Analysis method of temporal relationship of social network events

Abstract

The invention discloses a method for analyzing the time series relationship of social network events, which includes acquiring event detection result data, extracting the time series of event short text clusters, dynamically adjusting the time series, constructing a quantile-quantile diagram, and analyzing the time series relationship of events . The present invention first extracts the event short text cluster time series from the event short text cluster set, and uses a dynamic time warping algorithm to match the event time series, and then quantitatively calculates the timing correspondence between the event short text clusters according to the matching results Time-series distance, and qualitative analysis of the time-series relationship between event short text clusters through quantile-quantile map visualization method, can significantly improve the recognition accuracy of event time-series relationship in social networks.

Description

technical field [0001] The invention belongs to the technical field of event detection and tracking, and in particular relates to a method for analyzing the temporal relationship of social network events. Background technique [0002] Topic Detection and Tracking (TDT) technology is derived from the earlier Event Detection and Tracking (EDT) technology. The original TDT research defined topics as events. Events were originally described as things that happened at specific times and places. With the development of TDT technology, the definition of topic becomes more extensive. A topic includes not only the subsequent events caused or caused by the original event, but also other events or activities related to it. TDT defines a topic as: a topic consists of a seed event or activity and events or activities directly related to it. The tasks of TDT include segmentation tasks for news reports, tracking tasks for known topics, detection tasks for unknown topics, detection task...

AI Technical Summary

Problems solved by technology

However, this method needs to estimate the distribution of event text streams in advance. In practice, the text distribution characteristics of short text events in social networks may not be obvious, which brings great challenges to the establishment of distribution models.

Images