You are looking at the HTML representation of the XML format.
HTML is good for debugging, but probably is not suitable for your application.
See complete documentation, or API help for more information.
<?xml version="1.0"?>
<api>
  <query>
    <search>
      <p ns="0" title="The chaining algorithm" snippet="...wo different modes: addition or replacement. The default mode is addition, &lt;span class=&#039;searchmatch&#039;&gt;meaning&lt;/span&gt; that the declared output features for the web service are added to the ones&#10;" size="4318" wordcount="720" timestamp="2013-01-02T14:31:09Z" />
      <p ns="0" title="WebLicht Overview" snippet="...hat can process the data described in the profile. On a successful answer (&lt;span class=&#039;searchmatch&#039;&gt;meaning&lt;/span&gt; that the repository found at least one such service) WebLicht updates the g&#10;" size="2376" wordcount="379" timestamp="2012-07-02T09:29:16Z" />
      <p ns="0" title="WebLicht in Detail" snippet="...rent modes: &#039;&#039;addition&#039;&#039; or &#039;&#039;replacement&#039;&#039;. The default mode is addition, &lt;span class=&#039;searchmatch&#039;&gt;meaning&lt;/span&gt; that the declared output features for the web service are &#039;&#039;&#039;added&#039;&#039;&#039; to th&#10;" size="7410" wordcount="1138" timestamp="2012-05-25T14:29:58Z" />
      <p ns="0" title="POS Tagger" snippet="...k tag set (for English). Part-of-speech taggers must take into account the &lt;span class=&#039;searchmatch&#039;&gt;meaning&lt;/span&gt; of a word, its position in a sentence, and its relation to surrounding word&#10;" size="1745" wordcount="276" timestamp="2012-03-07T16:16:48Z" />
      <p ns="0" title="Named Entity Recognition" snippet="Although the &lt;span class=&#039;searchmatch&#039;&gt;meaning&lt;/span&gt; of such types of named entities as person, location and organization is rel&#10;" size="2388" wordcount="360" timestamp="2012-03-07T16:20:11Z" />
      <p ns="0" title="Word Sense Disambiguation" snippet="...ing to that word. The automatic identification which sense of a word (i.e. &lt;span class=&#039;searchmatch&#039;&gt;meaning&lt;/span&gt;) is used in a sentence is described as the problem of Word Sense Disambigua&#10;...rpus-based, where the collocates of a word are used to define the words’ &lt;span class=&#039;searchmatch&#039;&gt;meaning&lt;/span&gt;.&#10;" size="1027" wordcount="185" timestamp="2012-03-07T16:20:30Z" />
      <p ns="0" title="Semantic Relations" snippet="...led lexical units) where all the words are taken to have (almost) the same &lt;span class=&#039;searchmatch&#039;&gt;meaning&lt;/span&gt;. Thus a synset is a set-representation of the semantic relation of synonymy&#10;" size="949" wordcount="148" timestamp="2012-03-07T16:20:44Z" />
      <p ns="0" title="Tools in Detail" snippet="...ilarities. Words with the same PoS are in some sense similar in their use, &lt;span class=&#039;searchmatch&#039;&gt;meaning&lt;/span&gt;, or function.&#10;...gy is the study of how words and phrases change in form depending on their &lt;span class=&#039;searchmatch&#039;&gt;meaning&lt;/span&gt;, function and context. Morphological tools sometimes overlap in their funct&#10;" size="78119" wordcount="11174" timestamp="2024-07-22T12:42:38Z" />
    </search>
  </query>
</api>