Open Theses  

Locality-sensitive hashing (LSH) is used to speed up near-neighbor search in high dimensional space. LSH works by hashing the elements to discrete buckets. However, in some cases the hash function has to make a decision which leads to similar points being hashed apart. This, for instance, happens when a point is close to a hyperplane in RHH. One solution to this problem is to hash several small perturbations of the points and insert all of them into the indexes. Other solutions also exist. This thesis will look into the different options for improving the performance of LSH by hashing points to multiple buckets instead of just one.

Locality-sensitive hashing is used to speed up near-neighbor search in high dimensional space. When the distance of interest is cosine distance, Random hyperplane hashing (RHH) is used. This technique is based on randomly selecting hyperplanes. However, in some cases (when we have more information about the dataset) it seems reasonable to not choose the hyperplanes completely randomly. Further, if normal RRH is performed with a low number of hyperplanes, then the hyperplanes are likely to not cover the space very well. This thesis will be about choosing the hyperplane in a data dependent way and try to sample the hyperplanes such that they cover the space nicely (including a comparison with angular quantization).

Prototypes have been proposed recently as a new way to represent knowledge. Typically, one allow any kind of changes in a dataset. However, when a dataset is distributed with potentially malicious parties, it would be beneficial to make the dataset immutable and use some form of signature to prove authenticity. Moreover, making the data immutable has beneficial properties for caching. In the immutable scenario, the only way to make changes is by adding more prototypes. The student's task is to investigate the different options to make this immutable prototype store happen. This requires studying things like block chains and the internal git storage model. Further, the thesis can include some mechanisms to simulate some sort of mutability (e.g. by combining immutable and mutable parts).

An immersive Augmented Reality (AR training development framework needs support for combinations of a wide range of appropriate devices for different use cases. In a surgery training situation this can be an AR headset and a handtracking device capturing exact movements in order to evaluate the accuracy. Therefore, it is necessary to get information from different sensors and interpret them in a common sensor fusion framework, in order to record the actions. The goal of the thesis is twofold. First, the WEKIT.one framework will be extended. New sensor hardware, which can be used to improve the training experience of apprentices, will be implemented, tested and evaluated against the given requirements. The implementation is based on the Unity SDK in combination with Microsoft HoloLens and other AR relevant devices. Second, we are interested in visual learning analytics of the gathered data. Therefore, means for storing learner traces both locally and externally will be evaluated with the intention to use the stored data for long-term analytics. Contributions will be specified and implemented with the help of the upcoming IEEE standard on Augmented Reality Learning Experience Models (ARLEM). As a Bachelor thesis, the scope can be adjusted.

This thesis aims to develop social recommender system service for Open Source Software (OSS) Communities.

A business model is an abstract model of the business of one or more cooperating organisations. It is a conceptual and architectural implementation of a business strategy and the foundation for the implementation of business processes and information systems. The research objective of this thesis is the design, implementation and evaluation of an interactive tool for the engineering of a business model.

Alice wants to sell a knowledge graph (KG) to Bob, who wants to use it to extend his own KG. Bob wants to know how much it should pay Alice for the data. The only way seems to be that Alice gives her KG to Bob for evaluation. The problem is now that Alice cannot know that Bob is not going to keep the dataset. In this thesis, the student will look into algorithms to securely estimate the gain from combining these two KGs without actually sharing them.

Analyzing and processing open RDF data sources is vital for stimulating new discoveries in medicine. The Taverna workflow management system is an Apache incubator project ( https://taverna.incubator.apache.org/ ). The system enables scientist to create and store re-executable data analytics workflows which are described in the XML-based language SCUFL. Taverna is adopted by many research groups working in a variety of fields including bioinformatics (such as transcriptomics, proteomics and metabolomics), text mining, biodiversity and the Virtual Physiological Human. For life scientists, it is important to access distributed, changing sources such as Bio2RDF and DisGeNET-RDF to improve their outcomes. In this thesis, a Taverna plug-in will be developed to access open RDF data sources. The plug-in will browse services in RDF endpoints, query metadata, and add them to Taverna workflows.

Prototype ontologies are a new approach for knowledge representation. The task of the student is to create an editor for prototype based ontologies, based on the prototype knowledge base code provided. The editor must be intuitive to use and give suggestions to the user. Further, it must show how final values of the prototypes have been derived.

Investigate the relationship between communities and topics by applying overlapping community detection to the social network of contributors and subsets of intrawiki link networks on different Wikipedias.

Assigning automatically badges that reflect the learning progress of self-directed learners.

Social Bots (software robots) are computer algorithms that automatically produce content and interact with humans on social media.

Functional encryption is an emerging approach for encrypting data in a way which allows executing an a priori specified function on the encrypted data. While FHE/SMPC allows general functions to be executed, for functional encryption, one function needs to be specified beforehand, and only that function can be executed on the encrypted data. Can this be used as part of a personalisation process?

PIR is an approach for database queries in which the database can neither see the query nor the result, yet it can trust that constraints such as access control are followed. Can this be extended to enable personalisation if the query represents a user profile, and the constraints represent the logic/rules of the recommender system?

There are several algorithms to perform a hierarchical clustering, resulting in approximate dendrogram. This makes it possible to perform a clustering on big data sets. In this thesis the student will evaluate of several existing algorithms in terms of resource use and clustering quality. As part of this work, the student has to implement some of the algorithms to work on a GPU as they are not scalable enough for CPU computing.

Assuming that best practices for organizing work and learning emerge from the visual analytics and comparison of real communities, the goal of this thesis is to analyze media transcription processes from real communities using data mining and machine learning.

Goal of this thesis is an integration of post-mortem community data dumps with the MobSOS real-time community information systems success awareness framework.

Semantic technologies and RDF data representation can improve the reusability of scientific data and enable scientist to reproduce the experiments. However there is no tool to support researcher for making their data semantically interpretable by computers. Aim of the thesis is to understand the benefits of semantic web technologies for reproducible research and develop a tool which can convert experimental data to RDF by annotating with selected data models.

Prototypes have been proposed recently as a new way to represent knowledge. In recent years many datasets have been published using RDF. Your task is to find out how the RDF dataset can be converted to prototypes in an efficient manner. For a master thesis, you also have to work on the optimal conversion for different requirements such as updates in the original RDF data.

Often only noticed as a technology that enables the digital currency Bitcoin, blockchain is a novel protocol that allows the distributed and secure storing of information and untempered execution of program code in trust-less environments. Did you ever feel the intense desire to write a thesis about blockchain or do you have a slight hope that blockchain is the one-and-only topic that touches your heart? Use your chance now! We are looking forward to hear from you.