Research

The goal of our research

Data science lab.은 추천시스템과 지식진화의 연구그룹으로 구성됩니다.

Recommendation system

Recommendation systems are computer programs that provide personalized recommendations to users. They analyze users’ past behavior to understand their preferences and interests, and then recommend various products, services, or content to them accordingly.

  • Recommendation algorithms: This is the core technology of recommendation systems. Researchers develop algorithms that analyze users’ past behavior data to provide personalized recommendations.

  • User experience: Researchers study user experience to improve the quality of recommendations. This includes factors such as transparency, diversity, and trustworthiness.

  • Use of recommendation systems: Researchers study how users interact with recommendation systems. For example, how recommendation systems may influence users’ preferences or purchasing behavior.

  • Application areas: Recommendation systems are applied in various fields, such as music, movies, books, shopping, and travel. Researchers study how recommendation systems can be tailored to specific application areas.

Knowledge evolution

The topic of the knowledge evolution is to combine artificial intelligence, the semantic web, and (stream-)database integration techniques to solve complex, fast-changing information on the Web. We leverage general research techniques across information-intensive disciplines, including smart energy, data integration and the social web.

🌿 Sustainable Web

  • Energy-Efficient Web Design and Frontend Optimization
  • Carbon-Aware Web Workload Scheduling
  • Lifecycle Sustainability Analytics and User Engagement

🧠 Multi-modal Knowledge Graph Reasoning

  • Automated construction of multimodal knowledge graphs integrating text, images, and sensor data
  • Designing Graph Neural Networks for reasoning over multimodal KGs
  • Event reasoning and explainability using KGs
  • Semantic Table Annotation for KG Enrichment
  • Anomaly Detection using Multimodal KGs

⚡ Carbon-aware Computing

  • Cloud workload scheduling algorithms based on carbon emission predictions
  • VM placement optimization using power mix and time-dependent carbon intensity
  • Edge-cloud collaborative environments for carbon-performance trade-off optimization
  • Real-time monitoring and reduction techniques for carbon footprints during AI model training and inference

Improving your academic talk

데이터 사이언스 실험실의 주요 목적 중의 하나는 지식과 아이디어를 전달하기 위한 여러분의 academic talk 활동을 향상시키는 것입니다. 국내외 전문가를 위한 Academic talk는 다음과 같은 요소를 시간과 장소에 따라 적절하게 포함되어야 합니다.

  • Title: This is the first item somebody will check if your talk is worth their while. A title should be short and still capture the essence of your presentation. Detailed presentation can be expressed in a subtitle.
  • Abstract: People interested in your title will read your abstract to figure out if your talk is worthwhile. Therefore the abstract needs to be well crafted, interesting and to the point. It should include the following information:
    1. Introduction: What is the underlying problem? Characterize the state of the art.
    2. Motivation: What are the challenges? What makes it worthwhile for people to listen to you?
    3. Sketch Results: What is the core of your solution idea? Quantitative results?
    4. Conclusion: What is your conclusion? (Is the problem solved?)
  • Your biography: potential listeners will want to know if the person is experienced in the field and has something interesting to say. Your biography should capture this information.

Source: Stepan Decker의 How to announce academic talk

What is hot topics in Data science

  • the whole lifecycle of the data from the past to the present and the future
  • the analytics from explicit (known) analytics and reactive understanding to implicit (unknown) analytics and proactive early prediction and intervention
  • the journey from data exploration (by descriptive and predictive analytics) to the delivery of actionable insights and decisions through prescriptive analytics and actionable knowledge delivery

Source code

This site is based on the Bedford lab website, which makes available the codes from their site in GitHub. In that site there is a good description of how it works and a list of derived sites, such as ours.