I spent quite a bit of this week working on the final part of my new interface mockups for the Dictionaries of the Scots Language: the search results. This took rather longer than I’d initially anticipated as there was much to consider and implement, such as the filter options, the navigation, the sparklines and how to handle results for both quick search and advanced search, and for both one dictionary and two.
I can’t really share the mockups at this stage, unfortunately, but I’ll describe how the new search results page functions. The SND and DOST search results (i.e. the ‘quick search’) work in the same way: the page begins with a search bar, allowing you to perform another search immediately from the results page. Next there is a ‘header’ section like the new ‘entry’ page containing information about your search: the term, the dictionary, the search type (e.g. Headword) and how the results are ordered, which I’ve added as text rather than as a tooltip. Then there is a row of options as buttons, as with the ‘entry’ page. As these searches are headword searches I’ve added in the ‘Filter results’ option. Pressing on this scrolls down a box containing the filter options, as with our still to be made live test site. This scrolls up/down when you press the button in the mock-up, but no filter options actually work.
Below the header is the main search results section, which currently has a similar layout to the current test site: a navigation bar above and below the results, which appear one per line, are numbered and have the first date of attestation and sparkline to the right.
For the ‘Advanced Search’ results the layout is mostly similar but differs in a few ways. The header now lists both dictionaries (assuming both have been selected in the search) and more complete information about the search options will be displayed. The option buttons are different for the advanced search, and would also need to include the ‘filter’ option for a different search type (it’s not relevant to the mockup search type). This might mean I’d need to rethink the layout of these buttons as they’re already the full width of the screen on narrow displays.
The advanced search results feature the snippets and highlighting, which are handled in a similar manner to the live site for now and for the results themselves I initially replicated the approach taken on the live site, whereby SND and DOST results are presented in two separate panels side-by-side. However, I’ve decided to go with a tab-based approach instead. As the majority of users access the site on mobile devices it is likely that the ‘side-by-side’ panels in my first attempt will never be seen by them anyway, but instead the DOST panel will appear below the SND one, making it unclear that the DOST results are even there. Instead I’ve added in a new tab section below the results header for ‘Modern Scots’ and ‘Older Scots’ with a count of the number of results for each included in the tab. You can easily press on the other tab to view the results for the other dictionary and there is much more room for the results. It does mean it’s no longer possible to view both SND and DOST results simultaneously and I don’t know how much of an issue this may be – but as I say, most users won’t see the results side-by-side anyway due to their narrow screens.
Also this week I packaged up the Bilingual Thesaurus code and a sample of the data and sent it to a researcher who will be repurposing it for a new project. It took a bit of time to reach an agreement about how much and which data could be included as the Bilingual Thesaurus people didn’t want too much of it to be shared, and I had to ensure that there was a decent spread of categories across multiple hierarchical levels otherwise the tree structure wouldn’t display properly. I got there in the end, though.
I also spent some time on issues relating to the new proposal the Books and Borrowing people are putting together. I can’t really go into too much detail about this and my potential role in the project, but it did take up a fair bit of my time. Also for the Books and Borrowing project this week my VPN access at Stirling was finally reinstated so I was able to complete the upload of the new borrowing register images for Orkney. This consisted of almost 200 page images, for which I also generated page records in the CMS.
I also continued to help Deven Parker prepare sample materials for her Playbills project that she will sent to the AI people in Computing Science ad then rounded off the week with some Anglo-Norman Dictionary duties. The editor Geert had contacted me to say that he’d spotted a part of speech for an entry wasn’t appearing in the ‘summary’ section of the entry page and wanted me to investigate.
It turned out that the issues was a <note> tag that was appearing immediately after the <pos> tag. The XSLT was expecting a <sense> or <subsense> element to immediately follow a <pos> and if this condition was not met the summary data was not extracted. I tweaked the XSLT so it can deal with <note> elements that appear between <pos> and <sense> or <subsense> now and this did the trick.
Finally, Delphine, the other editor of the AND contacted me with two further texts to add to the ‘textbase’ of source texts that we publish on the site. These texts had a slightly different XML structure to the previous new texts I’d uploaded so I needed to check how the integration process worked. All worked fine with the first text, although the <pb> element was missing ‘ed=”base”’ which caused the page breaks to fail until I fixed things.
The second text did however cause some problems. It was an updated version of an existing text so I needed to delete all data relating to the text that was already in the system, which meant I needed to write and test a script that would do this, deleting the text, all of its pages and all of the individual word records that were stored in the database for the concordance and proximity searches. I managed to create the script and delete all of the data, but when I processed the new XML file the notes were failing to load into the page. I realised this was because the page IDs contained characters that are not allowed in IDs, namely full stops, spaces and occasionally square brackets (e.g. n=”f. [211rb]”). Ihad to update these to remove the characters and I added a dash between the ‘f’ and the rest (e.g. n=”f-211rb”) and this has thankfully fixed the issue and all is now well.
I wil be on holiday for the next two weeks so there will be no further posts from me until August.
I was back at work this week after a lovely week’s holiday in Mexico. My biggest task of the week was to migrate the ‘map first’ place-names interface I’ve been working on for the Iona place-names project to the Ayr place-names resource. The version I’ve created for the Iona project has not yet launched and the Ayr project is nearing completion so it looks like this might be the project that gets to launch the new interface first, but we’ll see.
It took a bit of time to migrate the interface as the Iona project is slightly different to the previous place-names projects I’ve worked on (Ayr, Galloway Glens and Berwickshire). The resource has many bilingual fields to record Gaelic versions, it has a ‘flattened’ altitude range due to Iona being so close to the sea and it doesn’t feature parish boundaries as Iona is all in one parish. To get the resource to work with the Ayr data I had to address all of these issues, wit the parish boundaries being the trickiest to get working. In order to integrate these I needed to update the map’s ‘Display Options’ to add the option in and then update my code to incorporate a new display option. As this needed to be represented in the page URL I needed to shift all parts of the URL relating to the search, browse and viewing of records along by one, and my code then needed to ensure this shift was represented throughout. It was a bit of a pain to sort out but I got there.
I also made a new colour scheme for the Ayr map interface to differentiate it from the Iona one, but this is only temporary and we might change this before we go live. I also added in a logo for the map (an ‘A’ for ‘Ayr’ taken from the background map of the website) but again this might be changed to something else (e.g. a ‘C’ or the Coalfield Communities icon if this is allowed), or removed entirely before the resource launches. Below is a screenshot of the new map interface showing the results of a quick search for ‘Burn’ with place-name labels and parish boundaries (the orange lines) turned on using the OS1881 map:
I had two meetings this week, both on Tuesday. The first was with my line manager Marc Alexander to discuss the logistics of the new project the Books and Borrowing team are putting a proposal together for. I can’t say too much about this for the moment. The second was with Deven Parker to discuss her Playbills project with a couple of people from Computing Science who are researching AI. It was interesting to hear the possibilities that might be offered by AI in terms of extracting data form the Playbill images, although I think that we’ll need to discuss things with them in more detail if we are to ensure they fully understand the data and what we need to get out of it. More discussions will no doubt follow.
I had an email discussion with Sofia from the Iona project about further updates to the map interface and to explain how to successfully import the data exported from the resource into Excel in a way that would ensure accented characters are not garbled. I also gave some advice to Pauline Graham of the DSL about creating user accounts and helped Pauline MacKay of Scottish Literature with some issues she’d been having accessing the content management system for the Burns correspondence. This project also encountered an issue later in the week whereby the scripts weren’t executing but were instead downloading. This was very concerning and turned out to be a problem with our hosting company that we were thankfully able to fix.
Also this week I added feedback questionnaire popups, pages and menu items to the https://www.seeingspeech.ac.uk/speechstar/ and https://speechstar.ac.uk/ websites, made a few further tweaks to the text of each resource and gave some advice to Eleanor about accessing the Google Analytics stats for each site.
I was also asked to add a further library register to the Books and Borrowing resource. This consisted of around 190 images, which were supplied as PNG files. Unfortunately we need the images to be JPEGs to be consistent with all of our other images so I needed to figure out a way to convert the images. Batch converting images from PNG to JPEG seems like the sort of thing that should be straightforward to do using Photoshop or even just in Windows, but despite trying several methods I didn’t find anything that worked. Eventually I installed the ImageMagick command-line tool and used a single command:
magick mogrify -format jpg *.png
That converted all of the files in one go, as detailed here: https://imagemagick.org/script/mogrify.php. Unfortunately I then discovered that my access to the Stirling VPN had been blocked so I was unable to access the project’s server. Unblocking my access required authorisation from a few people and by the end of the week the process had still not been completed so as of yet I haven’t been able to complete this task.
Finally this week we received some feedback from the testing of the new Wales ‘Speak For Yersel’ resource, which required me to make many changes to the project’s data. This included replacing existing and adding new sound files, adding new answer options, updating questions and adding new ones. I also fixed a bug that had been caused by a difference in the way blank fields were stored in the database for new questions that I’d added a while back. These were classed as ‘empty’ for previously created data but ‘null’ for the new data (two different things in databases) and my code wasn’t dealing with the ‘null’ values properly. I made the fields blank instead, which has thankfully sorted things. Note that the data was all recorded successfully for survey responses – the issue was purely with their display on the maps. I also spotted that some map markers for the ‘Mate’ question were not displaying properly due to the sheer number of possible options being greater than my code was set up to work with. I added in some new marker colours and this addressed the issue.