This article lists the search indexes required by the PubMan solution, focusing the external-repository view as described in the use case PubMan Func Spec Search. Please note that additional indexes are needed to fulfill administrative tasks on the repository.

Initially, the content of this page was copied from the file Systemspecification_Pubman.doc (FE_PM_04 Indexing item data), but the requirements will only be maintained on this page. Please use the discussion page to comment on the existing or to submit new requirements.

Required Indexes[edit]

All released items have to be retrievable via the search engine. Lucene should always index the last released version of database should provide following indexes:

Any Field (escidoc.metadata)[edit]

All elements defined in the functional Metadata set specification are indexed for the search.
In addition, following attributes are indexed: <Id of item>, <PID of item>, <PID of file>
Files of any type are indexed, except of type "correspondence" and "copyright transfer agreement".

Special requirements for identifiers[edit]

In "any-field" index identifiers should be indexed as "identifier-type" and "identifier-value" as two separate tokens e.g.

ISSN
ISBN 232323184738
URI http://mpdl.mpg.de/123

Please note remarks on normalization

Example Queries:

escidoc.metadata=ISSN - to return all records which have an identifier of type ISSN
escidoc.metadata="ISBN 1234*" - to return all records which have an identifier of type ISBN starting with 1234
escidoc.metadata="URI http://mpdl.mpg.de" - to return all records which have an identifier of type URI with the exact value "http://mpdl.mpg.de"

Genre (escidoc.publication.type ??)[edit]

Publication.Genre

Persons (escidoc.any-persons)[edit]

all Creator.Person.CompleteName with Creator.CreatorType = "Person"

Organization by name (escidoc.organization-name)[edit]

all Organization.Name for each language separately and for all languages at once

Organization by identifier (escidoc. ??)[edit]

Organization.PID and PIDs of all Organizations in the organizational path hierarchy to the top level organizations (i.e. the PID of the authors affiliations and PIDs of all parent organization).

Example:

A PubItem "Title1" has authors:
- Udo Müller, affiliated to Max-Planck Institute for Psycholinguistics (PID2)
- Johanna Müllerin, affiliated to Department1 (PID4) of Max-Planck Institute for Plasma Physics

If the organizational unit structure is as follows:

MaxPlanckSociety (PID1)
 |---Max-Planck-Institute for Psycholinguistics (PID2)
 |---Max-Planck Institute for Plasma Physics (PID5)
      |__Department 1 (PID4)
PlasmaphysicsSociety (PID6)
 |---Max-Planck Institute for Plasma Physics (PID5)
      |__Department 1 (PID4)

Outcome: The index on Organization.PID should contain the following values for the PubItem "Title1":

   PID2, PID1, PID5, PID4, PID6

Even if the Author affiliations in the descriptive metadata are directly related only to PID2 and PID4

Please note: clarification

Title[edit]

Publication.Title and Publication.AlternativeTitle for each language separately and for all languages at once

Topic[edit]

Publication.Title, Publication.AlternativeTitle, Publication.TableOfContents, Publication.Abstract and Publication.Subject for each language separately and for all languages at once

Dates[edit]

Publication.Date

Event[edit]

all Event.Title,Event.AlternativeTitle and Event.Place for each language separately and for all languages at once

Identifier[edit]

ID and PID of item, PID of files

Special requirements for indexing identifiers as "any-identifier" index[edit]

identifier type and identifier value should be indexed together so that users are able to find itemA with identifier type ISSN and identifier value 123-234 in all of following "search by identifier" criteria:

ISSN:123-234 ISSN:12* ISSN:* ISSN 123-234

Source[edit]

all Source.Title and Source.AlternativeTitle for each language separately and for all languages at once

Components[edit]

We to index also some properties of the components in the search, such as:

content-category
visibility
component-name
pid (of component)

Maybe the best option to index all component properties by default.

PubMan Indexing

Contents