “For a large university library it would not be possible set up a prediction pattern for each title.”

    I’ve a beef with that. For any Library, it’s possible to set up a prediction pattern for any given title. I can assert uncomfortably that the process probably becomes more onerous as the number of titles to predict rises. (This might *not* bear out in the unlikely event that a University subscribes to a large set of very regular periodicals.)

    That said, it might not be a particularly *good* pattern. With that in mind, the folks that best know the pattern are the publishers. The onus ought be on them to provide Libraries with data about their serials so that we can then set our ILS to comply.

    I shall not ramble on about my personal theories about collection development mating nicely with stocks.

  2. Tara — this topic has come up before as a possible area of standardization, but I don’t think there has been a groundswell of support in the past to carry the idea through the process. If that is starting to happen now I would be happy to make the connection with the right community leaders in NISO. (I’m a member of one of the three topic committees that guides the standardization efforts.)

  3. Thanks for the comment BWS. I think at a certain point it becomes not possible for a large library to do this, well…I guess it’s possible, it’s just not feasible or cost effective. Even if publishers shared this information there’s no standard that’s being used across most ILSes. So, each vendor or ILS community would make their own, likely hinky, way of dealing with this.

    12 years ago CONSER did a survey of 11 major ILSes and MFHD. One of the questions they asked was: Have users of your system successfully shared exported publication patterns with users of other serials systems in any of the following ways: Exchanged title-specific patterns with users of other ILS systems. All 11 of them said no.

  4. Peter–I met someone from NISO at code4lib and have her card but haven’t followed up yet. Do you recall the reasons that this didn’t get off the ground before? (Yanno, besides the fact that serials prediction patterns are hard, cumbersome and not at all sexy 🙂

  5. Yep. Gotta wonder what’s changed in a dozen years, though, eh? Here’s hoping summat has. *crosses fingers*

    As far as each Library creating their own very hinky things, I’ve often thought there should be a nicer way to crowd source summat into an authorised version. (Specifically local history shtuff which wouldn’t have entries in a “proper” repository.) Appropos of nearly naught here, but it’s a thought for how to eventually handle non mainstream serials that might not be widely held but are sought after by Patrons. Course, it all boils down to standards first, cool stuff with data and crowdsourcing latah.

  6. Nettie Lagace from NISO was at Code4Lib; she would be a great person to start with on this idea. As for why it hasn’t taken off, I think it is a case where a number of people think it is a good idea, but no one has stepped up yet with enough knowledge about the problem and energy to get the process started.

  7. Thanks! I found Nettie’s business card and got in touch. She pointed out that there’s been a big shift in moving from print to electronic serials, so for some librarians prediction patterns are becoming a moot point. She also mentioned as another group of people who might be interested and have expertise.

    I’ll find out next week if there’s interest in the Evergreen community to investigate this or take it on.

Leave a Reply

Your email address will not be published. Required fields are marked *