Cambridge University Library

How does discovery fit into the overall library systems architecture? A joint local and web scale approach

Methods Used
This case study makes use of the following, which are described in more detail in the linked resources:

MethodUsed?Also relevant
A - What? - Function PaletteYes
B - What? – Shared Services Spectrum
C - What – Profiles, Scenarios & User StoriesPossibly
D - Who? - Business Ownership Map
E - Where? - Service Location Map
F - How? – Requirement v. Resource Matrix
G - Why? – Business Benefits Ranking
H - Difficulty - Dependencies MatrixDocumented
I - Difficulty – Potential Risk RegisterYes


Cambridge University Library

  • Ed Chamberlain, Systems Development Librarian
  • John Norman, Director of Digital Services
  • Lesley Gray Administrator & Journal Coordination Scheme Manager

Project team

  • Owen Stephens, Owen Stephens Consulting
  • Ken Chad, Ken Chad Consulting

1. The Problem

A major selling point of the university is the high value of its collections—the library is trying to turn this into a service USP. That suggests a library centric discovery (as opposed to Google etc) approach should be part of that value proposition

2. Findings

a) Complexity of discovery landscape

Cambridge is an organisation with a plethora of niche/expert needs that library discovery products like Summon (from Serial Solution) do not fully address –at least at the moment.

The library recognises that it is a player in a wider ‘discovery’ environment that includes:

  • Pedagogical recommendations (inc informal ones and more formal reading lists)
  • Open web discovery services-typically Google (Scholar). The library is thinking how it can add ‘library’ value to these ‘generic’ services. For example this might be by embedding its services like authorisation, and authentication or ‘context’ (e.g. course, year,) Maybe there are now more (open) web centric ways to do this thing better than library centric approaches such as OpenURL. There may be potential for example in the work going on in libraries around
  • Web based niche services (e.g. subject services- e.g.  A&I like PubMed etc). These will still be used directly by STM researchers for example. For that reason the library provides citation databases even where they will show what library does not have. Seeing ‘all’ material is important to researchers—they don’t want to be limited to library holdings/licensed material.

At the A&I (e.g. PubMed) and Open web (e.g. and Google Scholar) level of discovery there is still a real problem in making links to appropriate (library) holdings/licensed content. Google scholar automatically harvests data whereas PubMed requires the library to initiate a process which is typically viewed by the library as too much of an overhead to do. More and more services seem to be enabling link resolution to direct the user to library resources. For example JSTOR is taking steps to be more generic service that will pick up the user ID (via IP or login) and will direct back to Cambridge university library resources

b) Diversity of systems and dependencies

Local Discovery -Aquabrowser

Aquabrowser harvests metadata from the Voyager Library Management System (LMS) and Institutional Repository (IR). However Archives are not included in the Voyager/Aquabrowser mix.

Aquabrowser does a live ‘scanback ‘to Voyager to get holdings/availability information. Voyager in Cambridge University is a complex installation being made up of eight separate databases

Web scale discovery

Web scale discovery (Summon) works on top of Aquabrowser. It ingests everything in Aquabrowser and in addition the associated Knowledge Base (KB) gets metadata (electronic material) from publishers etc

The discovery service themselves are dependent of a number other components

    • Resolvers/KB/ERM (all packaged with the vendor’s Discovery system)
    • Proxy server
    • Shibboleth
    • LMS- Physical material exposed to discovery service via LMS APIs

c) Value of a library centric approach

There is still some perceived value in local indexes. For example Summon is journal article based and an index of journal title also has value.

Having local control have control of indexes and the user interface (UI) at Voyager/Aquabrowser level can enable the library to meet the needs of niche searchers. For example improved music search where, for example a user can enter ‘Chopin’ for and get facets -concertos, piano music. Cambridge has enough music scholars to make this approach viable

3. Methods Used

a) Dependencies Matrix

(3=strong dependency > 0= no dependency)

CUL Dependencies Matrix


Web search and global index have the same dependencies although dependency is weaker for web search. Maybe Google et al could do better? So could Google be a better Summon?  ‘Publishers will feed Google for their e-commerce model and summon to keep their librarian customers happy’ (Ed Chamberlain)

4. Observations on methods

Cambridge University Library participants were equivocal about the insights gained from the investigation. However whilst is was limited in drawing out new insights for the university library the exposition of the issues in a very complex library systems environment, as illustrated in the Dependencies Matrix may have value for university libraries.