research

Here I’ve collected some of my past and current research topics and interests. Feel free to contact me if you have any questions.

CURRENT

Trends and Terms: Analysis of Brand Sentiments and Occuring Trends
Brands are very important when it comes to buying decision. In this experiment I try to discover occuring trends, sentimental values and writing behavior of brand related web content. In addition to that, I’m doing research on the question: What is “media buzz”? Can we find an anatomy of this term and visualize it for the brands? (log, trendinc)

What’s up in Question Answering?
In recent years a lot new possibilities occured that can be used to create smarter Question Answering systems. This work will function as a round up for the state of the art systems and serve as pool of new ideas and combinations. (log)


UPCOMING

What’s up in Recommendation Systems?
Since I’m working on recommendation systems for almost 2 years now the state of the art moved significantly. With this work I try to catch up with the state of the art systems, categorize and classify them as well give an outlook. (log)


PAST

2017, History of German Settlements: Can we find old counties and lands in data?
Can we find a pattern in naming of German settlements? And if yes, where can we find them? I try to find significant pre- & suffixes of settlement names. The first publication on this topic was made some in April 2017 with Dreamcity (loglink, article1).

2017, Data Sets for German Question Answering (link), Repository, Alexander Bresk, Different state of the art systems in Question Answering are dealing with English data sets and are also specialized for English language. This work introduces different data sets to train classifiers for Question Answering systems. (article)

2016, Phonetizz: Phonetic Distances of Phrases (link), Repository, Alexander Bresk, Most phonetic algorithms are using a code or symbols that are not easily comparable. This motivates the development of a phonetic algorithm that uses the simplest forms of text comparison. With Phonetizz the Levenshtein distance metric was utilized to perform such calculations for such a phonetic measure. Another important point is that all known algorithms are just working on words. With Phonetizz it is possible to work also on phrases instead of just one word. (article)

2015, Question Answering using Unstructured Data (pdf), Master Thesis, Alexander Bresk, In this thesis I tried to implement a Question Answering system based on different frameworks for German speaking data. The pipeline transforms unstructured data into structured data. After that the system ist able to perform questions like: “Wann ist Konrad Zuse geboren?”. Nominee for Carl-Zeiss Informatikpreis

2012, 3D Head Pose Estimation using Consumer Depth Cameras, Bachelor Thesis, Alexander Bresk, This work was done at Cognitec Systems GmbH. I implemented a system that estimated the pos (3 euler angles) of the head using a Microsoft Kinect camera.

 


TEACHING

  • 2017, Machine Learning 101 (pdf), Machine Rockstars, Alexander Bresk