Author Archives: admin

Parsing, Recursion and Observer Pattern

I have worked for a while now with two serializations of the XLIFF Object Model: XLIFF and JLIFF (which is still in draft). I have had occasion to write out each as the result of parsing some proprietary content format in order to facilitate easy interoperability within our tool chain, and round-tripping one serialization with the other.

Whilst both are hierarchical formats when parsing them recursively they require different strategies.

With XLIFF (XML) each opening element has all of its attributes available immediately. This means you can construct an object graph as you go: instantiate the object, set all of its attributes and make any decisions based on them, and add the object to a stack so that you can keep track of where you are in the object model. This all works nicely with the Observer pattern: you can subscribe to events which fire upon each new element no matter how nested.

<xliff xmlns="urn:oasis:names:tc:xliff:document:2.0" version="" srcLang="">
<file id="f1">
<group id="g1" type="cms:field">
<unit id="u1" type="cms:slug">
<segment id="s1">
<source>/>
<target/>
</segment>
</unit>
</group>
</file>
</xliff>

With JLIFF (json) you have to wait (assuming you’re doing a depth-first token read) to read all of the properties of nested objects until you can access all the properties of the parents. Thus you have to build an object graph before you can then traverse it again and use the Observer pattern in an efficient way to build another representation.

{
"jliff": "2.1",
"srcLang": "en-US",
"trgLang": "fr-FR",
"files": [
{
"id": "f1",
"kind": "file",
"subfiles": [
{
"canResegment": "no",
"id": "u2",
"kind": "unit",
"locQualityIssues": {
"items": []
},
"notes": [],
"subunits": [
{
"canResegment": "no",
"id": "s2",
"kind": "segment",
"source": [],
"target": []
}
]
},
]
}
]
}

Differences are also apparent when dealing with items which require nesting to convey their semantics. This classically happens in localization with trying to represent rich text (text with formatting).

XLIFF handles this nicely when serialized.

<source>Vacation homes in <sc id="fmt1" disp="underline" type="fmt" subType="xlf:u" dataRef=""/>Orlando<ec dataRef=""/>

Whilst JLIFF is somewhat fragmented.

"source": [
{
"text": "Vacation homes in "
},
{
"id": "mrk1",
"kind": "sm",
"type": "term"
},
{
"text": "Orlando"
},
{
"kind": "em",
"startRef": {
"token": "mrk1"
}
}
]

Content Interoperability

I am working on a project which is very familiar in the localization industry: moving content from the Content Management System (CMS) in which it is authored to a Translation Management System (TMS) in which it will be localized and then moved back to the CMS for publication.

These seemingly straight-forward scenarios often require far more effort than seems worthy. As the developer working on the interoperability you often have to have:

  • Knowledge of the CMS API and content model. (The content model being the representation which the article has inside the CMS and when exported.
  • Knowledge of the TMS API and the content formats that it is capable of parsing/filtering.

In this project the CMS is built on top of a “document database” and stores and exports content in JSON format.

One of the complexities is that rich text (text which includes formatting such as text emphasis – bold, italic – and embedded metadata such as hyperlinks and pointers to images) cause sentences to become fragmented when exported.

For example. the text:

“For more information refer to our User’s Guide or Community Forum.”

Becomes:

{
"content": [
{
"value": "For more information refer to our ",
"nodeType": "text"
},
{
"data": { "uri": "https://ficticious.com/help" },
"content": [{
"value": "User's Guide",
"nodeType": "text"
}],
"nodeType": "hyperlink"
},
{
"value": " or Community Forum.",
"nodeType": "text"
}],
"nodeType": "document"
}

If I simply let the TMS parse the JSON I know it will present the rich text sentence as three segments rather than one and it will be frustrating for translators to relocate the hyperlink within the overall sentence. Ironically, JLIFF suffers from the same problem.

What I need is a structured format that has the flexibility to enable me to express the sentence as a single string but also have the high fidelity to convert back without information loss. Luckily the industry has the XML Localization Interchange File Format (XLIFF).

I have three choices for programming the conversion, all of which are open source:

I wanted to exercise my own code a bit so I went with the third option.

JliffGraphTools contains a Jliff builder class and Xliff12 and xliff20 filter classes (for XLIFF 1.2 and 2.0 respectively). These event based classes allow a publish/subscribe interaction where elements in the XLIFF cause subscribing methods in the Jliff builder to be executed and thus create a JliffDocument.

I decided to use this pattern for the conversion of the above CMS’ JSON content model to XLIFF.

Slide1

It turns out that this approach wasn’t as straight-forward as anticipated but I’ll have to document that in another post.

Counting Sheep

I am affected by insomnia on a fairly frequent basis. I don’t use any sleep related gadgets or applications because I’d probably scare the crap out of myself. Let’s see if this post has any cathartic properties.

Tonight’s/this morning’s musings (in no particular order of appearance): graph based machine learning, building an object model library in Typescript, my next career move, health impact of not sleeping well, industry integration/fragmentation dichotomy, current development projects, Brexit, personal relationships, and quality estimation.

It’s going to be tough getting through tomorrow. Sweet dreams.

Exploring JLIFF

I have published a web application where you can submit XLIFF 2.x files and get back a JLIFF serialization as text.

JLIFF is a new XLIFF serialization format currently in working draft with the OASIS Object Model and Other Serializations Technical Committee.

The application uses my open source JliffGraphTools library.

I am working on a conversion of XLIFF 1.2 to JLIFF but as the content model is structurally different it’s tricky.

I was careful to implement it in a way that means no data is persisted. I don’t even collect any metadata about what is submitted. That way people can feel confident of about the privacy of their data.

The Devil is in the Detail

I really enjoyed attending Unicode 41 this week. Following changes to my role some years back and the fact that the conference is always held on the West Coast of the US, I hadn’t been in a while but I will definitely put it back on my conference agenda. It was great bumping into customers and old friends and seeing the new generation of researchers and engineers address what is essentially the challenge of worldwide communications.

The conference kicked off with a very interesting, entertaining and thought-provoking keynote entitled “Can We Escape Alphabetic Order”, given by Thomas S. Mullaney.

The remainder of the conference sessions I attended covered: predictive models used by Google in their Android keyboards; dynamic translation resource bundles developed by Uber for their mobile apps; enhancements to ICU (International Components for Unicode); Nextflix’s approaches to bi-directional and vertical subtitles and captions; Javascript libraries for internationalization; support for Emoji in Unicode; and NLP techniques for identifying fraudulent names across many languages.

It is quite incredible the degree to which companies are enabling and adapting their products in order to have them accepted in target regions. I’m not talking about translations and number formats here: it’s about supporting all writing directions, accurate and detailed rendering of complex scripts and perfect fluency in generated messages that involve levels of plurality, gender, formality and style. And the open and collaborative nature of the efforts to document this information in the form of the Common Locale Data Repository is commendable.

Temporarily Satiated

I’ve not updated my blog for a couple of months because I’ve been binge learning and sailing the East and South Coasts of Ireland.

Despite several unsuccessful past attempts, due to lack of free time, to complete a Deep Learning course, I’m delighted to have now finished and passed the first course in deeplearning.ai’s Neural Networks and Deep Learning specialization on Coursera.

Certificate

I blew most of my annual leave on cruising between Dún Laoghaire and Schull, Co. Cork: my justification being that the only way to improve my technique is to get out onto the sea. A great combination of learning and adventure.

IMG_1860

Dolphins off of the South Coast of Ireland

FastNet Rock

Rounding the Fastnet Rock Lighthouse

Serverless Machine Translation

It is well known that you can produce relatively good quality machine translations by doing the following:

  • Carry out some processing on the source language.
    Such as remove text which serves no purpose in the translations (say, imperial measurements in content destined for Europe); re-order some lengthy sentences; mark the boundaries of embedded tags, etc.
  • Use custom domain trained machine translation engines.
    This is possible with several machine translation providers. If you have an amount of good quality bilingual and monolingual corpora relevant to your subject matter then you can train and build engines which will produce higher quality output than a general public domain engine.
  • Post process the raw machine translation output to correct recurrent errors.
    To improve overall fluency; replace specific terminology, etc.

We decided to implement this in a fully automated Azure Functions pipeline.

NOTE: Some MT providers have this capability built into their services but we wanted the centralized flexibility to control the pre- and post-editing rules and to be able to mix and match which MT providers we get the translations from.

The pipeline consists of three functions: preedit, translate and postedit. The json payload used for inter-function communication is Jliff. Jliff is an open object graph serialization format specification being designed by an OASIS Technical Committee.

NOTE: Jliff is still in design phase but I’m impatient and it seemed like a good way to test the current snapshot of the format.

The whole thing is easily re-configured and re-deployed, and has all the advantages of an Azure consumption plan.

We can see that this pipeline would be a good candidate for durable functions so once we have time we’ll take a look at those.