Analysis of web page content

This page has been translated automatically

We want to provide you with the latest help content in your language as soon as possible. This page has been translated automatically and may contain grammatical errors or inaccuracies. We want this content to be useful to you. Please let us know at the bottom of this page if this information was helpful.

View the original article in Russian: Анализ содержимого web-страницы

Context Recognizer and the Google Menagerie

As you know, Google has set its beasts against SEO sites all over the world. One of the main (and, probably, the most important and complex) parameters that animals are trained for is the non-thematic nature of the sites with which your site is connected by incoming and outgoing links.

To protect your sites from rabid animals, we have come up with a new feature - Context Recognizer .

The Context Recognizer will help you determine the subject of the text of the web page or just the text that you specify. Instead of spamming your links to all resources in a row, you can first define the topic of the text of the page on which you want to link. If the text on this page does not fit the theme of your site, then you should not leave links there.

Similarly, you can find links on your site, and check the subject of the pages to which they link.

Let's say you have a posting linkbase. With the help of the new feature Context Recognizer, you can split your database into several databases by context. Then, when you need to advertise the site, you will not take a complete base, but the one that corresponds to the subject of the site. You will be able to post an article about car insurance on an automotive blog, rather than on a blog that publishes new movie announcements.

You can parse a site (like a blog) and find pages that best match your ad's topic. Leaving relevant comments and posts, you will not only get thematic links, but a greater chance of being moderated, which is very important on high-quality resources.

Context Recognizer is now in beta testing, despite this, it has a good recognition rate and we will improve it.


When configuring, specify the text for analysis. Please note that in order not to search for text on a web page yourself, use the function of highlighting the main text on a web page. You can find it in the Select Main Article action.

You can define a general theme of the text (about 20) or a specific direction (about 250) (will be available a little later).

Next, you need to configure two filters:

  • specify the maximum number of topics that the analyzer should display;

  • specify the minimum threshold of relevance to the topic, after which the topic will be considered inappropriate. This parameter ranges from 0 to 100.

Topic of the text

For example, three topics and at least 30 percent coincidence. In this case, no more than 3 suitable topics will be issued, which correspond to your text at least 30.

Please note that less than 3 topics can be displayed or none at all, if the analyzer does not see the similarity of your text with any of the topics known to it.

The variable will contain topics separated by commas.


In the toolbar of the project editor, there is a button for testing the Context Recognizer .

Note that Context Recognizer currently only works with English texts.