Customizing plugin

Successfull installation of the plugin results in immediate appearance of the recommendation list on your posts. Now, plugin comes in two different flavors. Content based is the one that recommends posts with the most similar contents to the current one. While collaborative filtering (CF) yields in recommending posts that are more likely to the user. So the first one is content focused, while the second one is user focused.

Changing the flavor of recommender is done by altering the Recommender type option. The content based approach does not require any additional conditions to work. But CF does. Because CF learns what items to recommend from the browsing history of your users. This browsing history is stored on this server, therefore you must register and create an account. By creating an account you obtain the account identifier and a secret key that you insert into options (at the bottom of the plugin settings page, see the figure). Note also that CF type does not produce recommendations immediately. Some period of time must elapse before enough data is collected. This time depends from the traffic your site receives. Therefore it is more appropriate for the pages serving thousands of pages per day.

Other opions are a bit more easy to understand. The way list is displayed can be changed by altering the Tagline, List Size, and Display Type values. Using css can help you fit the design (for those who intend to do so, recommendations are stored within a <div> with an id __recommendations__).

Display Position allows you to select whether you want to display recommendation list on posts, pages or both. Note that selecting a particular type, e.g. posts, also recommends items from this group only.

Display Percentage enables you to show recommendations to a certain subgroup only. This is useful in two scenarios. First, you are not sure how people will react therefore you want to have some try period on some subset of visitors. Second, you would like to measure how browsing habbits of visitors differ when presented with recommendations. This is only possible in combination with Google Analytics (GA). Statistics for such subgroup are accessible via Visitors | User defined menu item in GA (see the leftmost figure below). By clicking the __aliiiked__ value in the User Defined Value list (see the figure below) you obtain statistics for this subgroup of visitors compared to the average. The statistics figure below shows an increase of the average time on site and pages viewed per visit, and a decrease of bounce rate. Of course this is a sample scenario.

Changing the Ajax Path comes handy if the path to your plugins directory was changed for any particular reason.