On this page, we present our project ideas for Bachelor's and Master's theses. Most projects are also suitable as internship project or Ph.D. thesis project. Please note that you need to register an account to see the projects. There are various options of how you can approach your Bachelor's or Master’s thesis. You can develop a novel software tool, create novel algorithms, evaluate existing algorithms, and many things more. In the following, we explain what type of research projects exists and list our project ideas

...

What makes them ‘good’ is that they answer a question that has not been answered before (paper 3), or they propose a novel algorithm/concept (papers 1 and 2). In either case, they provide evidence that their answer is true, or their novel algorithm is better than the state of the art.

Automated Machine Learning (AutoML) & Algorithm Selection

Machine Learning

Recommender-Systems

Others

...

Automated Machine Learning (AutoML) & Algorithm Selection

Machine Learning

Recommender-Systems

...

Applied Research Paper

In “applied research” you aim at improving one field of application -- e.g. movie recommendation, lung-cancer prediction, face recognition, or stock market prediction. Typically in such a project, you ‘throw’ a large number of existing algorithms on the novel scenario, and see what algorithms perform best. For a Bachelor’s or Master’s thesis, such a project is fine. However, from a scientific point of view, such projects are normally considered second-class. Such projects often involve a lot of trial and error, and less theoretically founded idea. To illustrate the point: Imagine a person A) who proposed the idea of Support Vector Machines and evaluated the first SVM on a dataset with handwritten digits and a person B) who later proposed to apply SVMs on classifying images of cats and dogs. Person A) clearly made a much more significant contribution to the world. Nevertheless, applied research papers can make valuable contributions to the field (we have published many applied research papers ourselves).

...

Our Priority Projects

We are particularly interested in supervising the following projects. However, please be aware that it is important to us that these projects are completed very thoroughly. If you pick one of these topics, we expect a very high commitment from you. If you just want to pass your thesis, please pick another project. 

Recommender Systems, Personalization & User Modeling

Real-World Recommender-Systems Research (Darwin & Goliath and Mr. DLib)

We are operating Mr. DLib and Darwin & Goliath, two real-world recommender systems that deliver thousands of recommendations every month (see our publications for some additional details). Working on such a project is a great experience, though it also often is slower than working just with some dataset locally on your computer. Your work would be integrated in Mr. DLib or D&G, and you could test with thousands of users if your e.g. novel approach is really better than a baseline. The following projects relate to Mr. DLib or Darwin & Goliath. Some of them might even be done “offline”. Please note, to join the development team, you must have excellent Python knowledge.

Evaluation

These projects relate to the evaluation of recommender systems. To do them, you should already be familiar with the difference between online evaluations, offline evaluation, and user studies; and you should have heard of and calculated metrics like precision, recall, F1, and RMSE. If you are not, please read our paper A Comparison of Offline Evaluations, Online Evaluations, and User Studies in the Context of Research-Paper Recommender Systems before contacting us.

Novel Algorithms

These projects relate to the development of novel algorithms in the field of recommender systems and user modeling. These projects are challenging and most of them require good Math knowledge, creativity, thoroughness, and a bit of luck. With such projects there is always the risk that the novel algorithm will not be better than the state-of-the-art. While this will not necessarily affect your grades (you can also receive high marks for a ‘failing’ algorithm), the work can be frustrating. On the other hand, if it is successful, then it is extremely rewarding.

Comparative Studies

The following projects are comparative studies. This means, your goal is to run a large number of tools on a large number of datasets and identify, which tool is best (under which circumstances). The advantage of these projects is that they are relatively easy with a low risk of failure. They also provide a good learning experience as you will work with many different software tools. They are particularly suitable for internships or Bachelor’s theses. If you do a Master’s thesis, your work must be really comprehensive to receive high marks. We would recommend these projects to students who aim at a career in the industry e.g. as an engineer. If you aim at a research career, you should choose a different project.

Reproduce Existing Work

These projects are not novel. Instead, they focus on reproducing other researcher´s results. Hence, these projects are particularly suitable for internship students and group projects who want to learn about a certain aspect of recommender systems. They are not suitable for FYP or PhD projects.

User Experience / Interfaces

The projects relate to recommender-system aspects beyond accuracy, i.e. aspects relating to user experience and user interfaces. UX is a neglected topic in recommender-systems research and has a high potential to improve recommender-system performance. 

Novel Applications

The projects relate to applying non-recsys technologies to recommender systems. In other words, you will try how successful promising technologies that have not been applied to recommender systems, can be applied to recommender systems. 

Software Development

These projects relate to purely software-engineering related topics with no or little data analysis and research required. They may be suitable as group projects for some modules at TCD, or for internship students. They are probably not suitable as a Final Year Project and certainly not for a PhD thesis.

Non-Technical Topics (Surveys / Questionnaires / Ethics ...)

These topics relate to non-technical aspects of recommender systems. They suitable for non-computer-science students (and maybe for business-computer science students). 

Machine Learning

Meta-Learning & Automated Machine Learning

Evaluation

These projects relate to the evaluation of machine learning. You should be familiar with the difference between online evaluations, offline evaluation, and user studies; and you should have heard of and calculated metrics like precision, recall, F1, RMSE, and click-through rate. Please read our paper A Comparison of Offline Evaluations, Online Evaluations, and User Studies in the Context of Research-Paper Recommender Systems before contacting us.

Novel Algorithms

These projects relate to the development of novel algorithms in the field of machine learning. 

Comparative Studies

These projects relate to rather simple comparisons of different frameworks e.g. for machine translation. The projects are particularly suitable for internship students (3 months) who want to gain lots of practical experience. The projects are less suitable for final year projects (and certainly not suitable as PhD project), because they are intellectually not particularly challenging. 

Reproduce Existing Work

These projects are not novel. Instead, they focus on reproducing other researcher´s results. Hence, these projects are particularly suitable for internship students who want to learn about a certain aspect of machine learning.

User Experience / Interfaces

The projects relate to machine-learning aspects beyond accuracy, i.e. aspects relating to user experience and user interfaces. While these projects are technically not particularly challenging, UX is a neglected topic in machine-learning research and has a high potential to improve recommender-system performance. 

Novel Applications

The projects relate to applying non-ML technologies to ML. In other words, you will try how successful promising technologies that have not been applied with ML, can be applied to ML. 

Software Development

These projects relate to purely software-engineering related topics with no or little data analysis and research required. They may be suitable as group projects for some modules at TCD, or for internship students. They are probably not suitable as a Final Year Project and certainly not for a PhD thesis.

Non-Technical Topics (Surveys / Questionnaires / Ethics ...)

These topics relate to non-technical aspects of ML. They suitable for non-computer-science students (and maybe for business-computer science students). 

Assignments (CS7CS4/CS4404)

These projects are suitable as assignments for our Machine Learning (CS7CS4/CS4404) module taught at SCSS. 

Document Engineering

Metadata Extraction (Title, Authors, ...) 

Reference/Citation Extraction

Comparative Studies

Document Relatedness

Plagiarism Detection

Academic Search Engines (SEO, Spam, Ranking) 

Blockchain

IT Security

Mobile Apps (Android / iOS)

Software Development Focus

e-Business Assignments

The following tasks are suitable as a group project in my e-Business II module