AI-based advertisement audience identification method and system

Abstract

The application relates to the technical field of audience identification, in particular to an advertisement audience identification method and system based on AI, which comprises the following steps: obtaining user advertisement click sequence and jump data, identifying event-intensive and interval segments, extracting page stay, jump depth and jump-out rate, screening credible segments, analyzing label track stability, constructing a feature vector for identifying a group, and judging reverse index density to mark non-target audiences. In the application, time intervals between behaviors and page jump depth are used to construct a time sequence set and identify event-intensive and interval segments, statistical indexes of jump-out rate, stay time and co-occurrence frequency are used to determine the credibility of behavior segments, the identification accuracy of advertisement audiences and the consistency of label aggregation are improved, the reverse index density in the behavior track and the label group behavior mean deviation are combined to complete the exclusion judgment of non-target users, and the precision control of advertisement resource delivery and the crowd matching efficiency are enhanced.

Images