Abstract: Many trajectories have user ID, location ID, time-stamp, and activity type (like “topic” in text mining) due to the widespread use of positioning devices. Existing works use latent activity types from topic models or…
Abstract: Pattern mining (PM) finds user-interested data patterns. Most studies have examined one pattern, like frequent or high-utility. Single-objective PM methods struggle to meet the growing needs of businesses in various industries. This paper proposes…
Abstract: Heterogeneous information networks (HINs) model objects of different types and their relationships. HIN objects usually have more attributes to enrich their data. AHINs are attributed HINs. We cluster objects in an AHIN based on…
Abstract: HIN embedding learns low-dimensional representations of nodes while preserving HIN structures and semantics. Most methods consider heterogeneous relations and perform well, but they use one model for all relations, which limits HIN embedding. We…
Abstract: GPS and wireless technology generate massive trajectory data. LBSN activity trajectory adds user semantic activities like visiting work/home/entertainment places to traditional trajectory data. Comparing activity trajectories in time, location, and semantics measures their similarity….
Abstract: Representation learning has helped machine learning models excel in many tasks, including machine translation, face recognition, and recommendation. Most representation learning methods require many consistent and noise-free labels. In real life, labels are limited…
Abstract: Density-based clustering methods identify intrinsic clusters by classifying dataset regions as high- or low-density based on neighborhood information. They automatically count clusters of any shape and are popular and effective. Datasets from different applications…
Abstract: Heterogeneous information networks (HIN) can model complex real-world data because their nodes and edges have different semantic types. Thus, HIN embedding, which learns node representations in a low-dimensional space to preserve HIN structural and…
Abstract: We can create a process model from an event log of business process execution traces using automated process discovery. These techniques generate process models with several criteria, including fitness, which measures how well the…
Abstract: Cyberattacks are more organized, persistent, and weaponized. Faced with this situation, a growing number of organizations worldwide are willing to leverage the open exchange of cyber threat intelligence (CTI) to get a full picture…
Abstract: Databases have struggled to find top k frequent items. Finding top-k persistent items is a new problem that has gained attention recently. Users want to know which items are significant—frequent and persistent. No prior…
Abstract: Non-negative tensor factorization can automatically find phenotypes in EHRs with minimal human supervision. In practice, the correspondence between different modalities (e.g., medications and diagnoses) is often missing, so such methods require an input tensor…
WhatsApp us