Science Gossip Talk

How to classify the unclassifiable?

  • jules by jules moderator

    The aim of Science Gossip is find illustrations and "to help identify, classify and correlate them." However, some of the newer journals (such as Hooker and Quekett) place all their illustrative plates at the end. The articles they illustrate are elsewhere in the journal far removed from the illustrations and nigh on impossible to find. Many, like this example, contain no title or species information so all we can do is mark it as a drawing along with any contributors. Is this still helpful?

    It should be possible to automate a search of the text for references to plate numbers and tie the two together so I wondered if it was worth noting plate numbers somewhere when an illustration comes up - to help any automated search? Often there is no plate key - just a mention somewhere in a page of text that is easily overlooked. Also, "Hooker's journal of botany and Kew Garden miscellany" unhelpfully refers to the plates as "Tabs" in the text.

    I think some guidance would be really useful here!

    Thanks in advance.

    @geoffrey.belknap @VVH @trosesandler @eatyourgreens

    Posted

  • geoffrey.belknap by geoffrey.belknap scientist

    Hi @Jules,

    Thanks so much for your question - and it is a very important one, as we are going to be coming across more and more journals where the illustrations are put on plates in the back, completely disassociated from the text which describes what is in the illustration.

    To answer the first part of your question - is it still helpful if all you can do is mark it as a drawing with contributors? The short answer to that is absolutely! Each of the tasks we are asking of a page are helpful in differing ways - but they are not exclusive to each other. So, if all we had was descriptions of where images were in a journal - and what kind of illustrations they were - that would allow me as a researcher to say something important about the degree to which journals are illustrated, and would allow BHL to make their search functions better.

    This is not to say, however, that knowing what the illustration depicts is not still very important - as it will allow me to describe in much closer detail what kinds of subject required illustration in journals - whether plants, animals, the weird/wonderful etc.... As well as allowing the BHL to create a much more targeted search function for historical taxonomy within their dataset.

    As for the question of can we automate the search for plate descriptions - I don't have the answer for that yet, but maybe this is the perfect place to start that discussion. I'll wait for @VVH @trosesandler @eatyourgreens to say what they think, but if we can find a way of finding the descriptions of the illustrations without downloading the whole book on BHL and doing a OCR search that would certainly make these kinds of journals a whole lot easier.

    What does everyone else think?

    Geoff

    Posted

  • trosesandler by trosesandler scientist

    Thank you @Jules for bringing this up. I agree with Geoff having at least identification that its an illustration and the contributors that would be more info that what we currently have. If the content of the illustration is unknown that's OK. In the case of this one I couldn't find any reference to the plate in the text but as you said if we could search the OCR then we might find it easier. Full text searching in BHL is currently not possible but we are working on adding that functionality. If you go to the original page in BHL you'll notice the illustrations have already been noted and plate numbers added. THis is because when BHL staff have time they will paginate books (but this is a very small portion of the BHL books ). So in this case the page info is already there as "Plate II" but it wouldn't hurt to add a keyword of "pl. II" as well when you classify.
    If you want to pass on this info in an FAQ that would be great!

    Posted

  • jules by jules moderator

    Thanks for the speedy replies @geoffrey.belknap and @trosesandler. It's good to know that even "incomplete" information is useful.

    If some people were up for hunting through the journal to find these illusive references to the plates would this be useful? And if so where would be best to note the page numbers? If this is done while the classification page is open we could add page numbers as keywords - otherwise they could be added as #tags in Talk. Which leads on to a question of how/if researchers use the hashtags. I ask because I've managed to locate a few plate references but only because I hate being beaten! I'm not sure how useful this would be generally - or if anyone other than me would even be happy to look for them. Just thought I'd ask.

    Yes - I'll definitely add this to the FAQ - I'll hang on in case there are any other comments.

    Posted

  • jules by jules moderator

    Just found another example (ASC0003i24) with the species added in handwriting. Someone's clearly already had this idea! 😃

    Posted

  • trosesandler by trosesandler scientist

    Jules,
    yes if someone wants to look through the journal to find more find more info by all means do! Page numbers would be best put in keywords. or even inscription if its handwrittten. The researchers aren't sure yet how we will be making use of the hashtags. Because the vocabulary is less controlled there its not as consistent as the classification tags but they do seem to be useful to many of the volunteers on the project.

    Posted

  • djshuman by djshuman

    When we come across a page like this, should we outline each individual illustration, or can we draw the rectangle around the entire page (all the drawings) and mark it as one species (if it is indeed all one species)? Or do you want us to mark each separately?

    Posted

  • JANEJ by JANEJ

    I usually try to find the article the plate refers to. If I do, I put the page number in the comments, not in the keywords. Which would you prefer?

    Posted

  • geoffrey.belknap by geoffrey.belknap scientist

    Hi @djshuman and @JANEJ - the more info that you can put in each image, the better. However, with plates like this that can be very tedious. So the solution that @JANEJ suggests is a good one - putting the page number in the comments is ideal - then we can pinpoint where the text is that relates to the image. If you put it in keywords it will likely be lost in the metadata.

    Thanks both for your hard work!

    Posted