Open thesis topics:
Topic 1: Conceptual model of platform for digital asset management to enable cross-company collaborative digital twin development for transport infrastructure management
Background
In the transport sector, implementing digital twins is part of the digitization measure to improve resource efficiency in infrastructure management. However, the use of digital twins is still limited due to challenges such as the lack of a shared understanding of digital twin models, complex model integration, security issues, lack of access to essential data, and high costs due to inefficient business models. A platform for asset management platform will enable cross-company collaborative development and use of digital twins (DC). The platform aims to reduce the cost and effort of creating and using digital twins that improve transparency in the planning and operation of transport infrastructures.
Research Questions
- What transport infrastructure management scenarios require digital twin solutions?
- What are the essential components of digital asset management platform for digital twin?
- Which assets and data sources are required?
Tasks
- Analysis of existing digital twin platform solutions: literature review
- Analysis of typical transport infrastructure management scenarios including actors, use-cases, digital asset and data requirements: literature review
- Identification of essential components of the platform
- Development of the conceptual model
- Validation with experts
References
BMVI (2021) Masterplan BIM Bundesfernstraßen – Digitalisierung des Planens, Bauens, Erhaltens und Betreibens im Bundesfernstraßen-bau mit der Methode Building Information Modeling (BIM) [online].
https://aws.amazon.com/de/iot-twinmaker/
https://docs.microsoft.com/en-us/azure/digital-twins/overview
https://www.bentley.com/en/products/product-line/digital-twins/itwin
https://vertexvis.com/products/vertex-digital-twin-platform
Topic 2: Analysis and development of a standardized asset model for digital twin for transport infrastructure management
Background
The introduction of DZ in companies in Germany, Austria and Switzerland is still limited. Although 54% of companies have an end-to-end strategy for DC, only 8% of companies are already fully utilizing their DC. The central challenges are understanding the data and integrating models of the digital assets (DA) of individual DCs in a simulation of the overall system. For SMEs, the questions and uncertainties lie in the business models, costs, security aspects and access to essential data/information, among other things. Therefore, having a standardized information model expressing the standard terms, taxonomy, and relations of assets is essential.
Research Questions
- What transport infrastructure management scenarios require digital twin solutions?
- Which assets and data sources are required?
- What standardized information model in transport infrastructure management exists?
- How do we express standardized terms, taxonomy, and relations the best?
Tasks
- Analysis of typical transport infrastructure management scenarios including actors, use-cases, digital asset and data requirements: literature review
- Analysis of existing standards for transport infrastructure management: literature review
- Development of asset taxonomy for digital twin for transport infrastructure management
- Formalization in OWL/RDF
- Validation
References
https://learn.microsoft.com/en-us/azure/digital-twins/concepts ontologies
https://protege.stanford.edu/publications/ontology_development/ontology101.pdf
https://www.iaarc.org/publications/fulltext/ISARC_2019_Paper_117.pdf
Topic 3: Object detection and semantic annotation from 3D images of digital twin assets for transport infrastructure management using machine learning
Background
One of the typical data formats of digital twin assets is 3D images/drawings, e.g. car, bridge, and train models. These 3D images have to be integrated into the digital twin platform by annotating and linking the 3d images to the semantic structure. Thus, objects in the 3D images have to be automatically detected based on labels/terms defined in the semantic model/ontology.
Research Questions
- What object detection method is the best to detect objects from 3D drawings?
- How can we integrate the detected object into the digital twin of transport management ontology?
Tasks
- Finding the ontology (creating a simple ontology) as the source of labelling
- Image/drawing selection and image engineering/manipulation to enrich the dataset
- Labelling
- Feature engineering
- Object detection using e.g. YOLO
- Validation
Topic 4: Development of metrics to measure the explainability of machine learning models for digital twin
Background
Extracting explainable insights from the data managed by digital twins will improve the trust of the full spectrum of stakeholders in transport infrastructure management. This can be achieved through explainable AI (XAI). However, a metric is required to assess the quality and degree of the explainability of XAI.
Research Questions
- What criteria are required to assess the explainability of machine learning?
- Are the criteria the same for all stakeholders?
- How can we calculate the score of explainability based on the identified criteria?
Tasks
- Identification of stakeholders in transport infrastructure management and their level of knowledge about AI
- Literature review on explainability measurement/assessment
- Selection of measurement/assessment criteria
- Development of calculation model
- Validation
References
https://arxiv.org/abs/1812.04608
Topic 5: Deep Reinforcement Learning for industrial microgrid management
Background
The thesis will build on top of previous thesis work where a reinforcement environment was built for microgrid with an independent supply and demand infrastructure. The microgrid in the previous work was managed independently to meet local demand. In this thesis the microgrid will also be connected to the main electricity supply grid with through which it can continuously buys or sells energy on the electricity markets.
Research Questions
- How to build a such Reinforcement learning environment in python?
- What relevant Packages to use?
- What features and models are relevant to model different aspects of the environment?
- What will be the goal of the system?
- What will be the reward?
- Is the reinforcement architecture better in predicting Short term forecasts or long term? Compared to what models?
Tasks
- Systematic and comparative literature review on the application of Reinforcement learning: Identification of research that has implemented reinforcement learning for micro grid management.
- Identification of factors influencing the learning, predicting and forecasting strength of the architecture.
- Building Reinforcement learning environment in python. Use of Keras and TensorFlow is preferred.
References
Taha Abdelhalim Nakabi, Pekka Toivanen,Deep reinforcement learning for energy management in a microgrid with flexible demand,Sustainable Energy, Grids and Networks,Volume 25,2021, 100413,ISSN2352-4677,https://doi.org/10.1016/j.segan.2020.100413.
https://paperswithcode.com/paper/deep-reinforcement-learning-for-time-series
José R. Vázquez-Canteli, Zoltán Nagy, Reinforcement learning for demand response: A review of algorithms and modeling techniques, Applied Energy, Volume 235, 2019, Pages 1072-1089, ISSN 0306-2619, https://doi.org/10.1016/j.apenergy.2018.11.002.
Contact
Atit Bashyal
a.bashyal@jacobs-university.de
Topic 6: Deep Reinforcement Learning for Industrial task scheduling
Background
The thesis will explore reinforcement learning environment where the task of the RL agent will be to optimally schedule tasks related to industrial operations. The agents aim will be to minimize electricity cost while optimizing the schedule.
Research Questions
- How to build a such Reinforcement learning environment in python?
- What relevant Packages to use?
- What features and models are relevant to model different aspects of the environment?
- What will be the goal of the system?
- What will be the reward?
- Is the reinforcement architecture better in scheduling in the Short term forecasts or long term? Compared to what models?
Tasks
- Systematic and comparative literature review on the application of Reinforcement learning: Identification of research that has implemented similar tasks
- Identification of factors influencing the learning, predicting and forecasting strength of the architecture.
- Building Reinforcement learning environment in python. Use of Keras and TensorFlow is preferred.
References
Taha Abdelhalim Nakabi, Pekka Toivanen,Deep reinforcement learning for energy management in a microgrid with flexible demand,Sustainable Energy, Grids and Networks,Volume 25,2021, 100413,ISSN2352-4677,https://doi.org/10.1016/j.segan.2020.100413.
https://paperswithcode.com/paper/deep-reinforcement-learning-for-time-series
https://towardsdatascience.com/reinforcement-learning-for-production-scheduling-809db6923419
Contact
Atit Bashyal
a.bashyal@jacobs-university.de
Topic 7: Autoregressive models in Tensorflow
Background
The thesis will explore various Autoregressive models and Their implementation in Tensorflow.
Research Questions
- What are Autoregressive models and how do they work?
- What relevant Packages are available for Autoregressive models?
- Are there any implementations in Ternsorflow ? How to implement Autoregressive models in Tensorflow ?
Tasks
- Systematic and comparative literature review of Autoregressive models and their application in energy forcast. Identification of factors influencing the learning, predicting and forecasting strength of these models
- Exploring Autoregressive models present in Tensorflow and building autoregressive models in TensorFlow.
References
https://www.tensorflow.org/probability/api_docs/python/tfp/sts/Autoregressive
https://www.georgeho.org/deep-autoregressive-models/
https://ml.berkeley.edu/blog/posts/AR_intro/
Contact
Atit Bashyal
a.bashyal@jacobs-university.de
Topic 8: Implementing Explainable DL models for sentiment analysis in the Food Delivery Services
Research Questions
How to apply Explainable Deep Learning models for Sentiment Analysis in the Food Delivery Services ?
Tasks
- Literature Review about Explainable Deep Learning and Sentiment Analysis
- Exploratory Food Delivery Data Analysis
- Model Comparison between Conventional ML/AI and Explainable AI or Interpretable ML
- Result Analysis
References
Adak, A., Pradhan, B., & Shukla, N. (2022). Sentiment Analysis of Customer Reviews of Food Delivery Services Using Deep Learning and Explainable Artificial Intelligence: Systematic Review. Foods, 11(10), 1500.
Drus, Z.; Khalid, H. Sentiment Analysis in Social Media and Its Application: Systematic Literature Review.Procedia Comput. Sci.2019,161, 707–714
Johnson, R.; Zhang, T. Effective Use of Word Order for Text Categorization with Convolutional Neural Networks.arXiv 2014,arXiv:1412.1058.
Moreno Lopez, M.; Kalita, J. Deep Learning Applied to NLP.arXiv2017,arXiv:1703.03091.
Pang, B.; Lee, L.; Vaithyanathan, S. Thumbs Up? Sentiment Classification Using Machine Learning Techniques.arXiv2002,arXiv:cs/0205070.
Contact:
Rahmat Hidayat
r.hidayat@jacobs-university.de
Topic 9: Anomaly Detection in Building Energy-Saving Information Systems Using Internet of Things and Explainable Agent-Based Algorithms
Research Questions
How to apply Explainable Agent-Based Algorithms in Building Energy Saving Information System?
Tasks
- Literature Review
- Exploratory Data Analysis
- Model Comparison
- Result Analysis
References
Forestiero, A. Metaheuristic Algorithm for Anomaly Detection in Internet of Things leveraging on a Neural-driven Multiagent System. Knowl.-Based Syst.2021,228, 107241.
Forestiero, A.; Mastroianni, C.; Spezzano, G. Reorganization and Discovery of Grid Information with Epidemic Tuning. Future Gener. Comput. Syst. 2008,24, 788–797
Forestiero, A.; Papuzzo, G. Agents-Based Algorithm for a Distributed Information System in Internet of Things.IEEE Internet Things J.2021,8,
Contact:
Rahmat Hidayat
r.hidayat@jacobs-university.de
Topic 10: Ontology and Knowledge Graph Development for Sustainability/ESG* Performance Measurement in Manufacturing
*ESG: Environment, Social, Governance
Background
The complexity of the sustainability aspects in manufacturing has grown that include factors in the environment, social, and governance (ESG). As a comprehensive sustainability report is mandatory and must meet certain minimum requirements, failure to maintain sustainability aspects can lead to potential issues, e.g. failure of risk management and incompatibility with regulations that fail business activities.
Research Questions
- How are current sustainability performances measured in the manufacturing industry?
- How to build an Ontology and Knowledge Graph (KG) to help manage ESG performance measurement and solve interoperability issues in manufacturing?
Tasks
The whole tasks are estimated will be finished in March 2023, that consists of:
- Develop ontology and KG related to ESG performance measurement.
- Utilize and analyze current sustainability performance reports and extract information using NLP and Deep learning. The candidate tool for text extraction is already available.
- Map and align extracted information to ontology and KG.
- Put data from data sources to KG.
- Visualize knowledge graph results, e.g. using Neo4j, Stardog, etc.
The work progress will manage by using a management project tool named JIRA.
References
https://www.sciencedirect.com/science/article/pii/S0167739X2033003X?via%3Dihub
Contact:
Annas Vijaya
a.vijaya@jacobs-university.de
Topic 11: Analyzing causal machine learning trends in manufacturing
Research Questions and Tasks
- Literature analysis of where and why causal machine learning is applied (companies), finding a case study for deeper analyzes
- Study causal machine learning algorithm implementation environments like dowhy or causalml, compare them (theoretically)
- Suggest one of the environments, and using the case study, implement the searched algorithms
- Using the case study, introduce the performance of the algorithms, and suggest one of the algorithms (if it is possible)
- This topic is better for one interested in programming in Python or any other language (JavaScript, Go…)
- This topic is feasible for Master’s students
Contact:
Tamas Fekete
t.fekete@jacobs-university.de
Topic 12: Knowledge graph building for artificial intelligence for better decisions in manufacturing
Research Questions and Tasks
- What are the knowledge graphs, and how do you construct them?
- Search/introduce companies where the knowledge graphs are applied (main focus is the automotive industry)
- How can knowledge graphs and AI be used together?
- Chose a (simple) problem for a case study (earlier points can help)
- Search and compare AI algorithms/techniques focusing on the earlier introduced case study/problem
- Formulate suggestions on which algorithm/technique can use to solve the selected case study better, design the way
- This topic is based on literature research and design
- Programming is not required (kind of black box)
- This topic is feasible for Bachelor students
Contact:
Tamas Fekete
t.fekete@jacobs-university.de
Topic 13: Creating an ontology from the different data formats and comparing with other tools
Background
Many approaches create an ontology model from a relational database model and other data formats like CSV and JSON and migrate the database’s contents to the generated ontology. They require mappings between each created ontological component (concept, property, etc.) and its original database component (table, column, etc.).
Research Questions
- Which language is more flexible and suitable to express ontology-to-data mappings formally?
- What tools are needed for mapping approaches to create a new ontology from the different data formats?
- Can you develop the presented tools to create an ontology from several data sources?
- What algorithms can be applied to implement the approaches?
- How to improve the existing data-to-ontology mapping approaches?
- Can you present an approach to get a full automatic mapping process?
Tasks
- Discovery of additional semantic relations between data components
- Constructing ontology concepts and relations between them
- Mapping definition- the transformation of data schema into ontology structure
- Data migration- the migration of data contents into ontology instances (ontology population)
Tools: Owlready2
Programming Language: Python
References
[1] Ghawi R., Cullot N., (2007) Database-to-ontology mapping generation for semantic interoperability, Conference: Third International Workshop on Database Interoperability (InterDB 2007), held in conjunction with VLDB 2007, Vienna, Austria.
[2] Munir K., Odeh M., Mc clatchey R H., (2009) Managing the mappings between domain ontologies and database schemas when formulating relational queries, IDEAS ’09: Proceedings of the 2009 International Database Engineering & Applications SymposiumSeptember Pages 131– 141 https://doi.org/ 10.1145/1620432.1620446
[3] Shah, T., Rabhi, F. & Ray, P. (2015) Investigating an ontology-based approach for Big Data analysis of inter-dependent medical and oral health conditions. Cluster Compute 18, 351–367. https://doi.org/10.1007/s10586-014-0406-8
Contact:
Tina Boroukhian
t.boroukhian@jacobs-university.de
Topic 14: Comparing and analysing different ontology mapping tools
Background
Ontologies are a convenient way for knowledge representation. When ontologies are involved in system integration, it requires communication between two or more ontologies. Data to-ontology Mapping is the process whereby data and ontology are semantically related at a conceptual level, i.e. correspondences are established between the data source components and the ontology components. There are some tools for different mapping formats of data into ontologies. The goal here is to compare and evaluate the existing mapping tools.
Research Questions
- What ontology mapping tools exist?
- How do these tools perform?
- What methods were these tools used?
Tasks
- Finding all the mapping tools that are exist.
- Comparing and analysing the mapping tools.
- Determining the techniques that these tools are used.
References
[1] https://link.springer.com/chapter/10.1007/978-3-319-25639-9_24.
[2] https://arxiv.org/pdf/1906.08092.pdf.
Contact:
Tina Boroukhian
t.boroukhian@jacobs-university.de
Topic 15: Causal machine learning/inference to identify the causality of factors driving people doing voluntary carbon offsetting.
Background
Voluntary carbon offsetting is an approach to compensate activities that cause intensive greenhouse gas emissions and cannot be avoided, for example, intercontinental flying, package shipping, moving trucks, etc. The compensations are done by offsetting the carbon emission to fund projects that reduce greenhouse gas emissions, for example, installation of renewable energy power plants, electrification of transportation, the introduction of efficient cookstoves, etc. Therefore, a global net-zero emission can be achieved. This thesis focuses on empirical research and should apply causal machine learning which is a novel approach in AI.
Research Questions
- What factors contribute to people who want to do voluntary carbon offsetting for their activities?
- How are the causal relationships among those factors based on collected dataset?
Tasks
- Identify potential factors in the literatures that drive the intentions of people to contribute to global net-zero emission
- Develop hypotheses and formalize causal relationship e.g. using causal graph.
- Design a survey to test the hypotheses
- Collect the data through a survey
- Causal machine learning to analyze the collected data using tools such as DoWhy, ShowWhy
- Description and discussion of the results
References
https://ideas.repec.org/a/eee/touman/v45y2014icp194-198.html
https://www.sciencedirect.com/science/article/pii/S0261517720302028
https://www.sciencedirect.com/science/article/pii/S0959652621031206 (analysis of survey data)
https://arxiv.org/abs/2206.15475 (causal machine learning)
Topic 16: Causal machine learning/inference to identify the causality of factors driving people shopping online of products having social -cultural values
Background
eCommerce customers intend to purchase products not only because of price, quality, and environment-friendliness but also because of the social-cultural values of the products. For example, a customer wants to purchase products that can contribute to the operation of an orphanage or products that are made by a rural poor community from a certain country with a particular culture. However, since the products are purchased online, it is difficult for the customers to experience the social-cultural values of the products directly. This thesis focuses on empirical research on how social-cultural values and perceptions influence purchasing behaviors of eCommerce customers. The thesis should apply causal machine learning which is a novel approach in AI.
Research Questions
- What social-cultural factors of products contribute to people purchasing online?
- How are the causal relationships among those factors based on collected dataset?
- How to measure social-cultural value of an eCommerce product?
Tasks
- Identify potential social-cultural factors on products that drive the intentions of people to purchase online through literature review
- Develop hypotheses by correlating among the identified factors. The correlations can be positive or negative
- Design a survey to test the hypotheses
- Collect the data through a survey
- Causal machine learning to analyze the collected data using tools such as DoWhy, ShowWhy
- Development of social-cultural score based on identified factors
- Description and discussion of the results
References
https://www.sciencedirect.com/science/article/pii/S0959652621012129
http://www.journals.researchub.org/index.php/JMAS/article/view/351
https://www.sciencedirect.com/science/article/pii/S0959652621031206 (analysis of survey data)
https://arxiv.org/abs/2206.15475 (causal machine learning)
Topic 17: The Role of Causal Machine Learning (CML) to Improve Decision Making in Supply Chain Management
Background
CML is an emerging and growing field in applying AI in business. Causal inference in machine learning is a statistical tool that enables AI and machine learning algorithms to reason based on cause and effect. Understanding cause and effect would make existing AI systems more intelligent and more efficient. Moreover, it can help organizations understand their data. Some possible scenarios that may apply CML [1]:
#1: In an e-commerce context, we could determine which specific factor impacts the decision to purchase a product most. We could better allocate resources to improve a particular KPI with this information. We could also rank the impact of different factors on the purchasing decision. We could determine if a given customer would have purchased a specific product if he/she had not bought other products for the last two years.
#2: In a broader sense, we could discover how and what a given business strategy could have avoided negative impacts? We could also determine how much we expect our sales to increase by implementing a specific training program to our business developers. the impact of a specific training program
#3: In the agricultural field, we often try to predict if a farmer’s crop yield will be lower this year. However, using casual inference, it will become to better understand what steps should we take to increase the harvest.
Research Questions
- What scenarios in supply chain management may apply CML?
- Which CML approach and tool is the most appropriate to improve supply chain processes?
- How to measure the most appropriate approach?
- What improvements in terms of decision making can be achieved?
Tasks
- Define causality scenarios in supply chain contexts
- Find an appropriate dataset
- Literature review on related work, different approaches of causal inference (Uplift Random Forests, Meta Learners, SEM, Bayesian Network, etc.), and tools (CausalML [3], CausalNex [4])
- Data preprocessing and exploratory analysis
- Model implementation using chosen tools
- Evaluation of the results
- Discussions and documentation
Starting Literature
[1] https://towardsdatascience.com/introduction-to-causality-in-machine-learning-4cee9467f06f
[2] Dataset examples: https://github.com/rguo12/awesome-causality-data
[3] https://causalml.readthedocs.io/en/latest/about.html
[4] https://causalnex.readthedocs.io/en/latest/01_introduction/01_introduction.html#
[5] https://innovationatwork.ieee.org/what-is-causal-inference-and-why-is-it-key-to-machine-learning/
[6] https://www.sciencedirect.com/science/article/pii/S0278431921000578
Topic 18: Explainable Artificial Intelligence (XAI) Implementation in prediction of crop yield, soil properties, and irrigation requirements.
Research Questions
How to apply multiple XAI methods in predicting crop yield, soil properties, and irrigation requirement
Task Sequences:
- Literature Review
- Exploratory Data Analysis
- Model Comparison
- Result Analysis
“Ben Ayed et al. [16] analyzed 18 worldwide table olive cultivars by using morphological, biological, and physicochemical parameters and the Bayesian network to study the influence of these parameters intolerance, productivity, and oil content. They revealed that oil content was highly influenced by the tolerance of the crop.”
References
Ben Ayed, R., & Hanana, M. (2021). Artificial Intelligence to Improve the Food and Agriculture Sector. Journal of Food Quality, 2021.
Ayed, R. B., Ennouri, K., Amar, F. B., Moreau, F., Triki, M. A., & Rebai, A. (2017). Bayesian and phylogenic approaches for studying relationships among table olive cultivars. Biochemical genetics, 55(4), 300-313.
Elavarasan, D., Vincent, D. R., Sharma, V., Zomaya, A. Y., & Srinivasan, K. (2018). Forecasting yield by integrating agrarian factors and machine learning models: A survey. Computers and Electronics in Agriculture, 155, 257-282.
Zhang, C., Liu, J., Shang, J., & Cai, H. (2018). Capability of crop water content for revealing variability of winter wheat grain yield and soil moisture under limited irrigation. Science of the Total Environment, 631, 677-687.
Contact:
Rahmat Hidayat
r.hidayat@jacobs-university.de
Topic 19: Ontology Alignment using Multiple Similarity Measures
Background
Ontologies are convenient way for knowledge representation. When ontologies are involved in system integration, it requires communication between two or more ontologies. Ontology alignment (a process to generate a set of correspondences between similar entities of various ontologies) can be described as follows: given two ontologies, each describing a set of discrete entities (which can be classes, properties, etc.), find the similarities that hold between these entities. For automatic ontology matching, we need to combine some kinds of different similarity measures: lexical-based, Language-based, structure-based, and semantic-based techniques as well as using information in ontologies including names, labels, comments, relations and positions of concepts in the hierarchy and integrating WordNet dictionary.
Research Questions
- Which similarity measures approaches are more suitable?
- Which features are needed to extract from the ontologies for implementing the similarities?
- Which libraries are more suitable for this task?
- Can you develop the presented similarities to reach to the better F-measure?
- Which approaches is more flexible to compute the similarities for large ontologies?
Tasks
- Studying and comparing different similarity measures approaches
- Finding and extracting suitable features from ontologies
- Implementing different similarity measures (lexical-based, Language-based, structure-based, and
semantic-based techniques) - Combining different similarity measures techniques as a final similarity
Programming Language: Python
Library: owlready2, BeautifulSoup, nltk
Starting References
[1] Nguyen, Thi Thuy Anh, and Stefan Conrad. “Ontology matching using multiple similarity measures.” 2015 7th International Joint Conference on Knowledge Discovery, Knowledge Engineering and Knowledge Management (IC3K). Vol. 1. IEEE, 2015.
[2] Nguyen, Thi Thuy Anh, and Stefan Conrad. “Combination of lexical and structure-based similarity measures to match ontologies automatically.” International Joint Conference on Knowledge Discovery, Knowledge Engineering, and Knowledge Management. Springer, Berlin, Heidelberg, 2012.
[3] Bulygin, Lev, and Sergey A. Stupnikov. “Applying of Machine Learning Techniques to Combine Stringbased, Language-based and Structure-based Similarity Measures for Ontology Matching.” DAMDID/RCDL. 2019.
Contact:
Tina Boroukhian
t.boroukhian@jacobs-university.de
Topic 20: Mapping Data (JSON, CSV) to Existing Ontologies and populate the ontologies by the Data contents using Python Libraries
Background
Ontologies are convenient way for knowledge representation. When ontologies are involved in system integration, it requires communication between two or more ontologies. Database-to-Ontology Mapping is the process whereby a database and an ontology are semantically related at a conceptual level, i.e. correspondences are established between the database components and the ontology components. For automatic mapping data to ontologies, we need to find the related components between ontology and our data and applying the mapping between them.
Research Questions
- Which ontologies are related to our data? What are the ontology domains for our data?
- Which libraries are more suitable for this task?
- Which approaches is more flexible to map the big data to ontologies?
- How can you improve the program for running the big amount of data?
Tasks
- Studying and comparing different python libraries for purpose ontology
- Finding the related components between ontology and data
- Reading JSON file from the Api and mapping it to related ontologies
- Populating the ontologies by our data contents
- Saving the mapping file into the Neo4j
Programming Language: Python
Library: owlready2, RDFLib
Starting References
[1] Munir K., Anjum M. S., (2018) The use of ontologies for effective knowledge modelling and information
retrieval, Applied Computing and Informatics , 14(2), 116-126.
[2] Shah, T., Rabhi, F. & Ray, P. (2015) Investigating an ontology-based approach for Big Data analysis of
inter-dependent medical and oral health conditions. Cluster Compute 18, 351–367 (2015). https://doi.org/
10.1007/s10586-014-0406-8.
[3] https://owlready2.readthedocs.io/en/latest/onto.html.
[4] https://neo4j.com/labs/neosemantics/4.0/introduction/
Contact:
Tina Boroukhian
t.boroukhian@jacobs-university.de
Topic 21: Ontology Alignment using Machine Learning and Neural Networks Techniques
Background
Ontologies are convenient way for knowledge representation. When ontologies are involved in system integration, it requires communication between two or more ontologies. Ontology alignment is a process of establishing correspondences between semantically related entities in different ontologies. Applying machine learning between ontology elements as features for ontology matching. Logistic Regression, Random Forest classifier and Gradient Boosting are used as machine learning methods.
Research Questions
- Which Machine Learning and neural networks Techniques are more suitable?
- Which features are needed to alignment the ontologies?
- Which libraries are more suitable for this task?
- Can you develop the presented similarities to reach to the better F-measure?
- Which approaches is more flexible to compute the alignment for large ontologies?
Tasks
- Studying and comparing different machine learning and neural network techniques
- Creating predicted alignment from two lists of entities
- Applying machine learning methods to existing similarity measures and creating dataset and training a machine learning model
Programming Language: Python
Library: scikit-learn, owlready2, BeautifulSoup
Starting References
[1] Bulygin, Lev, and Sergey A. Stupnikov. “Applying of Machine Learning Techniques to Combine Stringbased,
Language-based and Structure-based Similarity Measures for Ontology Matching.” DAMDID/RCDL.
2019.
[2] Nezhadi, Azadeh Haratian, Bita Shadgar, and Alireza Osareh. “Ontology alignment using machine
learning techniques.” International Journal of Computer Science & Information Technology 3.2 (2011): 139.
[3] Bento, Alexandre, Amal Zouaq, and Michel Gagnon. “Ontology matching using convolutional neural
networks.” Proceedings of the 12th language resources and evaluation conference. 2020.
Contact:
Atit Bashyal
a.bashyal@jacobs-university.de
Topic 22: Querying Ontologies using NLP: A Natural Language based Query Interface to OWL Ontologies for Demand-Response System
Background
Ontologies are convenient way for knowledge representation. Accessing structured data in the form of ontologies requires training and learning formal query languages (e.g., Cypher or SPARQL) which poses significant difficulties for non-expert users. One of the ways to lower the learning overhead and make ontology queries more straightforward is through a Natural Language Interface In this project we need to Implement the Question-based Interface to Ontologies – a tool for querying ontologies.
Research Questions
- How we can extracting domain knowledge automatically?
- How we can translating the text query to formal query language, e.g. Cypher or SPARQL?
- Which libraries are more suitable for this task?
Tasks
- Extracting domain knowledge automatically
- Extracting a text query as an input and transforms it to the set of queries expressed in a formal language, e.g. Cypher or SPARQL
- Connecting to Neo4j and querying ontology based on formal language, e.g. Cypher or SPARQL
Programming language and libraries
- Python
- nltk, owlready2
Starting References
[1] Damljanovic, Danica, Valentin Tablan, and Kalina Bontcheva. “A Text-based Query Interface to OWL Ontologies.” LREC. 2008.
[2] Bernstein, Abraham, et al. “Querying ontologies: A controlled english interface for end-users.” International Semantic Web Conference. Springer, Berlin, Heidelberg, 2005.
Contact:
Tina Boroukhian
t.boroukhian@jacobs-university.de
Topic 23: How artificial intelligence can improve the management of large projects? A causal machine learning approach
Background
Even though the impact of artificial intelligence, predictive analytics and machine learning has proved to be significant in different areas of automation and engineering, there is a lack of research on their effect on project management.
Research Questions
- How artificial intelligence can improve the management of large projects?
- What aspects of project management can be improved through AI?
- Which factors influence the willingness of projects/project managers to adopt AI?
- How are the causal relationships among those factors based on collected dataset?
Tasks
- Systematic and comparative literature review on the applications of AI and machine learning in project management
- Identification of aspects/factors in project management that can be improved through AI.
- Identification of factors influencing the willingness of project managers/teams to adopt AI
- Development of causal graph based on the identified factors
- Survey
- Causal machine learning to analyze the collected data using tools such as DoWhy, ShowWhy
Starting References
https://medium.com/the-project-office/artificial-intelligence-in-project-management-68ddf2ad91d7
https://www.liquidplanner.com/blog/seven-future-trends-in-project-management/
Topic 24: Managing interoperability in circular economy
Background
The circular economy concept involves collaborations of different human actors and information systems throughout the supply chain. Those different actors and systems need to share reliable data and innovative solutions with each other to improve decision makings and process efficiency. In other domains like healthcare [1], and inter-organization collaboration, ontologies and standards are used to define common understanding on a semantical level. Those ontologies and standards provide agreed vocabularies that represent entities, processes, and resources in the whole system
Research Questions
- What are typical actors, systems, and solutions involved in the circular economy?
- What vocabularies and relation models are required to facilitate the interoperability of those entities?
Tasks
- Systematic and comparative literature review
- Identification of actors, systems, and exchanged information and solutions between actors/systems
- Identification of common vocabularies to facilitate interoperability
- Development of relation model representing the vocabularies and their relations
Starting References
[1] https://www.sciencedirect.com/science/article/pii/S1877050912004024
[2] https://link.springer.com/chapter/10.1007/978-3-642-15961-9_86
[3] http://ceur-ws.org/Vol-2044/paper10/paper10.html
Topic 25: The impacts of dynamic electricity price on production planning and control
Background
The introduction of dynamic electricity prices in the manufacturing (energy consumer) requires the production planners to consider the price fluctuation in their production planning and controls. The production processes have to be shifted to the time points where the electricity price is low.
Research questions
- What activities in production planning and control will be influenced by the dynamic electricity price?
- What objectives, variables, and constraints are required to be changed compared to conventional production planning and control?
- How can production planning and control be optimized?
Tasks
- Systematic and comparative literature review
- Analysis of different dynamic price models and their impacts on the planning time horizon
- Identification objectives, variables, and constraints for the optimization
- Development of optimization model
- Solving the optimization model using tools, eg. using linear programming, ant colony optimization, etc.
Starting References
https://www.sciencedirect.com/science/article/pii/S2212827116001281
https://publikationen.bibliothek.kit.edu/1000060882
Topic 26: Energy consumption and cost prediction of customized products using data analytics or machine learning
Background
To improve the competitiveness in the market, companies transform their business by providing customized products. A stainless steel manufacturing SME, who is currently the leader in a niche market of stainless steel application in the oil and gas industry, provides customized steel products that allow customers to configure the material, size, shape, heat treatment, etc. However, it leads to challenges in estimating the price and energy consumption of the customer configured products individually.
Research Questions
- What variables influence the energy consumption and costs?
- Which data analytics/machine learning methods can be applied to estimate the energy consumption and costs? Which method is the best?
- What would be the actions to save the energy consumption and costs after predicting them?
Tasks
- Comparative literature review
- Data collection, pre-processing, exploratory data analytics
- Building machine learning models
- Evaluation of the models
- Critical analysis of results and actions required to achieve energy and cost-saving
Starting References
https://publikationen.bibliothek.kit.edu/1000060882
Topic 27: How does COVID-19 give impact to the supply chain of different products (e.g. hygiene, electronics/IT, typical product )? A data analytics approach
Background
The global COVID-19 pandemic has a strong impact to world economy due to the limitation of supply chain activities. The supply chain of certain types of products are negatively affected but not for some other types of product. This thesis focuses on the analysis of the supply chain of different types of products during the global pandemic by analyzing the available data.
Research Questions
- What types of products are strongly affected by the COVID-19 global pandemic?
- How the pandemic affects the supply chain of those products?
- What model can be used to describe the correlations of the pandemic to the supply chain. (note: focus on the different time series models e.g. Holt-Winters, ARIMA, SARIMA, VAR, etc. or regression models, e.g. linear, polynomial, random forest, ridge, etc.)
Tasks
- Comparative literature review to analyze related works
- Identification of supply chain scenarios affected by COVID-19 pandemic
- Data collection
- Exploratory data analysis
- Building the time series and regression models
- Evaluation of the results
Starting References
https://www.ncbi.nlm.nih.gov/pmc/articles/PMC7413852/
Topic 28: The roles of digital twins in sustainable supply chain
A digital twin can be used as a virtual supply chain replica that consists of hundreds of assets, warehouses, logistics and inventory positions [1]. It is gaining more attention in the industry due to improvements in technical and computational capabilities with operations technology.
Research Questions
- What factors make a supply chain more sustainable?
- What are the requirements for applying digital twins to a supply chain?
- What scenarios in a supply chain can be improved through digital twins?
Tasks
- Comparative literature review on sustainable supply chains and digital twins
- Identification of essential factors in applying digital twins in supply chains
- Development of general sustainable supply chain scenarios
- Two possible methodology
- Cross case analysis: combining literature studies and interview on different supply chain scenarios
- Surveys
- Analysis of literature studies + interview or survey results
Starting References
[1] https://www.supplychaindigital.com/technology/evolution-digital-twins-supply-chain
[2] https://link.springer.com/article/10.1007/s11036-020-01557-9
Other topics (self-defined)
Industry 4.0
IC01. Platform for inter-organizational data sharing and data trading for industry 4.0
IC02. Platform for digital twin management and digital asset sharing
IC03. Industry 4.0 meets circular economy
IC04. Industry 4.0 meets sharing economy
IC05. Open Innovation Platform for industry
IC06. Industry 4.0 in agriculture – precision farming
Green Manufacturing and Circular Economy
GC01. The role of IoT and big data to green manufacturing
GC02. Materials lifecycle analysis methodology for the circular economy
GC03. Holistic energy-efficient manufacturing system management
GC04. Digital twin applications in sustainable manufacturing
Construction 4.0 and Smart Building
CB01. Industry 4.0 in construction industry (self-defined topic)
CB02. The role of digitization in building retrofitting
CB03. Smart operation of proactive residential buildings needs – challenges in control technologies, predictive maintenance, and data supply for the customer
CB04. Construction by Digital-Twin Reconstruction.
Student Theses
| Name | Topic | Year | Thesis Category |
|---|---|---|---|
| Edmundo Cuadra | Data Analytics Approach for Penetrating the Aircraft Manufacturing Market Duopoly | 2021 | B.Sc. IEM |
| Egor Sarafanov | Artificial Intelligence in Project Management: What factors influence the level of AI adaptation? | 2021 | B.Sc. IEM |
| Ishansh Gupta | Enhancing Football Analytics using Data Science | 2021 | M.Sc. DE |
| Korin Hoxha | Towards the Construction of a Supply Chain Management Digital Twin using Data Management and Data Analytics | 2021 | B.Sc. IEM |
| Mohid Qaiser | The effect of dynamic electricity price on production planning and control | 2021 | B.Sc. IEM |
| Oumeng Xia | How COVID-19 Give Impact to the Supply Chain of Food and Hygiene | 2021 | B.Sc. IEM |
| Pardis Sahraei | Integration of Drones Technology and the Public Transportation System for the Last Mile Delivery by Considering Dynamic Distribution Centers | 2021 | M.Sc. IEM |
| Sebastian Gonzalez | Impact of Machine Learning and Natural Language Processing on Products and their Features | 2021 | B.Sc. IEM |
| Tammy Siyakurima | How can Artificial Intelligence Improve the Management of Large Projects? | 2021 | B.Sc. IEM |
| Temirlan Aikenov | Energy consumption and cost prediction of customized products using data analytics or machine learning | 2021 | B.Sc. IEM |
| Thanh Tuan Nguyen | Handling interoperability in dynamic electricity market | 2021 | B.Sc. IEM |
| Unai Alapont | Novel Smart Contracts Model for the Infrastructure Sector: A Design Science Research Framework | 2021 | B.Sc. IEM |
| Vaibhav Apraj | A Proof of Concept for an incrementally enhanced Data Quality Dashboard | 2021 | M.Sc. DE |
| Yanny Budiaki | Reduction of food loss along the Sub-Saharan African food supply chain: The application of a sharing economy product | 2021 | B.Sc. IEM |
| Andrea Pin Morales | The Environmental Impact of Increased Transparency In Last Mile Delivery Emissions in E commerce A study on the effect of a sustainability driven point system on customer behavior | 2020 | B.Sc. IEM |
| Atit Bashyal | Moving Beyond Maximum Likelihood Estimates Application of Bayesian Inference techniques in Poverty Prediction. | 2020 | M.Sc. DE |
| Daniyar Abdimomunov | Opportunities and Challenges of Data-driven Applications Systematic comparison and validation of academic literature from the perspective of data users | 2020 | B.Sc. IEM |
| Daoyuan Ji | Collaborative Distribution Model between Logistics Service Providers and Public Transportation Operators: Investigation of parcel Movement in the Public Transit Network with Heuristics Approach | 2020 | M.Sc. SCEM |
| Helen Schmitz | How does bonus programs affect the customer behaviour towards sustainable packaging? A survey-based customer behaviour analysis | 2020 | B.Sc. IEM |
| Kataryna Taran | Current state of Facilities Management industry: Development and implementation of new standards, technologies, and strategies around the world | 2020 | M.Sc. SCEM |
| Merint Thomas Mathew | Monitoring Customer Activity and Forecasting Future Deals for improved Customer Success Management | 2020 | M.Sc. DE |
| Nathnael Taye | Energy efficiency and Digitalization: How to improve awareness on energy efficiency in residential buildings using digitalization? | 2020 | B.Sc. IEM |
| Paola Cevallos | Best Practices of Digitalization within Motorsport Events | 2020 | B.Sc. IEM |
| Paolo de Leon | How Industry 4.0 Secures the Future of the Shipbuilding Industry Data Analysis on the implementation of new IoT technologies to the Shipbuilding Industry | 2020 | B.Sc. IEM |
| Peter-Sleiman Mansour | Supplier selection model for sustainable supply chain in low-cost robotics production | 2020 | M.Sc. SCEM |
| Saurav Kedia | Carpooling in Nepal: User Prospective based on Sustainability: A survey-based approach | 2020 | B.Sc. IEM |
| Sebastian Reiser | Dynamic Electricity Pricing Comparative Study of different forecasting methods | 2020 | B.Sc. IEM |
| Surendra Ketkar | Quality Control in Smart Manufacturing for SMEs | 2020 | M.Sc. SCEM |
| Yousuf Farooq | Can Claiming Sustainability harm companies? A survey-based approach | 2020 | B.Sc. IEM |
| Zayed Balout | Industry 4.0: An Empirical and Conceptual Analysis Approach to the Implementation and Progression of Digitalization within the Construction Sector | 2020 | M.Sc. SCEM |
| Amin Houshidari | Big data analytics for forecasting in supply chain | 2019 | M.Sc. SCEM |
| Burulai Abdykapar Kyzy | The role of artificial intelligence, machine learning and predictive analytics in project management | 2019 | M.Sc. SCEM |
| Cedric Ohlms | Data and Interface Management for optimization of change processes | 2019 | B.Sc. IEM |
| Dazhi Zhan | Systematic Comparison of Impact and Readiness of Industry 4.0 Technologies between China and Germany | 2019 | B.Sc. IEM |
| Diego Lainfiesta | Capacity and Risk Planning in a high Mix Product Environment | 2019 | M.Sc. DE |
| Katharina Landers | The Use of Artificial Intelligence and Internet of Things for Anomaly Detection and Predictive Maintenance for Construction Machinery | 2019 | M.Sc. SCEM |
| Mifrah Aftab | Analyzation and Development of Logistics Chain KPIs for Sustainable Ports in Europe. | 2019 | M.Sc. SCEM |
| Milos Delikladic | How Industry 4.0 technologies can be incorporated in future farming? Survey and interview investigation in Serbia | 2019 | B.Sc. IEM |
| Miruthula Ramesh | To what extent can sustainable/green manufacturing benefit from the integration of IoT and Big data? | 2019 | B.Sc. IEM |
| Murtaza Khawaja | The Role of After Sales in an IoT based Startup | 2019 | B.Sc. IEM |
| Osama Bin Waheed | Construction 4.0 and smart building Development of the reference process model in construction 4.0 Case study: Sweden and China | 2019 | B.Sc. IEM |
| Rahul Upadhyay | Industry 4.0 in Farming: Precision Agriculture A case study in Nepal | 2019 | B.Sc. IEM |
| Sharath Abraham Peter | Application of augmented reality technology in the manufacturing process of pressing tools | 2019 | M.Sc. SCEM |
| Tianran Ni | Automated Information System Development for Medium- to Short-term Capacity Planning in MTO Industry | 2019 | B.Sc. IEM |
| Xuqi Bai | Ergonomic Performance Indicator Development for Assembly Systems: Extension of EAWS Method with Assessment of Environmental Factor | 2019 | B.Sc. IEM |
| Ali Naweed | The Role of Digitalization for transformation to Industrie 4.0 oriented Business Models in the German Automotive Industry. | 2018 | B.Sc. IEM |
| Anton Shavtvalishvili | Additive Manufacturing’s impact on supply chain in aerospace industry | 2018 | B.Sc. IEM |
| Hussein Hegazy | Outsourcing and Swap practice in the petroleum industry | 2018 | B.Sc. IEM |
| JongHack Yi | Development of Sales Demand Forecast through process Improvement | 2018 | M.Sc. SCEM |
| Kemal Tumysch | Outsourcing and Enterprise Resource Planning: Challenges and Opportunity in Petroleum Supply Chain | 2018 | B.Sc. IEM |
| Kirsten Mueller | Analysis of NBA supply chain network for its expansion to Europe | 2018 | M.Sc. SCEM |
| Osaid Khan | Assessing the Maturity Level of Industry 4.0 in the Automotive Industry | 2018 | B.Sc. IEM |
| Sergi Drago Gonzalez | Do machine learning models provide accurate macroeconomic GDP forecast? | 2018 | B.Sc. GEM |
| Soman Tariq | Reinventing the Manufacturing Sector of Aerospace Industry by Implementing Industry 4.0 Technologies | 2018 | B.Sc. IEM |
| Waleed Asif | Improving Energy and Resource Efficiency in Different Level of Production Using Industry 4.0 Technologies | 2018 | B.Sc. IEM |
| Young Joon Lee | Product Lifecycle Information Model for Reference Architectural Model Industrie 4.0 (RAMI 4.0) | 2018 | M.Sc. SCEM |
| Akarawint Chawalitanont | Power Consumption Prediction of Customised Stainless Steel Ring Products using Hierarchical Bayesian Models and Artificial Neural Networks | 2022 | M.Sc. DE |
| Nadhir Khoubeieb Mechai | Causal Machine Learning (CML) for Demand Forecasting of Semiconductor Products | 2022 | M.Sc. SCM |
| Xiaotong Fan | Advanced Time Series Analysis for Electric Vehicle Demand Forecasting | 2022 | M.Sc. SCM |
| Seyed Taha Raeisi | Delivery Assurance and Digitalization Utilizing Expert’s Knowledge, Internal Key Performance Indexes and Exogenous Risk Factors to Drive Operational Supply Chain | 2022 | M.Sc. SCM |
| Vamsi Sai Pidikiti | Managing interoperability in the circular economy | 2022 | M.Sc. SCM |
| Muhammad Abdullah Shah | Using Machine Learning to Optimize Dynamic Pricing in Production at Machine Level | 2022 | B.Sc. IEM |
| Nurgun Rafizade, | Exploration of Radial Basis Function Networks and Adaptive Network Based Fuzzy Inference Systems for Time Series Forecasting | 2022 | B.Sc. CS |
| Hamza Hayak | Explainable AI for Crop Yield Prediction | 2022 | B.Sc. CS |
| Toska Ymerhalili | Deep Reinforcement Learning for Industrial Microgrid Management | 2022 | B.Sc. CS |
| Nguyen Duy Bao | Ensemble Learning to forecast Wind/Solar power generation | 2022 | B.Sc. IEM |
| Carlotta Buck | Incentivising Voluntary Carbon Offsetting and its Driving Factors: Integrating Causal and Statistical Data Analysis | 2022 | B.Sc. IEM |
| Abel Teklearegay | Ontology Matching Using Supervised Learning Techniques | 2022 | B.Sc. IEM |
| Christopher Mclaughlan | Querying Ontologies using NLP: A Natural Language based Query Interface to OWL Ontologies for Demand-Response System | 2022 | B.Sc. CS |
| Kaleb Kristo | What social-cultural values of products make people shop online? A causal data analytics method | 2022 | B.Sc. IEM |
| Moonkyoung Choo | Effect of Governmental Policies and Healthcare Systems on COVID-19 Hosptalization: A Machine Learning Prediction Approach | 2022 | B.Sc. IEM |
| Karen Okoli | The Role of AI in Circular Economy: Automotive industry | 2022 | B.Sc. IEM |
| Kezia Adesanya | Mapping JSON data to Existing Ontologies: Populating the ontologies by the Data contents using Python Libraries | 2022 | B.Sc. IEM |
| Mariam Bukhrashvili | The impact of Covid-19 on Quality Audits The Future of pharmaceutical quality audits | 2022 | B.Sc. IEM |
| Mohamed Ghandour | Thesis 2022 21. Mohamed Ghandour, Product Portfolio Complexity Management a data-driven approach using ‘Analytical Hierarchy Process’ and the ‘Technique for Order Performance by Similarity to Ideal Solution | 2022 | B.Sc. IEM |
| Mulisa Kevine Umwiza | The Role of Digital Twin in Sustainable Supply Chain in the Automotive Industry | 2022 | B.Sc. IEM |
| Rauan Beibit | COVID-19 Impact on the Supply Chain Effects of the Semiconductor Shortage on the Automotive Industry | 2022 | B.Sc. IEM |
| Stefan Krstevski | The Influence of Dynamic Electricity Price on Production Scheduling: A Mixed Integer Linear Programming Model | 2022 | B.Sc. IEM |
| Ylva Reinhold | How Do Industry 4.0 Applications Help in Designing Sustainable Forest Management? A Conceptual Framework of IoT in Novel Sectors | 2022 | B.Sc. IEM |
| Xuchong Du | Application of GNN in supply chain for link prediction and explainability analysis | 2022 | B.Sc. CS |
| Apivut Ourairat | Gap Analysis, State of the Art, Existing Artificial Intelligence in Agrifood supply chain | 2022 | B.Sc. IEM |
| Abumansur Sabyrrakhim | Mapping SQL (Pre)-clinical Data from Cancer Radiooncology and Radiobiology Studies to Ontology | 2022 | B.Sc. CS |