Scoping Study: Final Report Aggregations of Metadata about Images and Time-based Media

Cite <blockquote cite='http://mass.blogs.edina.ac.uk/2010/11/05/c-1-3federated-search-model/#1'>This model is not strictly a model of metadata aggregation, since the metadata remains in the collection owners’ databases. It is included in this appendix due to references from several interviewees, but is not directly related to an aggregation of metadata since a service provider could build a search service entirely without reference to the metadata aggregation.</blockquote> Permalink:
1 This model is not strictly a model of metadata aggregation, since the metadata remains in the collection owners’ databases. It is included in this appendix due to references from several interviewees, but is not directly related to an aggregation of metadata since a service provider could build a search service entirely without reference to the metadata aggregation.

Cite <blockquote cite='http://mass.blogs.edina.ac.uk/2010/11/05/c-1-3federated-search-model/#2'>This model is similar to the common schema aggregator model, in that there is a central service with a common schema which links to each of the source collections. The main difference is that the source collections provide metadata in real-time to the requesting service, using a protocol such as Z39:50 or SRW. An example is the initial version of the VSM Portal Demonstrator (this was replaced by the standardised schema aggregation model as it became clear all contributors were willing to support harvesting; some ruled out Z39:50 or SRW completely). This is described in the following diagram:</blockquote> Permalink:
2 This model is similar to the common schema aggregator model, in that there is a central service with a common schema which links to each of the source collections. The main difference is that the source collections provide metadata in real-time to the requesting service, using a protocol such as Z39:50 or SRW. An example is the initial version of the VSM Portal Demonstrator (this was replaced by the standardised schema aggregation model as it became clear all contributors were willing to support harvesting; some ruled out Z39:50 or SRW completely). This is described in the following diagram:

Cite <blockquote cite='http://mass.blogs.edina.ac.uk/2010/11/05/c-1-3federated-search-model/#3'> Figure C3: Federated Search Model</blockquote> Permalink:
3
Figure C3: Federated Search Model

Cite <blockquote cite='http://mass.blogs.edina.ac.uk/2010/11/05/c-1-3federated-search-model/#4'>The main advantages and disadvantages of this approach are summarised in the following table.</blockquote> Permalink:
4 The main advantages and disadvantages of this approach are summarised in the following table.

Cite <blockquote cite='http://mass.blogs.edina.ac.uk/2010/11/05/c-1-3federated-search-model/#5'> Advantages Disadvantages Metadata is always as up to date as possible, as the queries are real-time into the source metadata set. Collection owner remains responsible for maintaining the database. Collection owners only need to make schema and metadata available once, the search service will request updates as-needed. Search is optimised for the end-user. Limited model: focus on search service only limits opportunity for other metadata-based services. Search functions are controlled by the content-provider s so will vary between source databases. The speed and reliability of the network becomes an issue when connecting remotely. Higher traffic load on collections as every search will query their metadata. Metadata enrichment is possible at the collection level, and search only. Some metadata are forced into fields they are not natural fits for (if using DC). </blockquote> Permalink:
5

Advantages

Disadvantages

Metadata is always as up to date as possible, as the queries are real-time into the source metadata set. Collection owner remains responsible for maintaining the database. Collection owners only need to make schema and metadata available once, the search service will request updates as-needed. Search is optimised for the end-user.

Limited model: focus on search service only limits opportunity for other metadata-based services. Search functions are controlled by the content-provider s so will vary between source databases. The speed and reliability of the network becomes an issue when connecting remotely. Higher traffic load on collections as every search will query their metadata. Metadata enrichment is possible at the collection level, and search only. Some metadata are forced into fields they are not natural fits for (if using DC).

This entry was posted on Friday, November 5th, 2010 at 4:03 pm and is filed under Uncategorized. You can follow any responses to this entry through the RSS 2.0 feed. Both comments and pings are currently closed.