Unit 8: Service Provider Review

The service providers I have selected are as follows:

1) METALIS: Service Provider for the Library and Information Science field. We harvest metadata from institutions that offer full-text papers and documents about Library and Information Science

2) OASIC : Search engine for french articles, memories and thesis in Communication and Information Sciences. It provides a single point of access to a significant number of Open Archives in these areas.

3) Scirus: Concentrates on scientific content only and by searching both web and (often proprietary) databases. Scirus' aim is to provide scientists with one comprehensive search platform covering both the web and the normally "invisible" databases. Scirus now harvests Open Archives data.

A good federated search I would offer has several characteristics. Some characteristics that come to mind are that a federated search should…

• Have a high level retrieval effectiveness of individual collections/resources. If each collection/resources does not have good metadata the results will also mediocre.
• Retrieving unique results across different resources
• The ability to be able to present the search results is a meaningful way to the user, either by categorizing the results that came from a specific collection, by a certain author and the like.
• Be sustainable as the number of collections being searched grows.

I do feel that the service providers were good overall examples of employing federates searches for the reasons discussed above and especially for the reason I address below in being a source of similar information and thus producing a better searh results page from a specified query.

Some sites may choose to index metadata from a large number of repositories and my outlook on this is that has both its advantages as it does its drawback. For example, if the College of Arts and Humanities is implementing a repository then it would serve a great purpose if the contents and collections were of relevancy and within a certain scope so that its community could make most of the information. An individual not seeking information in this area of expertise or interest would not find this repository of the highest value. In that sense, if a repository is made up of a large quantity of unlike and broad disciplines, then the usefulness to the focused user would diminish. Generally speaking, we categorize the like with the like and similarly my thought would agree with a repository. The amount repositories indexed is not a problem so much as that resource has a defined purpose. Have repositories of unique and unrelated meta data can slow things down and just present other challenges, such a process of elimination for search results for the user.