Below are some of the most common Content processing issues. The main areas were identified during our review of site qualities and may have various impacts on the function of our platform’s features, starting from the processing of content in the first place up to the resulting recommendations that are presented to users.
Cookie consent notification
Upon their initial visit, each user is usually presented with a notification to consent to the setting and processing of cookies. In case the cookie notice is being fetched as the body or content of an article, this severely limits the NLP processing our system can do. This Natural language processing plays an important part in content classifications.
The impact of this issue with the incorrectly fetched cookie notice is a high rate of wrong content classifications (as tech-and-computing is prevalent since cookie notices usually contain tech keywords, even if the content of the article has an altogether different classification).
This leads to wrong User Interest segments and high risks of recommending the wrong articles to users.
How to fix this issue?
In order to fix this, you will need to whitelist the crawler in your Consent Management Platform (CMP) or similar system. More information about the whitelisting of the crawler is available here.
Missing publication date
The date and time of when an article’s been published is commonly used in filters to ensure only fresh articles are recommended when using Piano Content. This information also has to be correct, and in addition to the date, the relevant meta tag should also include the hour and minute of the article’s publication.
An incorrect publication date leads to a high risk of recommending too old articles to users.
How to fix this issue?
In order to fix this, you will need to add a meta tag to all articles with the correct publication time (date, hour and minute). More information about this can be found here.
Duplicate URLs
Due to lower site quality, where multiple URLs have the same titles, issues with duplicated content recommendations may occur. For example, the same article can be recommended twice or more times to the same user. This error is very visible to users and disrupts the overall experience of the visit.
How to fix this issue?
You need to ensure that all articles have a meta tag with the canonical URL. The canonical URL is expected to stay the same regardless of any updates to the article. These may include changes to the title or any other tags.
The URL that identifies a document is normally based on the canonical URL. It can be defined explicitly via cXenseParse:url markup, a canonical link, or open graph og:url meta data.
The supported meta tags, in prioritized order, are:
<meta name="cXenseParse:url" content="https://www.website.com/section/123.html"/> <link rel="canonical" href="https://www.website.com/section/123.html" /> <meta property="og:url" content="http://wwws.website.com/section/123.html"/>