Grouping of articles under a particular story

The title and content of a news article are the two main entities that decide the story to which an article will be added. Thus, for every article entering the Sprinklr database, the title and content of the article are analyzed using our advanced AI algorithms, which are, then, either –

1. matched to an existing story with a similar title and content, or

2. a new story is generated with that article being the primary article of the story.

By default, the first article mapped to the story is selected as the primary article. The primary article of the story gets updated based on the MOZ Rank. If an article with a greater MOZ Rank than the primary article is added to the story, the new article is crowned as the primary article. The image, title, and summary of the story are also updated according to this new primary article.

For creating a story, apart from messages from online news and blogs, TV, print, and tweets from verified profiles are also considered. Any shares of the news article on social channels such as Twitter, Facebook public pages, and Reddit (i.e. if a tweet/post contains a news article URL) are also grouped under the same story.

What is the lifetime of a story?

The lifetime of a story is 4 days. Incoming messages are not added to the story if the corresponding story has been created before 4 days. We have this criterion to ensure that the stories are relevant and up to date. Any similar message published after 4 days is used to generate a new story. This is also the reason why users might also see similar stories because the story creation times of both stories are more than a week apart. However, the social shares count is tracked for 15 days since the article was added to the story. Finally, the social shares of the story will be the combined shares of all the news articles attached to that story.

Determining the title and the summary of the story

Out of all the articles grouped under a story, the article which has the highest MOZ Rank, i.e. highest popularity is selected as the primary article. The image, headline, and summary are derived from it.

• The title of the story would be equal to the headline of the primary article. Note that the title of the article is based on the title available on the online website or print media.

• The summary of the story is derived as follows –

  • The entire content of the selected news article is broken into sentences and meaningful phrases. Then, each sentence/phrase is given a rank based on its importance using a proprietary algorithm. Now, based on the significance scores, the top 2 sentences that are related to the title are selected for summary.

    Note: For premium content, the summary is based on the first 256 characters of the text content.

​How is the language of the story card determined?

A story card is generated in all the 32 languages available in MMA, namely English, Arabic, Chinese - Simplified, Chinese - Traditional, French, German, Italian, Japanese, Korean, Dutch, Polish, Portuguese, Spanish, Thai, Turkish, and Russian. The language in which the title and summary are displayed in story cards can be controlled through the user language preference in users' settings. More details are in this document.

Please find below the different scenarios on how stories can be leveraged with the language filter –