Janus Wawrzinek

General

Janus Wawrzinek, M.Sc.



Technische Universität Braunschweig
Institut für Informationssysteme
Mühlenpfordtstraße 23, 2.OG
D-38106 Braunschweig


Phone: +49 (531) 391 7447
Email: wawrzinek@ifis.cs.tu-bs.de
Room: 231

Detail
My main focus is the analysis of Neural Language and Deep Learning Models applied in the field of pharmaceutical/medical digital libraries. Besides investigating the semantics that have been learned between pharmaceutical entities by these models, I investigate whether these models can be used not only to learn, but also to predict new relationships between pharmaceutical entities such as drug-disease associations.
 

Latest News

 

2020 Covid-19 and Sars-CoV-2 Web-Services

We re-trained our Neural-Network approaches with the actual biomedical literature (PubMed + WHO Database). As next we will also include the preprints from bioRxiv.org.
 
You can already search/explore the results for covid-19 and 50.000 other disease-related terms in PubPharm (Link). The facets “Related Substances”, “Related Symptoms/Diseases” and “Related Genes” show the relationships that were found/predicted by the Neural-Network.
 
In addition we also present the results for Sars-CoV-2 (Link).
 
2020 JCDL PC-Member

I am glad to be a JCDL PC-Member this year.

 

2020 New PubPharm Prototype
The exponential growth of publications in the bio-medical field makes it increasingly difficult to access the information contained in literature. For example, tens of thousands of publications are published annually about the disease "diabetes". Such huge amounts of publications simply cannot be read by individuals anymore. In such a scenario, intelligent systems are needed that can automatically extract useful information for scientists from the literature.
One possible solution is the use of AI, which can automatically detect or even predict relationships between active substances, genes and diseases based on millions of publications. On the other hand, from a user’s perspective, it is often difficult to assess what an artificial neural network has learned explicitly. Based on our latest publications, we have developed a prototype that facilitates the exploration of learned and predicted drug-disease associations. In this context, network views provide a simple overview of the complex relationships between the different entities. Our prototype is currently being evaluated by the (pharmaceutical) community. The goal is to integrate it this summer.
 

 

 

 

2018 Innovative service for the pharmaceutical digital library PubPharm
Based on our work „Semantic Facettation in Pharmaceutical Collections using Deep Learning for Active Substance Contextualization“ we implemented an innovative service for the pharmaceutical digital library PubPharm. Our service provides an alternative access path to literature beyond mere keyword or bibliographic search.

klick me!

 

2018 Presented innovative services on the CeBIT

We presented our research output on the community exhibitor stand of the Lower Saxony Ministry for Science and Culture.

We got also the chance to present our work to the secretary of state Dr. Sabine Johannsen on the largest and most

internationally representative computer expo (CeBIT).

 
 

 
2017 "Best Paper Award" awarded at 19th International Conference on Asia-Pacific Digital Libraries (ICADL’17), Bangkok, Thailand
 

http://www.ifis.cs.tu-bs.de/webfm_send/2252

 

 

 

 

 

 

Supervised Thesis

Thesis Type Student Name Title
Master Thesis Vidya Mohan Sathya Drug-Repurposing by Exploring the Semantic Similarity of Drugs
Project Thesis Philipp Markiewka Topic Modeling and Topic Labeling of Deep-Learned Facets

Research Project

Vidya Mohan Sathya

Semantic Context of Drugs using Neural Networking Mode

Lectures

Semester Role Lecture
Summer 2018 assistant Software Entwicklungs Praktikum "TEActive"
Winter 2017 assistant Seminar "Narrative Information Systems and Storytelling"
Summer 2017 assistant Relational Database Systems
Publications
2020
Wawrzinek, J., S. A. R. Hussaini, O. Wiehr, J. M. G. Pinto, and W. - T. Balke, "Explainable Word-Embeddings for Medical Digital Libraries – a Context-Aware Approach", Joint Conference on Digital Libraries (JCDL’20), Xi'an, Shaanxi, P. R. China, ACM/IEEE-CS, 08/2020. Abstract  Download: Wawrzinek82.pdf (692.25 KB)
Wawrzinek, J., J. M. G. Pinto, and W. - T. Balke, "Mining Semantic Subspaces to Express Discipline-Specific Similarities", Joint Conference on Digital Libraries (JCDL’20), Xi'an, Shaanxi, P. R. China, ACM/IEEE-CS, 08/2020. Abstract  Download: Wawrzinek85.pdf (782.76 KB)
2019
2018
Balke, W. - T., K. Keßler, A. T. Krüger, K. Stump, J. Wawrzinek, and S. Wulle, "FID Pharmazie: Zwischen Spitzenforschung und verlässlicher Infrastruktur (German)", ZfBB - Zeitschrift für Bibliothekswesen und Bibliographie, vol. 65, no. 2-3, pp. 114-117, 2018.  Download: ZfBB-PubPharm.pdf (4.62 MB)
Wawrzinek, J., J. M. G. Pinto, P. Markiewka, and W. - T. Balke, "Do Scaling Algorithms Preserve Word2Vec Semantics? A Case Study for Medical Entities", 13th International Conference on Data Integration in Life Science (DILS2018), Hannover, Germany, 11/2018. Abstract  Download: Camera-Ready of DILS2018 Paper 7.pdf (862.63 KB)
Wawrzinek, J., and W. - T. Balke, "Measuring the Semantic World – How to Map Meaning to High-Dimensional Entity Clusters in PubMed?", The 20th International Conference on Asia-Pacific Digital Libraries (ICADL), Hamilton, New Zealand, 11/2018. Abstract  Download: Camera-Ready of ICADL2018 Paper 38.pdf (631.95 KB)
2017
Wawrzinek, J., and W. - T. Balke, "Semantic Facettation in Pharmaceutical Collections using Deep Learning for Active Substance Contextualization", The 19th International Conference on Asia-Pacific Digital Libraries (ICADL), Bangkok, Thailand, 11/2017. Abstract  Download: Camera-Ready of ICADL2017 Paper 19.pdf (524.06 KB)