Source

orange-text / docs / old-html / widgets / catalog / Text / WordNgrams.htm

Full commit
mitar 4d843d6 






























































<html>
<head>
<title>Word n-grams</title>
<link rel=stylesheet href="../../../style.css" type="text/css" media=screen>
<link rel=stylesheet href="style-print.css" type="text/css" media=print></link>
</head>

<body>

<h1>Word n-grams</h1>

<img class="screenshot" src="../icons/WordNgram.png">
<p>Construct the word n-grams representation of documents.</p>

<h2>Channels</h2>

<h3>Inputs</h3>

<DL class=attributes>
<DT>Examples (ExampleTable)</DT>
<dd>Attribute-valued data set.</dd>
</dl>

<h3>Outputs</h3>
<DL class=attributes>
<DT>Examples (ExampleTable)</DT>
<DD>Attribute-valued data set with word n-grams as metaatributes.</DD>

<h2>Description</h2>

<p>The word n-grams widget constructs the representation of documents using
word n-grams. Word n-grams are sequences of n consecutive words that appear
in the text, uninterrputed by punctuation. Same as in the bag of words widget,
text features (in this case word n-grams) are added as metaatributes to documents.
The value corresponding to a metaatribute is the frequency of that metaatribute
(word n-gram) in the particular document. The No. of words box lets you choose
how many consecutive words make a word n-gram. It is possible to choose word
n-grams of two, three, or four letters. The last option in the box, Named entities,
adds named entities as features for documents. Named entities can have any length
and are extracted based on capitalization of words. The association measure box
is used to choose according to which association measure are the word n-grams extracted.
For named entities, it is not possible to choose an association measure because
of the way they are extracted.
Threshold box is used to input the minimal score a word n-gram has to receive,
according to the chosen association measure, to be kept as a feature. The Stopwords
File box is used to input the list of stop words for the language in which the text
is written. Note that stop words should not be removed using the Preprocess widget
if word n-grams are meant to be used as features. The number of different word n-grams in
the entire collection is shown on the bottom of the widget.</p>

<a href="WordNgram.png"><img class="schema" src="WordNgram.png" alt="Word n-grams widget"></a>

<h2>Examples</h2>

<p>Below is a simple example how to use this widget. The input is fed
from the <a href="Preprocess.htm">Preprocess</a> widget, and the output
is sent to the <a href="TextFeatureSelection.htm">Feature selection</a> widget.</p>

<a href="WordNgram-Example.png"><img src="WordNgram-Example.png" alt="Schema with WordNgram"
 class="schema"></a>

</body>

</html>