Localisation / translation workflow management

Background

I manage my company’s JIRA intranet, but at the same time I also develop our public facing “Global” (ie. non-Japanese) website.

At the moment the global website runs in 5 languages: English, French, German, Chinese and Korean, with all the content coming originally from our seperate Japanese site.  So we have a multilingual site and we have need to manage translation of all the contents. It cannot be managed manually for various reasons:

  • Volume is too high – 300 pages in 5 languages is 1500 translation sets.
  • Version management – Sometimes we have to restart translation because we find an error in the master.  We must then correct the master and resubmit it to all translators.
  • Time lag – Sometimes a translation gets stuck for months because of a sticky point in the text such as legal or cultural problem.

So I decided to set up a project on JIRA to manage the translations.

Statuses and Workflows

It basically works like this:

  1. Myself or a regional office requests a page to be translated from Japanese to their language
  2. Japan HQ approves/disapproves the request
  3. If approved, the Reporter assigns the issue to a suitable translator
  4. The translator submits it back to the reporter
  5. The reporter proof reads the translation then submits it to Japan HQ to create the webpage.
  6. I create the draft webpage (and Flash/images etc)
  7. Reporter proofs the draft and usually requests changes
  8. Repeat above two steps until satisified
  9. The page is then published live.
  10. We then enter ‘localization’ stage where subtasks are created automatically for the other languages.  They are auto-assigned to each regions default translator, then cancelled, re-prioritized or completed depending on each regions needs.

This translates into the following workflow diagram:

translation-workflow

It is important to note that this workflow only manages the translation of a Japanese page to one other language.  It does not track the status of the other languages – that is done by Sub-tasks of this parent issue.

Dashboards and Search Filters

To track the progress we require two search filters:

  1. Parent Issue statuses (Japanese -> Lang1)
  2. Subtask issue statuses (Lang1 -> Chinese/German/French/etc)

This is the Parent Issue filter:

web-translations

It is a standard filter but it has some nice extra columns that really help us:

  • Published field is a multi-checkbox field.  As we publish a page live on the website, we check the box for that language.  So we can see quickly that the first issue in the filter above is published live in English and French.
  • Global URL is a ‘URL’ type custom field that contains a direct link to the published English page on the website.  I dont bother to add the links for other langs on JIRA because the URL is almost identical for each lang.
  • Sub-tasks column lets me jump directly to the status of a specific language translation, e.g. German, for that webpage.  I think I had to add a plugin to show this column.

Here is the second filter showing the status of all subtasks:

trans-subtasks

To get the summary of the parent issue to display above, I used a the parent issue summary plugin from the JIRA plugin library.

Inside an Issue

I only need to attach the master translation file (we use excel files) to the parent issue.  The assignees of the subtasks know to go back up to the parent to get the master.  This makes it easy to update the master later if needed.

I mentioned that when the status of the parent issue moves to ‘Localization’, it automatically creates subtasks for each language. To do this, I used the create sub-task on transition plugin.  This is the result:

subtask-list

The subtask get assigned automatically (this is set up using the plugin), so there is not much more to do.  I just click ‘pass to translator’ and then the assignees get informed by email notification and later provide me the translation.  Sometimes I hold off on passing them the work, if I know they already have a big workload or that region has put all translation work on hold for a few months etc – in that case I just leave it as Open and it will not appear on their dashboards.

By the way these subtasks have their own workflow, though it is pretty similar to the workflow of the parent.

Conclusion

Well thats all I can think of right now.  I probably missed out a lot of other stuff here, please leave a comment if you have any questions about this and I will try to expand.

Advertisements

Synchronising JIRA custom field values with legacy systems

About 6 months ago, I was moving a legacy ticketing system to JIRA.  In addition to importing the old data, I also needed to synchronise not the JIRA issue data, but the actual options displayed in the drop down lists on the Create Issue screen.

I was planning to write my own plugin to do this, but just before I began I thought ‘lets have one more look in the user created JIRA plugin library first’.  By some amazing luck, a new plugin had been added in the last few months that did exactly what I required: Wim Deblauwe’s Database Values plugin.

searching-exampleThis plugin allows you to connect to an external database, fetch some values, and use them to populate the options in your custom fields.

We are using it in several ways, the most useful of which is to display our product model range in our JIRA.  When a user reports a problem with a product, then can select the product from a huge cascading drop down list (a 2 level drop down list containing Product Category/Model).  This prevents any ambiguity and reduces chance of user error.  It also allows us to standardise searches for specific models across multiple JIRA projects.

The list of models comes from a legacy system.  The added bonus of this is that when we later export the JIRA issue data to that legacy system, the drop down list values are the same so the data can be exported as is without any filter program in between.  The plugin retains not only the values (e.g. “Product XYZ”) but also any key associated with it (e.g. “018”).

I presume it would also be possible to connect to an MS Outlook address book in order to display customer names. This is something I may need to set up in future as we start to use JIRA for more CRM related activity.

You can also sort the values however you wish using an SQL query.  Another cool feature is that it can display the values using the JIRA AJAX textbox.  See Database Values for further information on this great plugin.

Monitoring JIRA for suspicious user activity

My company is in a very specialised industry with very few big players and very few new developments.  As as result, any activity to make new products/services is extremely valuable information.  My company is therefore understandably very concerned about data security.

In the past, confidential documents have found their way out of the company and into the hands of competitors, enabling them to use our new product ideas and try to get a similar product to market, or find some other way to steal our thunder when it is released.   Other times, sensitive customer information has been taken and distributed to the customer in an attempt to show that we do not properly protect sensitive customer information and therefore cannot be trusted/are incompetent etc.

So we needed a way to track who has accessed certain documents and discussions on JIRA, so that if such an incident occurs again, we can have some way to find out who did it, or at least narrow the options.

The things we wanted to achieve were:

  • be able to see who accessed Document-A between date X and Y.
  • have the system send an alarm email when a user downloads >50 binary files in 30 mins.

jira-user-filterThe solution we came up with was to use Sawmill which is a log analysis tool.  You install it on the web server and it parses log files regularly, updates it own database, and presents the information to you via a web interface.  You can set up various filters for the report information, and also set up email alarms such as above.  It was quite cheap and works well for our needs.   Probably there are other solutions, but this was the best I could find at the time.

I tried to use Sawmill for some other log analysis such as for my website, and for the SMTP mail server but it didnt work particularly well and I gave up on it without trying too much.  I use Google Analytics anyway for the website which is great, and I just look through the mail logs manually for now.