A Genetic Algorithm for Finding a Small and Diverse Set of Recent News Stories on a Given Subject: How We Generate AAAI’s AI-Alert

Abstract

This paper describes the genetic algorithm used to select news stories about artificial intelligence for AAAI’s weekly AIAlert, emailed to nearly 11,000 subscribers. Each week, about 1,500 news stories covering various aspects of artificial intelligence and machine learning are discovered by i2k Connect’s NewsFinder agent. Our challenge is to select just 10 stories from this collection that represent the important news about AI. Since stories and topics do not necessarily repeat in later weeks, we cannot use click tracking and supervised learning to predict which stories or topics are most preferred by readers. Instead, we must build a representative selection of stories a priori, using information about each story’s topics, content, publisher, date of publication, and other features. This paper describes a genetic algorithm that achieves this task. We demonstrate its effectiveness by comparing several engagement metrics from six months of “A/B testing” experiments that compare random story selection vs. a simple scoring algorithm vs. our new genetic algorithm.

Assessing the behavior and performance of a supervised term-weighting technique for topic-based retrieval

“…Topic-based retrieval can be used to generate topic-based alerts [12], support expert as they organize domain knowledge [21], and build vertical portals [25], among many other applications. Depending on the application at hand, the emphasis of topic-based retrieval can be placed on attaining high precision, high recall or a combination of both.…”

Evolution of a Robust Artificial Intelligence System: A Case Study of the Association for the Advancement of Artificial Intelligence's AI‐Alert

“…During the timespan of our recent experiment, which we detail in this article, we gathered a variety of more specific user‐engagement metrics. These metrics were published previously in our Innovative Applications of Artificial Intelligence conference paper about our experiment (Eckroth and Schoen, 2019). An average of thirty percent of recipients open any given week's alert, although most (about sixty percent) have never clicked a link in any alert (albeit we have anecdotal evidence that these subscribers derive value from reviewing the titles alone).…”

“…With this in mind, we next describe NewsFinder's implementation and the experiment we conducted with the new genetic algorithm for story selection. These sections were documented previously in our Innovative Applications of Artificial Intelligence‐deployed application paper, “A Genetic Algorithm for Finding a Small and Diverse Set of Recent News Stories on a Given Subject: How We Generate AAAI's AI‐Alert ” (Eckroth and Schoen, 2019).…”