Learning Behaviorally Grounded Item Embeddings via Personalized Temporal Contexts

2026-04-16

AI Summary

This work addresses the limitation of existing item embedding methods that overlook temporal dynamics in user behavior, thereby failing to distinguish between short-term session patterns and long-term interest evolution. To overcome this, the authors propose TAI2Vec, a model that enables fine-grained temporal modeling through a user-adaptive time-aware mechanism. Specifically, it integrates anomaly-detection-based session segmentation with user-specific continuous-time decay functions to dynamically capture the temporal semantics of interactions. Implemented within a lightweight embedding learning framework, TAI2Vec consistently outperforms static baselines across eight real-world datasets, achieving state-of-the-art performance in over 80% of them and yielding relative improvements of up to 135%.

Technology Category

• Machine Learning: Time-Series/Data Streams
• Humans and AI: Intelligent User Interfaces
• Data Mining & Knowledge Management: Anomaly/Outlier Detection

Application Category

• User Modeling, Personalization and Recommendation: Attacks and countermeasures in recommendation systems
• Graph Algorithms and Modeling for the Web: Graph embeddings and representation learning for Web-related graphs
• Semantics and Knowledge: Methods to enhance, augment, integrate or synergize semantic models such as knowledge graphs and LLMs

Abstract

Effective user modeling requires distinguishing between short-term and long-term preference evolution. While item embeddings have become a key component of recommender systems, standard approaches like Item2Vec treat user histories as unordered sets (bag-of-items), implicitly assuming that interactions separated by minutes are as semantically related as those separated by months. This simplification flattens the rich temporal structure of user behavior, obscuring the distinction between coherent consumption sessions and gradual interest drifts. In this work, we introduce TAI2Vec (Time-Aware Item-to-Vector), a family of lightweight embedding models that integrates temporal proximity directly into the representation learning process. Unlike approaches that apply global time constraints, TAI2Vec is user-adaptive, tailoring its temporal definitions to individual interaction paces. We propose two complementary strategies: TAI2Vec-Disc, which utilizes personalized anomaly detection to dynamically segment interactions into semantic sessions, and TAI2Vec-Cont, which employs continuous, user-specific decay functions to weigh item relationships based on their relative temporal distance. Experimental results across eight diverse datasets demonstrate that TAI2Vec consistently produces more accurate and behaviorally grounded representations than static baselines, achieving competitive or superior performance in over 80% of the datasets, with improvements of up to 135%. The source code is publicly available at https://github.com/UFSCar-LaSID/tai2vec.

Problem

• item embeddings
• temporal context
• user modeling
• preference evolution
• recommender systems

Innovation

• time-aware embeddings
• personalized temporal context
• behavioral grounding
• session segmentation
• user-adaptive modeling