A5039040673- Recommender Systems and Techniques
- Human Mobility and Location-Based Analysis
- Advanced Bandit Algorithms Research
- Image Retrieval and Classification Techniques
- Caching and Content Delivery
- Privacy-Preserving Technologies in Data
- Data Management and Algorithms
- Image and Video Quality Assessment
- Traffic Prediction and Management Techniques
- Data-Driven Disease Surveillance
- Context-Aware Activity Recognition Systems
- Time Series Analysis and Forecasting
- Advanced Clustering Algorithms Research
- Advanced Graph Neural Networks
- Topic Modeling
- Mobile Crowdsensing and Crowdsourcing
- Data Stream Mining Techniques
- Complex Network Analysis Techniques
- Web Data Mining and Analysis
- Indoor and Outdoor Localization Technologies
- Transportation Planning and Optimization
- Privacy, Security, and Data Protection
- Remote-Sensing Image Classification
- Expert finding and Q&A systems
- Advanced Image and Video Retrieval Techniques
Beijing University of Technology
2025
RMIT University
2015-2024
The Royal Melbourne Hospital
2024
MIT University
2018-2023
Tarim University
2022
Xinjiang Production and Construction Corps
2022
Deakin University
2010-2015
Zhengzhou University
2008-2009
Ranking algorithms in recommender systems influence people to make decisions. Conventional ranking based on implicit feedback data aim maximize the utility users by capturing users' preferences over items. However, these utility-focused tend cause fairness issues that require careful consideration online platforms. Existing fairness-focused studies does not explicitly consider problem of lacking negative data, while previous methods ignore importance recommendations. To fill this gap, we...
Traditionally, recommender systems modelled the physical and cyber contextual influence on people's moving, querying, browsing behaviors in isolation. Yet, searching, moving are intricately linked, especially indoors. Here, we introduce a tripartite location-query-browse graph (LQB) for nuanced recommendations. The LQB consists of three kinds nodes: locations, queries, Web domains. Directed connections only between heterogeneous nodes represent influences, while homogeneous inferred from...
As a popular technique in recommender systems, Collaborative Filtering (CF) has received extensive attention recent years. However, its privacy-related issues, especially for neighborhood-based CF methods, can not be overlooked. The aim of this study is to address the privacy issues context methods by proposing Private Neighbor (PNCF) algorithm. algorithm includes two privacy-preserving operations: Selection and Recommendation-Aware Sensitivity. constructed on basis notion differential...
We investigate the impact of popularity bias in false-positive metrics offline evaluation recommender systems. Unlike their true-positive complements, reward systems that minimize recommendations disliked by users. Our analysis is, to best our knowledge, first show tend penalise popular items, opposite behavior metrics—causing a disagreement trend between both types presence biases. present theoretical identifies reason disagree and determines rare situations where might agree—the key...
Recommender systems aim to suggest relevant items users among a large number of available items. They have been successfully applied in various industries, such as e-commerce, education and digital health. On the other hand, clustering approaches can help recommender group into appropriate clusters, which are considered neighborhoods prediction process. Although it is fact that preferences vary over time, traditional fail consider this important factor. To address problem, social system...
Human mobility prediction is a core functionality in many location-based services and applications. However, due to the sparsity of data, it not an easy task predict future POIs (place-of-interests) that are going be visited. In this paper, we propose MobTCast, Transformer-based context-aware network for prediction. Specifically, explore influence four types context prediction: temporal, semantic, social geographical contexts. We first design base feature extractor using Transformer...
As one of the biggest challenges in research on recommender systems, data sparsity issue is mainly caused by fact that users tend to rate a small proportion items from huge number available items. This becomes even more problematic for neighborhood-based collaborative filtering (CF) methods, as there are lower numbers ratings neighborhood query item. In this paper, we aim address context CF. For given (user, item), set key first identified taking historical information both user and item...
Understanding and predicting human mobility is vital to a large number of applications, ranging from recommendations safety urban service planning. In some travel the ability accurately predict user's future trajectory for delivering high quality service. The accurate prediction detailed trajectories would empower location-based providers with deliver more precise users. Existing work on has mainly focused next location (or set locations) visited by user, rather than continuous (sequences...
Existing human mobility forecasting models follow the standard design of time-series prediction model which takes a series numerical values as input to generate value prediction. Although treating this regression problem seems straightforward, incorporating various contextual information such semantic category each Place-of-Interest (POI) is necessary step, and often bottleneck, in designing an effective model. As opposed typical approach, we treat translation propose novel through language...
As each user tends to rate a small proportion of available items, the resulted Data Sparsity issue brings significant challenges research recommender systems. This becomes even more severe for neighborhood-based collaborative filtering methods, as there are lower numbers ratings in neighborhood query item. In this paper, we aim address context filtering. Given (user, item) query, set key identified, and an auto-adaptive imputation method is proposed fill missing values ratings. The can be...
Network alignment is useful for multiple applications that require increasingly large graphs to be processed. Existing research approaches this as an optimization problem or computes the similarity based on node representations. However, process of aligning every pair nodes between relatively networks time-consuming and resource-intensive. In paper, we propose a framework, called G-CREWE (Graph CompREssion With Embedding) solve network problem. uses embeddings align two levels resolution,...
Understanding the association between customer demographics and behaviour is critical for operators of indoor retail spaces. This study explores such an based on a combined understanding Cyber (online), Physical, (some aspects of) Social (CPS) behaviour, at conjunction corresponding CPS We combine results traditional questionnaire with large-scale WiFi access logs, which capture cyber physical behaviour. investigate predictability user behaviors captured from both sources. find (1) strong...
Intelligent assistants can serve many purposes, including entertainment (e.g. playing music), home automation, and task management timers, reminders). The role of these is evolving to also support people engaged in work tasks, workplaces beyond. To design truly useful intelligent for work, it important better understand the tasks that are performing. Based on a survey 401 respondents' daily activities setting, we present classification work-related analyze their key characteristics,...
We analyze 18 million rows of Wi-Fi access logs collected over a one year period from 120,000 anonymized users at an inner-city shopping mall. The dataset gathered opt-in system provides users' approximate physical location, as well Web browsing and some search history. Such data unique opportunity to the interaction between people's behavior in retail spaces their behavior, serving proxy information needs. find: (1) use network maps opening hours mall; (2) there is weekly periodicity visits...
Fog computing is an emergent paradigm that extends the cloud paradigm. With explosive growth of smart devices and mobile users, no longer matches requirements Internet Things (IoT) era. a promising solution to satisfying these new requirements, such as low latency, uninterrupted service, location awareness. As typical network architecture, fog raises challenges, privacy, data management, analytics, information overload, participatory sensing. In this article, we present fog-based hybrid...
In this paper, we address the neighborhood identification problem in presence of a large number heterogeneous contextual features. We formulate our research as queue wait time prediction for taxi drivers at airports and investigate factors related to time, weather, flight arrivals, trips. The neighborhood-based methods have been applied type previously. However, failure capture relevant their weights during calculation neighborhoods can make existing ineffective. Specifically, driver...