The Catalogue search uses the Microsoft SQL Full Text Indexing (FTI) facility. Its behaviour is not the same as a Google search.
We have tried to organise data and searching to minimise irrelevant hits (but by no means perfect! Due to the diversity of data included in the system).
The initial search page allows for the entry of a search phrase and the search results can be limited to a specific Record Type and/or Library.
The limiting options can be viewed in the drop down list.
For more sophisticated requirements the Advanced Search can be used, this allows up to four search terms to be entered with Boolean logic (and, or, and not). Further, the field type searched can be specific e.g. in a Title, a date range, ….
See also FAQs.
Access to certain documents is restricted, login to be able to view everything permitted. If insufficient permission to view a document it will show 'Private' (rather than 'View').
Different libraries use different categories and what they mean by a category can vary. When importing data from a library their Record Type is translated to one of our standard Record Types. For instance: • If a Record Type of CDROM is supplied it is translated to Miscellaneous (but the original RecordType is maintained against the record) CDROM is in fact a media type., if the CD contained a Journal it should be assigned as such, however, without detailed inspection of the data it is usually too difficult to assign. Some libraries make the distinction between Government/Post Office publications, Catalogues and Books, others do not. Therefore, the reformatting process often cannot easily identify the correct allocation of Record Types. However, some 'Data Cleaning' has been attempted, but this often relies on manual inspection if the provided data (and hence is very labour intensive).
The system contains data supplied by many libraries, to view the current list, choose the 'All Libraries' drop down to see those included. Each library works to its own data standards and even within a library there are inconsistencies in the way data has been recorded. Therefore, it is important to bear in mind that a search that pinpoints a hit for one library won't necessarily work for another. The information held is a snapshot and isn't completely up-to-date, further, it does not always hold the full information available in the libraries own database, this system is a finding aid which may point you to a library to obtain more information.
Most of the libraries have a physical presence, however, the Global Philatelic Library (GPL) is a virtual library and used for recording information from diverse sources, e.g. Journal cumulative indexes, supplied by individual societies. See also RPSL/GPL Auction Catalogues, later.
The phrase entered into the search box is sought explicitly in full, a wildcard * can be added if you want to put in a word stem, rather than a whole word, e.g. putting in 'monitor*' would bring back 'monitoring' but without the asterisk a hit would not be identified. Initially, we used a word based search, so if you put in say 'Great Britain' the search returned every instance of 'Great' and every instance of 'Britain' – so, too many erroneous hits and we changed to phrase based.
The system contains a list of synonyms that RPSL can supplement. They are of two types: Replacement; Expansion. Replacement are alternative phrases meaning the same thing, e.g. TPO; Travelling Post Offices; Train Post Offices. A search of TPO will also bring back hits on all the other phrases. Expansion is a one-way alternative, e.g. South Africa; Natal; Zululand, a search on South Africa would bring back hits on all alternatives but a search on Natal would not bring back hits on South Africa nor Zululand. If you have suggestions for new synonyms please submit via the Comments facility. Not all suggestions will be included as they can have unintentional consequences, providing an explosion of hits that are not relevant.
The Microsoft FTI facility uses word-breaking feature to identify segments of words to match against. This is particularly important for different languages, like German, where compound words exist. The FTI facility attempts to break the compound words up to improve search word matching. The first release of the application does this to a certain extent, but we have identified a refinement that suggests an improvement, We will investigate, and if useful, we will seek funding for inclusion in a later release. Meanwhile, the wildcard facility offers an alternate solution.
The Catalogue is built with a hierarchy, providing a drill-down facility. A good example to explain it is The London Philatelist journal. The Journal as a whole is a 'Grouping', each issue is an 'Item' and each article is a 'Part'. Against each issue is an 'Associated Document' that references the pdf file. Against each 'Part' is another 'Associated Document' that contains the text of the article. Involved Parties' (e.g. people or organisations) can also be attached to the main hierarchy, this can be used for recording, for example, Editors against the Grouping (and in the case of Books an author and publisher against the 'Item')
A search hit on an article would show above it the journal name and its issue. The researcher can choose what level of the hierarchy they want to view and navigate up or down. All the search hits within the same journal/issue are sequenced together. The lowest row in a band is the actual search hit. Examples are shown later.
In many cases a journal index has been supplied, rather than an article list. This index would also be stored as a 'Part'.
For Books there is often no 'Grouping' but a dummy entry is always created, 'No Series Title'. So, an 'Item' relates to a physical edition, for its 'Parts' we could have a contents list or indeed an index. An example where a real 'Grouping' could exist is the Melville series of booklets, published in the early 1900s, to be able to group relies on the provision of a series title, as well as the 'Item' title. The variety of data standards used makes this difficult and usually, Grouping is not attempted (some exceptions).
As well as the main hierarchy, there are adjoining pieces of information: • Notes – these can exist at any level in the hierarchy, e.g. a book review • Associated Documents – these can exist attached to 'Items' or 'Parts' and store information, such as the text of an article (used for searching) and a link to the pdf document. If the article text is a search hit, the 'Context' can be viewed. • Accessions – these can be attached to an 'Item' and record the existence of a copy in a library. If full accession information has not been supplied an entry may not have been created. • Involved Parties – these are people or organisations and can be attached to any level of the hierarchy, via a role link.
For Auction Catalogues, in theory, an Auction House would be a 'Grouping' and each Sale Catalogue and 'Item' but whereas the RPSL records information in this hierarchy, other libraries record individual sale catalogues and the auction house name is stored inconsistently or embedded within other information and it is difficult to ascertain the 'Grouping' consistently.
Events record things that have a start and end date. They can have links into the main hierarchy, for instance, a meeting could consist of several participants (each would been an 'Item') and each 'Item' could have Associated Documents (links to a handout, video, …)
For RPSL museum items they are all put in one Grouping to enable browsing of all items held. Indicivual items can then be viewed and sometimes parts of an object have been individually recorded in the Adlib database (which feeds this online Catalogue). Some images are available as associated documents.
Archival material is recorded at RPSL withing the Adlib database (as per Museum items). The Adlib hierachy is very deep and has been simplified for the Catalogue down to the three levels of Grouping, Item and Part.
Search results open in a new tab / window so that results for multiple searches can be retained for viewing.
At the top is the search criteria used and the number of hits found. Pagination always defaults to 15 hits, to optimise the time taken to render the screen, alternative options can be chosen (after initial results displayed).
A filter box exists which can be used to refine results, note this only filters on text visible on the pages (e.g. doesn’t find text buried in notes).
This is a hit on the Journal Title, Click Grouping' to view the details (and drill down to lower levels). Note the highlighting of the search words found.
This is a hit on: top band – a book author in an 'Item'; lower band – an article author in a 'Part'.
Clicking 'Grouping' in either takes you to the Group record (with subsequent drill-down). Clicking 'Item' would take you to journal issue (with subsequent drill up or down). Clickink 'Part' would take you to the article (with subsequent drill up or down).
This is a hit on text contained in an Associated Document
Clicking on 'Context' brings up a pop up window showing the text around the search word hit. Clicking on 'Part' Document brings up the article and from there the Associated Document (pdf) can be downloaded.
This is a hit on text contained in a Note attached to an 'Item'
Similar to Associated Documents.
The Advanced Search offers Boolean logic (and, or, and not) for entering up to four search terms. When mixing these operands the sequence is important. For example: a search on 'Walton' AND 'Frank' OR 'FL would initially find all items that contained both 'Walton' and 'Frank' but then it would look for any containing 'FL' and join the two sets of results together. A search of: 'Frank' OR 'FL' AND 'Walton' would first of all find all entries with either Frank or FL and then exclude from that list any results that did not also contain 'Walton'. A different result from the first case. For Advanced searches the data hierarchy is important. If the Grouping record contains, say, the word France and an index entry (Part) contains 'Paris' (but not France) a search on 'France' AND 'Paris' would be found because the contents of the Grouping record is implied to cascade down the hierarchy.
By default, all fields are searched, but the search can be narrowed to only look at a specific set of fields by choosing the appropriate one from the drop down list.
Data has been stored to various standards, for example, sometimes an author has been stored as 'Frank Walton' and sometimes as 'Walton, Frank' (and other variations). Therefore, a search on 'Frank Walton' would not find 'Walton, Frank'. However, the Advanced Search facility can be used to search on 'Walton' AND 'Frank' which would find both.
An Advanced Search can be specific to a date range e.g. 1900 to 1910. However, it needs to be borne in mind that in many cases, dates are not known/supplied (or are incorrect) and so this search is not reliable.
Content under development
The above shows information about the journal plus a list of editors (held as an associated InvolvedParty). Their names are held in a non-search field top avoid too many hits.
Any Notes & Associated Documents would appear here.
At the bottom of the screen are any subordinate 'Items', in this case issues of the LP (in descending date sequence). Clicking on the Title drills down to the 'Item' level.
To follow.