Results 1 to 7 of 7

Thread: Correcting missing categories, eg "Family Notices"

  1. #1

    Correcting missing categories, eg "Family Notices"

    I'm currently checking and correcting Family Notices in mid 1944 in the SA Advertiser and Mail. I've found quite a few days when the Family Notice category hasn't been set in the newspaper (it's under Advertising instead). This really slows down the review.

    Apparently, I can't create a missing category, only add tags. If I could, it would at least make it easier for others, and for myself searching later. (I'm aware of searching using tags, but it's klunky and still slow - it doesn't really solve the problem.)

    So what are my options here. Can I notify these categorisation errors? If so, how? Is there any way *I* can fix them?

    Andrew

  2. #2
    Staff Moderator mraadgev [NLA]'s Avatar
    Join Date
    Jul 2010
    Posts
    355
    Hi Andrew,

    It is not currently possible to change article categories in Trove, and there are no plans to add the ability for users to do this. We may add the ability for users to add additional categories, or a similar function, at some stage, but we cannot provide any information when this may be.

    When searching, Tags are always searched, so if you are tagging the articles with "family notices" then a search for the name with "family notices" as an additional search term will return both the tagged articles and the family notices articles.

    Mark
    Mark
    Trove Support
    National Library of Australia

  3. #3
    That's sad, Mark, that users need to cope with gross errors in the categories as well as the OCR errors. The OCR errors are at least partly inevitable, but category errors should be fixable.

    It's looking like any except the most casual and compacent searchers need to become query wizards. I can handle this myself I guess, as long as the rules are clear.

    So some query questions:

    Can I specify the source (titles) and categories in the query?

    How would I search on eg Jones and Gwenda in both (a) Family Notices category and (b) Advertising category tagged with "Family Notices"?

    Andrew

  4. #4
    Staff Moderator mraadgev [NLA]'s Avatar
    Join Date
    Jul 2010
    Posts
    355
    Hi Andrew,

    Although it would be nice for category errors to be fixable, it is not possible to redefine the article boundaries after they have been loaded into Trove, so in the case of Family Notices not being pulled out of the advertising articles, if the category were changed to Family Notices, then you would have the problem that the majority of the article would actually be advertising.

    In terms of the specifics of your query, you can use the Advanced Search page to define the titles & categories that you are searching.

    With the actual search you have specified, the easiest way to search for this would be to search for:
    gwenda jones "family notices"

    Adding "family notices" in quotation marks adds this as a phrase to be searched, and this will pick up all articles that have this as a title, all articles that have this as a tag, and all articles that have this in a comment. By default Trove does an AND search so will only return articles that contain all search terms entered.
    Mark
    Trove Support
    National Library of Australia

  5. #5
    Thanks for the tips, Mark, I'm getting there. Where the family notices are in the Advertising category, I'm tagging that as "Family Notices". This means that a search on both Family Notices and Advertising with search text as follows should work:

    gwenda jones "family notices"

    To lift my chances of spotting the names when scanning, I'm using something like:

    (gwenda OR jones OR marten OR births OR deaths) AND "Family Notices"

    This finds most family notice or advertising sections containing same. It also highlights the words I've selected which makes it easier to spot at least some of the ones I'm looking for.

    Andrew

  6. #6
    Prolific poster
    Join Date
    Feb 2011
    Posts
    228
    G'day Andrew,

    I think you will find your query will be much more effective as:
    (gwenda OR jones OR marten) AND "Family Notices"

    This is because with the [OR births OR deaths] part in there, the query will also find all articles that have [births AND "Family Notices"] as well as all those with [deaths AND "Family Notices"]. Which of course will be the great majority of family notices anyway, regardless of whether they contain "gwenda" or "jones" or "marten".

    That is, your query:
    (gwenda OR jones OR marten OR births OR deaths) AND "Family Notices"

    ..... is exactly the same as:
    (gwenda AND "Family Notices") OR (jones AND "Family Notices") OR (marten AND "Family Notices") OR (births AND "Family Notices") OR (deaths AND "Family Notices")


    Tricky stuff these queries ..... the secret is to think very logically, like a calculator !!

  7. #7
    Tricky stuff these queries ..... the secret is to think very logically, like a calculator !!
    Very true. In fact, my query does more or less what I want it to. I'm scanning all days and this is the fastest way I've found to move from one day to the next without too many false alarms. It opens in the right article (usually) with my search keywords underlined, which helps.

    It's a compromise between getting too many articles and the annoyance in having to directly find the Family Notices section for missing days, which is slow. Using this technique I can scan the births and deaths, and make the odd correction, at a rate of about 1 minute per issue.

    I've modified the above now to be (gwenda OR jones OR marten) AND (births deaths), which is a bit more reliable.

    Andrew

Posting Permissions

  • You may not post new threads
  • You may not post replies
  • You may not post attachments
  • You may not edit your posts
  •