Relevance and Quality

Relevance of a topic

The relevance of a topic is determined as the sum of the relevancies of the articles that define the topic.

Important: Article relevancies decrease over time. For example, a topic with 5 articles published in the last hour has much higher relevance than a topic with 20 articles, but all of them are a day or two old.

Relevance of an article

The relevance of an article is composed of the following static and dynamic factors:

• Static: quality of the article (see below)
• Static: relevance of the source (see below)
• Dynamic: temporal relevance - Each article is assigned a temporal relevance that decreases over time and is re-determined as needed.

Quality of an article

DeepNews determines the quality of an article via various 'scores':

• 'Click-bait score' - How descriptive is the title of the article
• 'Focus Measure' (in brief) - Does the article focus on one topic or is it written about several different topics in a digressive manner.
• Length of Article - As a rule, the accurate presentation of a news item requires a certain minimum length of text.

Relevance of a source

In DeepNews, the relevance of a source is essentially measured by how much that source is read, or whether it is a regional or national newspaper.

Personal relevance

If the relevance of a topic in relation to individual interests is concerned, the similarity of the topic to these interests plays a significant role here.

Relevance of search results

The relevance of a search result is measured primarily by its similarity to the search query and to a lesser extent by its novelty.