Page 2 of 3 FirstFirst 123 LastLast
Results 11 to 20 of 26

Thread: Trove newspapers redevelopment - Discussion thread

  1. #11
    Staff Moderator mraadgev [NLA]'s Avatar
    Join Date
    Jul 2010
    Posts
    526
    Due to the way that the identifiers are structured I don't believe that this would be possible - essentially the article identifier is assigned to the full article when it is loaded into the content management system, and the part appended to the article URL to represent the second or subsequent page is the page identifier for that page.

    Having said this, however, we do plan to improve navigation between sections of multi-page articles, so as long as your citation identifies which page of an article is being referred to the changes should make it easier to move to the relevant section of the article.
    Mark
    Trove Support
    National Library of Australia

  2. #12
    Adult Member KerryRaymond's Avatar
    Join Date
    Oct 2010
    Location
    Brisbane, Qld
    Posts
    240
    Given that the number of comments is relatively small (from a computing perspective), is it possible to run a script over them and transform any of the URLs in this format into

    http://trove.nla.gov.au/ndp/del/article/36774395

    into the persistant URL:

    http://nla.gov.au/nla.news-article36774395

    It doesn't look like rocket science for the single-page articles.

    Kerry

  3. #13
    Staff Moderator mraadgev [NLA]'s Avatar
    Join Date
    Jul 2010
    Posts
    526
    Hi Kerry,

    That should be possible, and I'll forward it on as a suggestion for something to be done in future. It would not, however, be done as part of the current redevelopment.

    Mark
    Mark
    Trove Support
    National Library of Australia

  4. #14
    Staff Moderator mraadgev [NLA]'s Avatar
    Join Date
    Jul 2010
    Posts
    526

    UPDATE

    Although we’ve been a bit quiet over the last month, the newspapers redevelopment project has been moving along.

    As mentioned in the last update, the major piece of work involved in this redevelopment is an update to the underlying infrastructure of the service to increase the stability of the service.

    The IT team working on this project have been doing a significant amount of work in developing a prototype of the infrastructure that will be operating behind the scenes. The work that has been completed so far suggests that there will be improvements in the performance of Trove newspapers, in addition to providing a more streamlined browse path, which will reduce the load on Trove in comparison to the current browse paths.

    Alongside this development, the interface redesign has also undergone a number of iterations, and is reaching a point where the basic structure of the view article page, browse paths and homepage has been determined, with a lot of work still to be done on the exact content of these pages, and development of some of the other pages that are currently in Trove.

    Some of the changes that will be appearing with redevelopment are:
    1. Greater stability and improved scalability to accommodate growth
    2. Improved functionality on mobile devices
    3. Improved selection of States for facets and Advanced searching
    4. Updated and improved browse functionality



    There will be other changes in the redevelopment, and as the project advances and additional features are included we’ll let you know what is in and what is out.

    Technical details

    This information is focused on the underlying technology challenges the project is trying to tackle. There are more user focused changes to the UI that we will be covering elsewhere:


    Scalability
    • Our Solr search indexes are moving to a new version (4.3+), and we are adopting a clustered architecture (SolrCloud) that should prove more scalable as the collection size increases. Previously, Newspapers was making direct use of Lucene v2.9.3, and the rest of Trove is running Solr v3.5. This new pattern of deployment is also a model we will be interested in extending to the rest of Trove. Whilst performance improvements are not our goal from this project, we expect that the improved stability will at least assist in eliminating some the extremely poor performance that can be observed when our existing systems are under heavy load.
    • Beyond search, the Newspapers system also has a number of screens related to ‘discovery by Browse’, which have been a performance problem for us for some time. These screens can potentially aggregate a significant amount of data (and it is continually growing) from across the collection onto a single screen, and have been shown to cause significant stress on the database. This is far out of proportion to the tiny amount of actual traffic the screens receive relative to our other content. The new system will be taking steps to remedy this problem by rethinking what browse features are offered, how they are offered, as well as underlying implementation details (such as shifting some of the load from our database to the Solr indexes).

    Fault Tolerance
    • We are distributing our application components to make them more resistant to outages. For example, when the database is under heavy load or offline completely, we will of course need to gracefully degrade some features that depend on the database. But if the image service is online to serve image tiles, and the web server is online to provide you basic access to a viewer, then the task of viewing an article should still work.
    • We get a lot of traffic from ‘bots’ such as Google and Baidu indexing our content, in fact so much traffic that it dwarfs traffic from humans trying to use the site. This traffic is important, since we routinely observe that many of our customers are finding our content from these search engines and then paying us a visit to gain access. We are implementing mechanisms by which we can handle bot traffic more intelligently to minimize its impact on humans. For example, if we detect that our systems are under heavy load and struggling to keep up we can dynamically begin rate limiting (or block entirely) bot traffic to reduce most of the load. This prioritises service levels for the people that really matter, and we can gradually allow bots access to the system again as it recovers.



    Beyond these, we have our some of our own internal goals as well, such as improving the metrics/statistics we can get out of the system, which will help us with growth planning and health monitoring.
    Mark
    Trove Support
    National Library of Australia

  5. #15
    Staff Moderator mraadgev [NLA]'s Avatar
    Join Date
    Jul 2010
    Posts
    526

    Newspapers UI redevelopment

    Newspapers UI redevelopment

    The development team took into consideration the following information for the initial redesign for the Trove newspapers user interface (UI):
    • Extensive list of known issues with the current application
    • Continued consultation and user feedback
    • Requirements from other NLA delivery systems to create a consistent user experience



    This information was used to create a list of prioritised features which were was addressed in design workshops. The workshops allowed the team to explore various solutions quickly and share these with a wider audience. From the workshops a basic starting point for the revised Newspapers UI was agreed to be evaluated and refined with users through a HTML prototype. One particular feature that the development team was interested in testing was a revised interaction for discovering Newspaper content through browse. It appears that the revised browse feature we proposed made sense to most of the users who we have evaluated with so far and also provides the NLA with the ability to support more flexible browsing journeys for Newspapers users in future versions.
    Mark
    Trove Support
    National Library of Australia

  6. #16

    Suggestions for improvements...

    May I say how impressed I am with Trove, and being able to read old newspapers from the comfort of my own home. I am new to Trove, although certainly not new to reading old newspapers. I'm lucky to live in Canberra, and being able to visit the National Library, and I have done so for several years. However, as I'm getting older and petrol is getting more expensive and the traffic heavier, it is so much easier to read these from home. I'm also very happy to correct/edit items that I read, and have so do many times already.

    However, I have some suggestions for improvement, if I may (I hope this is the right place to suggest these). Could someone please advise me, if I am just missing something (and perhaps not yet knowing how best to use Trove).

    Firstly, I'd like an easier way to "turn" pages, as you read, rather than having to move the mouse up to those 2 small arrows "<" and ">" at the top of the page. Something similar to the way we turn pages on our e-Readers would be terrific, just clicking the mouse in a particular area.

    Secondly, I think the zoom control needs to be a bit finer ( [-] ! ! ! ! ! ! ! ! [+] ). For example, when reading the Canberra Times, the far left zoom mark makes the page too small to read, and tiny on the screen; however the next zoom level, +1, makes the paper too big to fit the screen. This results in the need to constantly move/scroll up and down in order to see the entire page. Ideally, another intermediate level/s of zoom would eliminate the need to do this. Changing my own screen size helps a bit, but isn't really helpful, as it makes those page-turning arrows even smaller!

    Finally, I'd like to see navigation improved slightly. When wishing to go back to a different edition of the same paper/magazine, for the same month/year (or even a different day/week/month) it's currently necessary - unless I'm missing something - to go all the way back to the Home page, and then start again (you can't just use the back arrow after reading 50-odd pages). Could there be an easier way?

    Anyway, just a couple of suggestions for improvement, but I am certainly very happy indeed to have this service at all. Many thanks again to those involved.

  7. #17
    Staff Moderator mraadgev [NLA]'s Avatar
    Join Date
    Jul 2010
    Posts
    526
    Hi Pembaca,

    Just a few notes on your suggestions:

    1) Improving navigation between pages is something that we hope to address in the redevelopment, although at this stage I can't give any specific details about this.

    2) The zoom levels are out of scope for the redevelopment project, as they require a lot more work. The levels between 3 and 7 have been selected to try and make sure that all different newspapers are readable, at the article level. How much appears on the screen will vary depending on the newspaper being looked at as well, and trying to create a one-size fits all zoom level is a bit difficult.

    3) Also being improved in the redevelopment. Currently, however, you can simply click on the 'about' link next to the newspaper title, then select the new date.

    Mark
    Mark
    Trove Support
    National Library of Australia

  8. #18
    Staff Moderator mraadgev [NLA]'s Avatar
    Join Date
    Jul 2010
    Posts
    526
    Yet again we've been quiet on the subject of the newspapers redevelopment. However work has been continuing, and we can now show you a sneak peak of what the new system will look like.


    Article view


    Article view showing annotations pane


    Browse panel


    Browse panel showing detail of list of pages of an issue


    Title information from browse panel.


    Much of the infrastructure work has been completed, and we're now working on making sure all of Trove's current features, as well as a few new features, are implemented and working as we expected.

    We do currently have a prototype in testing, and the screenshots show how the browse functionality will appear on your screens. Our next update will include a bit more information on some of the features we'll be enhancing in the new system.

    Please use this thread if you have and questions or comments about this update.
    Mark
    Trove Support
    National Library of Australia

  9. #19
    Member
    Join Date
    May 2013
    Location
    Melbourne
    Posts
    4
    I would like the ability to exclude (turn off) uncorrected hits when searching where the corrections remove the hit.

    For example if the original OCR (uncorrected) had Bunter but this had been corrected to Hunter I would like to be able to search for articles containing bunter but not include those that had been corrected to hunter.

  10. #20
    That sounds interesting. Hope that project can be launched soon.

Posting Permissions

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