2021-11-03: The Return of SHARI -- Bringing News and Web Archive Storytelling Together Again

A SHARI story covering the biggest news story of October 25, 2010, involving revelations about Facebook.
Follow @StormyArchives on Twitter for daily updates on yesterday's biggest news story.

SHARI is back. Each day, the Dark and Stormy Archives Project will apply the SHARI process to gather the news articles for the previous day's top stories and present them as a social media story, as seen above. Each card in the story links to a memento of a news article in a web archive. Each story will be shared via the @StormyArchives social media account. SHARI depends on Dr. Alexander Nwala's StoryGraph service. We thank the Internet Archive for helping us restore StoryGraph, thus making SHARI possible once again.

SHARI uses StoryGraph and the DSA Toolkit to apply all of the storytelling processes from the five-process storytelling model.

SHARI satisfies all of the processes from our five-process storytelling model. As shown above, SHARI uses StoryGraph to select exemplars from the news articles of different news sources. In this case, the exemplars represent the news articles from the biggest news story of the day. Nwala's StoryGraph applies graph weights to different news sources based on the named entities found in these articles. He used StoryGraph to identify the biggest news stories at different points in various years as part of "365 dots in 2018", "365 dots in 2019", and "366 dots in 2020." He presented StoryGraph as part of "365 Dots in 2018, 2019, and 2020: Quantifying Attention of News Sources" at the Computation + Journalism Symposium in 2021.

The full tool and process breakdown for SHARI.

As shown above, SHARI depends on the other tools in the DSA Toolkit to help visualize the story. 

  1. StoryGraph selects exemplars and the StoryGraph Toolkit creates a list of the URI-Rs for the biggest story of the day.
  2. Hypercane asks web archives to preserve the news articles at these URI-Rs as mementos and outputs these mementos as a list of URI-Ms.
  3. Hypercane generates a list of named entities and their frequencies from these articles.
  4. Hypercane generates a list of sumgrams (phrases) and their frequencies from these articles.
  5. Hypercane analyzes the images in the articles and generates a report recommending the overall story's striking image.
  6. Hypercane synthesizes the list of URI-Ms, the named entities, the sumgrams, and the images into a JSON file formatted for Raintale.
  7. Raintale extracts the URI-Ms from this JSON file and sends them to MementoEmbed. Raintale then uses the template to help it render each article as a card based on this information from MementoEmbed. Raintale also uses the template to render the other information from the JSON (story striking image, named entities, etc.).
  8. We push the Jekyll HTML code to GitHub Pages where it is distributed on the DSA Puddles website for anyone to view.
  9. (not shown in the diagram) We tweet the new story so that followers know about it, as shown below ⬇️.

In 2019, Alexander Nwala and I teamed up to pair his StoryGraph service with the DSA toolkit to provide yesterday's top news story in a storytelling format. Because of the tools involved we named it SHARI (StoryGraph Hypercane ArchiveNow Raintale Integration). We presented SHARI at the Web Archiving and Digital Libraries Workshop in 2020.

In the Summer of 2021, disaster struck. Old Dominion University was the victim of a computer security incident. I do not know the details, but all of our systems were taken offline. This included StoryGraph and the machine that ran the SHARI process. It took us months to find solutions for various projects. Dr. Nwala worked with the Internet Archive to find a home for StoryGraph and updated the StoryGraph toolkit with information about this new home so that SHARI could run again. Once that was in place, I dusted off an old MacBook Pro to serve as the scheduling and execution server for SHARI. The machine is semi-reliable and requires care and feeding in its old age. I will be looking for a more permanent solution for running SHARI as time goes on.

Long live SHARI on my desk.

Because StoryGraph has information about the top news stories starting in 2017, we will be backfilling the content of the DSA Puddles website with older stories over time. I will have to coordinate with Dr. Nwala to make sure we have good coverage and avoid days when we lost data. I am also working on a schedule for generating more stories from the web archive collections at Archive-It and the National Library of Australia, but I need to make sure that this processing does not affect SHARI. Dr. Nwala will be bringing StoryGraphBot back online soon to present updates of the top news stories as they happen.

So, follow @StormyArchives for daily SHARI updates, and thanks for your support of the StoryGraph and Dark and Stormy Archives Projects.