Libraries Australia members can take advantage of Trove’s powerful search features to better expose collection lists, finding aids and other uncatalogued or briefly catalogued material.
By following the instructions below, you can let Trove’s search engines know about documents on your website in text, unencrypted PDF, XML (including EAD) or HTML format. When Trove knows there is a searchable document available, it will index all the text on the page. When someone in Trove searches on a term only found in the document, it will still return the record for the document, and a link to the text.
Example: Summer is researching former WA premier John Tonkin’s life. She searches in Trove for eulogy John Tonkin. The first result returned in the Diaries, Letters and Archives zone of Trove is a link to the papers of Governor Sir Francis Burt, held at the State Library of WA. While there is no reference to John Tonkin in the catalogue record, Summer clicks through to the finding aid, discovers that Sir Francis gave the eulogy at the premier’s funeral, and a copy is available at the State Library.
Instructions:
- Choose some material you think would be useful for the public to be able to search. This could be finding aids, summaries, transcripts, or lists. This material must:
- Be publicly available online at a persistent URL
- Be in one of the following formats: XML (including EAD in XML format), txt, unencrypted PDF, or HTML.
(PDF’s must contain text, not a picture of text. To check, try searching for a word in the PDF using Adobe Reader. If the word is found, the PDF is ok to use. If the word is not found, Trove will not be able to find the word either. If you can’t search the PDF, it may be possible to convert it to a searchable type.)
(Only the first page of a HTML website will be indexed. Links to other pages won’t be followed. All words on the page will be indexed, including navigation and menus, so it’s best to have a page that doesn’t include a lot of extraneous information.) - Be less than 3 Mb
- Have a relevant catalogue record in (or able to be added to)Libraries Australia.
- Add the trigger to the MARC record in Libraries Australia: In the 856 for the link that Trove should follow, add $xsearchable resource. The trigger can be added to more than one 856 in the same MARC record. If you remove this trigger at any time, the list will stop being searchable in Trove. (If you are using the Record Import Service, let Libraries Australia know first, as they will need to ensure that the 856 is added correctly.)
Example:856 42 $uhttp://www.slwa.wa.gov.au/pdf/mn/mn2001_2500/mn2023.pdf
$zView the list of items in this collection$xsearchable resource - When the record has been updated in Libraries Australia, it should only take a few moments to appear in Trove. (If there is a lot of activity on Libraries Australia at the time, it may take longer). Search Trove for a word or phrase in the list, which is not in the record metadata. If there are too many results, you can also add the Libraries Australia number to the search terms to help your testing.
- If you update the linked documents, you will need to re-upload the records to Libraries Australia, to ensure that Trove re-indexes correctly.
Notes to consider:
- Trove will index the whole page at the URL link, not just the main “frame” of the web-page. The indexing will include any navigation terms on the webpage and any terms used to describe outside links. This can cause false hits, which are minimized by providing an indexing friendly webpage.
- If using html or xml, data needs to be in a format where the Trove indexing service can take most advantage of it. i.e. the data can’t be inside the tag. Formatted like<author>J.Smith</author> Trove would index ‘J. Smith’. However, if the data was formatted as <author name=”J.Smith/> then J.Smith would not be indexed.
- On Trove, the document text will only be indexed in keyword format. It won’t be available within command or advanced search; and the data won’t be findable through the facets on the left-hand side of the page. If users are limiting their search by facet, the results won’t appear unless the collection level record is included in that facet. (e.g. a collection that includes many maps, but where the collection level record is not coded as a map, will not appear in a search restricted to maps.)
- Trove’s relevance ranking considers terms indexed from a searchable resource link as less important than terms found in the catalogue record. If a search returns many results, it is not likely that matched records from searchable resources will appear on the first page.
If you have any further questions about using this service, the Help Desk would love to help you out!
