Traditional short video recommendations primarily enhance user retention by reinforcing existing user preferences, potentially leading to information cocoons. Conversely, proactive recommendations aim to diversify user interests by exposing users to content beyond their historical preferences. However, current proactive approaches face three limitations: (1) homogeneous receptivity assumption, neglecting individual differences in users' openness to new interests; (2) short-term item exposure without interest anchoring, focusing on item-level shifts rather than interest evolution; and (3) static feedback utilization, failing to incorporate dynamic user feedback during the recommendation adequately. To address these challenges, we propose ProRec-Video, a proactive framework that guides hierarchical interest transitions through three innovations. First, User Receptivity Profiling assesses individual openness for new interests, ensuring personalized transition pacing. Second, Hierarchical Interest Transition Planning decomposes complex interest shifts into intermediate steps to generate smooth interest transition paths and semantically coherent video sequences, addressing overemphasis on item exposure. Third, Dynamic Feedback Adaptation integrates agent-based simulation and Reflexion mechanisms to refine interest transition paths and video sequences based on real-time user feedback, enhancing adaptability and satisfaction. Extensive experiments on two datasets demonstrate that ProRec-Video achieves a significant improvement in proactive recommendation performance, with an interest transition success rate of 85% and a user satisfaction rate of 78.3%.

Abstract The typical aim of relational triple extraction is to identify entities along with their relations from unstructured text, which is a crucial task in information extraction. Recent methods achieve considerable performance by mining semantic associations within the input sentence but they hardly exploit the equally important semantic meaning of relations. Most methods simply represent relations as numeric labels and the rich semantic information is not fully utilized to enhance performance. To address the issue, we decompose the task into two sequential subtasks, entity pairing and relation matching, from a novel perspective and then propose a semantic-preserving model SPRel. Specifically, SPRel first extracts entity pairs associated with at least one relation, and then matches them with certain relations according to the descriptive text of relations. Comprehensive experiments on two widely used datasets demonstrate that SPRel outperforms previous methods, particularly in handling complex scenarios.

User identity linkage aims to correlate multiple virtual identities of the same person across different social networks. It is widely used in person profile integration, potential friend recommendation, user behavior prediction, identity verification, etc. Since the data inconsistency and network heterogeneity among social networks, extracting features from profiles, contents, and network structures may lead to severe random noises. To this end, we propose PsyLink, a user identity linkage method via psychological characteristic modeling. To reduce the data noise, PsyLink extracts both internal personality characteristics and external interest characteristics to represent an individual. Furthermore, we design a neighborhood enhancement strategy to capture latent higher-order structural features through a graph contrastive learning technique. Under various parameter settings, the experimental results demonstrate that the macro-f1 of PsyLink improves 19.23% on average compared with state-of-the-art deep learning models. Combined with the psychological characteristics and graph contrastive learning, PsyLink is able to adequately learn the similarities and differences between user identities across social networks.