Instagram’s algorithm for identifying trending hashtags involves calculating observed and baseline probabilities. For each hashtag, counters track the number of media posts tagged with it within a 5-minute window over the past 7 days. The observed probability, P(h, t), for a hashtag h at time t, is determined by normalizing the count of posts with that hashtag. This normalization accounts for the variability in hashtag usage over time. The method involves dividing the frequency of a hashtag in the current 5-minute window by either the sum of that hashtag’s frequency over all times or by the total count of all hashtags in the current window. This approach helps in understanding which hashtags are gaining traction in real-time, providing insights into what content is becoming popular on the platform.
Source: stackoverflow.com









