MachineTranslation.xml

<?xml version="1.0" encoding="UTF-8"?>
<!DOCTYPE book PUBLIC "-//OASIS//DTD DocBook XML V4.5//EN"
"../../../docbook-xml-4.5/docbookx.dtd">
<chapter id="chapter.machine.translate">
  <title>Machine Translation<indexterm class="singular">
      <primary>Machine Translation</primary>
    </indexterm></title>

  <section>
    <title>Introduction<indexterm class="singular">
        <primary>Machine Translation</primary>

        <secondary>Introduction</secondary>
      </indexterm></title>

    <para>As opposed to user-generated translation memories (as in the case of
    <application>OmegaT</application>) Machine translation (MT) tools use
    rule-based linguistic tools to create a translation of the source segment
    without the need for a translation memory. Statistical learning
    techniques, based on source and target texts, are used to build a
    translation model. Machine translation services have been achieving good
    and steadfastly improving results in research evaluations.</para>

    <para>To activate any of the Machine Translation services, go to
    <guimenuitem>Options &gt; Machine Translate ...</guimenuitem> and activate
    the service desired. Note that they are all web-based: you will have to be
    on-line if you want to use them.</para>
  </section>

  <section id="introduction">
    <title>Google Translate<indexterm class="singular">
        <primary>Machine Translation</primary>

        <secondary>Google Translate</secondary>
      </indexterm></title>

    <para>Google Translate is a payable service offered by Google, for
    translating sentences, web sites and complete texts between an
    ever-growing number of languages. At the time of writing the list includes
    more than 50 languages, from Albanian to Yiddish, including of course all
    the major languages. The current version of the service is based on usage,
    with the price of 20 USD per million characters at the time of writing.
    <emphasis role="bold"/></para>

    <para><emphasis role="bold">Important: </emphasis>Google Translate API v2
    requires billing information for all accounts before you can start using
    the service (see <ulink
    url="https://developers.google.com/translate/v2/pricing?hl=en-US">Pricing
    and Terms of Service</ulink> for more). To identify yourself as a valid
    user for the Google services, you use your private unique key sent to you
    by Google, when you have registered for the service. See chapter <link
    linkend="chapter.installing.and.running">Installing and Running</link>,
    section Launch command arguments, for details on how to add this key to
    the OmegaT environment.</para>

    <para>The quality of the Google Translate translation depends on one side
    on the reservoir of target-language texts and the availability of their
    bilingual versions, on the other hand on the quality of the models built.
    It is pretty much certain that while the quality may be insufficient in
    some cases, it will definitely get better with time and not worse.</para>
  </section>

  <section>
    <title><application>OmegaT</application> users and Google
    Translate</title>

    <para>The <application>OmegaT</application> user is not forced to use
    Google Translate. If used, neither the user's decision to accept the
    translation nor the final translation are made available to Google. The
    following window shows an example of a) the English source b) Spanish and
    c) Slovenian Google Translate translation.</para>

    <figure id="moby.dick">
      <title>Google Translate - example</title>

      <mediaobject>
        <imageobject role="html">
          <imagedata fileref="images/MobyDick.png"/>
        </imageobject>

        <imageobject role="fo">
          <imagedata fileref="images/MobyDick.png" width="60%"/>
        </imageobject>
      </mediaobject>
    </figure>

    <para>The Spanish translation is better than the Slovenian. Note
    <emphasis>interesar</emphasis> and<emphasis> navegar</emphasis> in
    Spanish, are correctly translated as the verbs interest and sail
    respectively. In the Slovenian version both words have been translated as
    nouns. It is actually quite probable that the Spanish translation is based
    at least partially on the actual translation of the book.</para>

    <para>Once you have activated the service, a suggestion for the
    translation will appear in the Machine Translate pane every time a new
    source segment is opened. If you find the suggestion acceptable, press
    <keycombo>
        <keycap><indexterm class="singular">
            <primary>Shortcuts</primary>

            <secondary>Machine Translate - Ctrl+M</secondary>
          </indexterm>Ctrl</keycap>

        <keycap>M</keycap>
      </keycombo> to replace the target part of the opened segment with the
    suggestion. In the above segment, for instance, <keycombo>
        <keycap>Ctrl</keycap>

        <keycap>M</keycap>
      </keycombo> would replace the Spanish version with the Slovenian
    suggestion.</para>

    <para>If you do not wish <application>OmegaT</application> to send your
    source segments to Google to get translated, untick the Google Translate
    menu entry in Options.</para>

    <para>Note that nothing but your source segment is sent to the MT service.
    The online version of Google Translate allows the user to correct the
    suggestion and send the corrected segment in. This feature, however, is
    not implemented in OmegaT.</para>
  </section>

  <section>
    <title>Belazar<indexterm class="singular">
        <primary>Machine Translation</primary>

        <secondary>Belazar</secondary>
      </indexterm></title>

    <para><ulink url="http://belazar.info/">Belazar</ulink> is a Machine
    language translation tool for the Russian-Belarusian language pair.</para>
  </section>

  <section>
    <title>Apertium<indexterm class="singular">
        <primary>Machine Translation</primary>

        <secondary>Apertium</secondary>
      </indexterm></title>

    <para><ulink url="http://www.apertium.org/">Apertium</ulink> is a
    free/open-source machine translation platform, initially aimed at
    related-language pairs, like CA, ES, GA, PT, OC and FR but recently
    expanded to deal with more divergent language pairs (such as
    English-Catalan). Check the web site for the latest list of implemented
    language pairs.</para>

    <para>The platform provides</para>

    <itemizedlist>
      <listitem>
        <para>a language-independent machine translation engine</para>
      </listitem>

      <listitem>
        <para>tools to manage the linguistic data necessary to build a machine
        translation system for a given language pair </para>
      </listitem>

      <listitem>
        <para>linguistic data for a growing number of language pairs</para>
      </listitem>
    </itemizedlist>

    <para>Apertium uses a shallow-transfer machine translation engine which
    processes the input text in stages, as in an assembly line: de-formatting,
    morphological analysis, part-of-speech disambiguation, shallow structural
    transfer, lexical transfer, morphological generation, and
    re-formatting.</para>

    <para>It is possible to use Apertium to build machine translation systems
    for a variety of language pairs; to that end, Apertium uses simple
    XML-based standard formats to encode the linguistic data needed (either by
    hand or by converting existing data), which are compiled using the
    provided tools into the high-speed formats used by the engine.</para>
  </section>

  <section>
    <title>Microsoft Translator<indexterm class="singular">
        <primary>Machine Translation</primary>

        <secondary>Microsoft Translator</secondary>
      </indexterm></title>

    <para>In order to get credentials for MS Translator, follow these
    steps:</para>

    <orderedlist>
      <listitem>
        <para>Log into Microsoft Azure Marketplace: <ulink
        url="http://datamarket.azure.com/">http://datamarket.azure.com/</ulink></para>

        <para>If you do not already have an Azure Marketplace account you will
        need to register one first.</para>
      </listitem>

      <listitem>
        <para>Click on the My Account option at the top of the page.</para>
      </listitem>

      <listitem>
        <para>Near the bottom you will see entries and values for:
        <itemizedlist>
            <listitem>
              <para>Primary Account Key (corresponds to
              <code>microsoft.api.client_secret</code> command-line
              parameter)</para>
            </listitem>

            <listitem>
              <para>Customer ID (corresponds to
              <code>microsoft.api.client_id</code> command-line
              parameter)</para>
            </listitem>
          </itemizedlist></para>
      </listitem>
    </orderedlist>

    <para>To enable MS Translator in OmegaT edit its launcher or see chapter
    <link linkend="chapter.installing.and.running">Installing and
    Running</link> to learn how to start OmegaT from the command line.</para>
  </section>

  <section>
    <title>Yandex Translate<indexterm class="singular">
        <primary>Machine Translation</primary>

        <secondary>Yandex Translate</secondary>
      </indexterm></title>

    <para>In order to be able to use Yandex Translate in OmegaT, you need to
    <ulink url="http://api.yandex.com/key/form.xml?service=trnsl">obtain an
    API key from Yandex</ulink>.</para>

    <para>The obtained API key needs to be passed to OmegaT at startup through
    <code>yandex.api.key</code> command-line parameter. To do that edit OmegaT
    launcher or see chapter <link
    linkend="chapter.installing.and.running">Installing and Running</link> to
    learn how to start OmegaT from the command line.</para>
  </section>

  <section id="Google.Translate.troubleshooting">
    <title>Machine translation - trouble shooting<indexterm class="singular">
        <primary>Machine Translation</primary>

        <secondary>Troubleshooting</secondary>
      </indexterm></title>

    <para>If there's nothing appearing in the Machine Translate pane, then
    check the following:</para>

    <itemizedlist>
      <listitem>
        <para>Are you online? You need to be online to be able to use an MT
        tool.</para>
      </listitem>
    </itemizedlist>

    <itemizedlist>
      <listitem>
        <para>What is the language pair you need? Check if the selected
        service offers it.</para>
      </listitem>

      <listitem>
        <para>Google Translate does not work: have you applied <ulink
        url="https://developers.google.com/translate/v2/faq">Translate API
        service</ulink>? Note that Google Translate service is not free of
        charge, see chapter <link
        linkend="chapter.installing.and.running">Installing and Running</link>
        (runtime parameters) for more on that.</para>
      </listitem>

      <listitem>
        <para>"Google Translate returned HTTP response code: 403 ...": check
        that the 38-characters key, entered in the pinfo.list file, is
        correct. Check that <ulink
        url="https://developers.google.com/translate/v2/faq">Translate API
        service </ulink>has been activated.</para>
      </listitem>

      <listitem>
        <para>Google Translate does not work: - with the Google API key
        entered as requested. Check in Options &gt; Machine Translate, that
        Google Translate V2 is checked.</para>
      </listitem>

      <listitem>
        <para>Google Translate V2 reports "Bad request" - check the source and
        target languages for your project. Having no languages defined elicits
        this kind or a response.</para>
      </listitem>
    </itemizedlist>
  </section>
</chapter>