392279 ISY Project: Word Embeddings for Material Science (Pj) (SoSe 2022)

Short Description:
Short: The goal is to i) learn word embeddings from material science literature, ii) systematically search for analogies in the embedding space, and iii) to present the results in a user-friendly web-application.

Automatically gathering information and generating recommendations by machine learning and natural language processing methods is a hot research topic and of major interest for domain experts like material scientists. Therefore, the task is to develop an approach to automatically extend an existing dataset of publications (which means: develop an approach that finds and adds related publications). Then you train a Word Embedding model and analyze this vector space systematically to find analogies, which either reflect existing knowledge or new directions of research. To make your system usable by domain experts, your task is to develop a user-friendly web-application.

This student-project will be offered in the context of DiProMag, a BMBF-funded interdisciplinary research project with the purpose to advance the digitalization of material science.

References:

• https://escholarship.org/content/qt082091b4/qt082091b4.pdf

• https://arxiv.org/pdf/1301.3781.pdf

• https://www.dipromag.de

In case this would not find enough interest for a team project, this project proposal would be also offered (in reduced/modified form)

• an individual project
• a project for only 2 students