Abstract
Local news organizations face myriad challenges in handling copious amounts of data and information. The current approach of manually reviewing, classifying and pub-lishing news or analyzing raw news source data from public or law enforcement agencies leads to reduced quality, scale and timeliness in handling important news. In this paper, we present AI/ML-driven solutions to automate mundane processes involved with news content curation and storytelling. We present two pipelines particularly targeting local newsrooms with resource limitations that can be aided by automation solutions. The first automated pipeline performs a newsworthiness assessment from e-mail data and involves a neural network approach to classify the news as “Worthy”, “Not Worthy” or “Unsure”. The second automated pipeline addresses news storytelling from police blotters encompassing multiple law enforcement agencies, and involves multi-format parsers to streamline the collection, formatting, and publication. Experiment results of our first pipeline shows our approach achieves an accuracy rate of 91 % after successful and seamless integration of the classifier with real-world datasets. Integration results of our second pipeline shows that our approach had 96 % effectiveness in integration with story generation technology for reporting crime incidents.