Power Gateway Report for 200406

This report analyses the MetaStar Gateway logs to provide the following summaries. See some explanations of terminology below each table.

• search terms used in the queries
• repositories that were searched by the queries
• responses from the repositories to the queries
• presentation requests asked of the repositories
• miscellaneous statistics
• any errors that occured in the gateways.

Warning: As with any statistics, you must take care with their interpretation.

Search Terms Used in This Gateway

Total number of searches conducted: 217

Note: Complex search queries can be composed, to yield accurate results. Multiple search terms are shown separated by delimiter, e.g. Title|Title|Abstract means 3 search terms combined with AND/OR

Multiple search terms were used 73 times

Singular search terms were used 144 times

• This table lists each different type of search that was used. It indicates how people are using the interface to build a complex search query.
• Multiple search terms are shown separated by delimiter, e.g. Title|Title|Abstract
• Recall that the interfaces provide multiple rows for text-based queries. In addition, the Power interface provides Spatial, Date, and Controlled terms capabilities.
• "spatial overlap" option is the default for spatial search, so it will be used more often than "spatial within"
• Full text ... Search anywhere in the document, i.e. all metadata fields
• ANZLIC search words ... Theme keywords

Repositories Searched by the Queries

• This table shows the total number of searches conducted at each node (including successful and failed searches).
• Repository ... A collection of documents that a certain custodian has published and made available for searching. Also known as a "node".

Responses Returned by the Repositories

• successful search ... The search succeeded and one or more metadata records were found to match. The value is the number of such successful searches.
• total hits ... The accumulation of all matches to all searches at this node. This statistic is skewed towards the larger repositories - more metadata records, means more possible hits.
• no results found ... The search did succeed, but no matches were found. The value indicates the number of searches that resulted in no matches.
• search failed ... The search failed for a node-specific reason. Possible causes include: server misconfiguration, out-of-date server software at this node, network failure during search.
• timed out ... The remote search server could not be contacted. Possible causes include: offline for maintenance, network congestion. The ASDD Help shows other reasons.
• connect failed ... The remote search server was contacted, but did not respond properly. Possible causes include: server software configuration error, software failure

Presentation Requests Asked of the Repositories

• Brief results ... Display the listing of titles returned from a search. The user has conducted a search which yields results. They have selected to see the results from a certain repository. That node sends a brief listing, showing hyperlinked titles.
• Full record ... Display a complete dataset description. From the brief listing from a certain node, the user has selected to see the full metadata record.
• You might think that this table does not correlate with other tables. However, remember that when the results come back from a search, the user may select to see the results from only one or two nodes, even though there may have been successful searches at other nodes as well.

Miscellaneous Statistics

ANZLIC search words used (top 10)

• The default term is AGRICULTURE, so this will sway the results.
• Terms beginning with a word in all uppercase characters indicates that they used the "Controlled Terms" option. Otherwise they must have used the text options which allow them to enter their own text to search against this field.

Number of repositories selected for simultaneous search

• Each search is conducted simultaneously against a certain number of repositories (as selected by the user).
• If the user forgot to specify a list of repositories to be searched, then the default is to search against all repositories.
• A high number of repositories probably indicates that they used the "select all" button.

Phrase search versus single-word search

• Phrase search ... more than one word occurring together

Other statistics

• ANZLIC Unique ID ... they searched for particular IDs using a Full-text option
• Right-truncation ... they used a wildcard * to match any word-ending (see ASDD Help)

Errors Detected by Gateway

• Unfortunately the log files do not record at which node the error occurred.