Temporal language grounding (TLG) aims to localize query-related events in videos, which explores how to cognize relationships of video content with language descriptions. According to selective visual attention mechanism in cognitive science, people's cognition and understanding of what happens often rely on dynamic foreground information in the video. Nonetheless, background usually predominates the scenes so that query-related visual features and irrelevant ones are confused. Thus, we propose a Foreground Enhanced Network (FEN) to diminish the background effect from two aspects. FEN at first in spatial dimension explicitly models the evolving foreground in video features by removing relatively unchanged background content. Besides, we propose a progressive contrastive sample generation module to gradually learn the differences between the predicted proposal and its elongated proposals that include the former as a portion, thereby distinguishing similar neighborhood frames. Experiments on two common-used datasets show the efficacy of our model.

Stance detection for unseen targets is designed to automatically identify the user's stance or attitude towards various new targets that are constantly appearing with no labels. Inspired by work in cognitive science, we distinguish functions between systems for syntactic and semantic to enhance stance detection. First, we construct a dual-view graph and utilize unsupervised graph contrastive learning to capture target-invariant features influencing stance expression from a syntactic structure perspective. Second, we use an attention mechanism to learn the relationship between syntactic pattern features and a given target, and fuse the two parts to enhance the model's ability to predict unseen targets. Meanwhile, we employ the interactive GCN to maintain the global semantics of the dual-view graph fusion and ensure the stability and validity of the learned syntactic representations. Comprehensive experiments on stance detection of unseen targets verify the effectiveness and superiority of our proposed method.