Self
Hi, my name is Wojciech Walczak*.
I solve machine learning and text mining problems.
I work at Samsung's R&D unit in Warsaw, Poland.


Projects
This non-code repo describes my professional conversion from a sociologist and amateur programmer to a professional software engineer and data scientist.

For some recent code, check out my short snippets of neural networks-related stuff.


Papers / talks
My recent research was focused on paraphrase detection. In 2016, along with my teammates, I have won the Semantic Textual Similarity task during the SemEval 2016 contest. Our paraphrase detection system was based on recursive autoencoders, WordNet-based semantic weighting, and various supervised learning methods. You can read the paper here. For slides, please, refer to the talk given at Institute of Computer Science Polish Academy of Sciences. Also, somewhat introductory presentation about the topic was presented during the talk at PyData Warsaw meeting.

Another branch of my research is related to Facebook, especially the strategies of sharing links. Here's a paper describing who is most likely to reproduce link to mainstream content. The second one discusses what kind of people receive most feedback for the link sharing, and why social appreciation goes to the same people over and over again.

Contact
Feel free to drop me an e-mail at ww@underfitted.org or check out my accounts on any of these obscure services: github, twitter, stackexchange, goodreads, and linkedin.