Update on Automated Data and Knowledge Discovery in Engineering Literature

Providing machine-readable engineering literature

First open access sample documents, in the form of journal articles from the publisher MDPI, can be found in the public test area of the TUstorage document repository. This area of the repository is divided into a catalogue and a distribution area. The catalogue area contains the metadata of all currently available articles, sorted by publisher, journal and volume, in the form of one data set per article (e.g. example data set). The landing page of each data set leads to the distributions (PDF or publisher XML) in the distributions area.

Work is currently in progress to make the associated image files available for all MDPI articles and to include more journals of this publisher. In the future, it should be possible to download large numbers of documents automatically from TUstorage via an API.

In addition to MDPI, the publishers Copernicus, Frontiers, PLOS and Springer Nature also agreed to make their open access literature available and provided access to their open access files. Scripts are currently being developed to enable these large amounts of literature to be imported into TUstorage automatically as well. Enquiries with further publishers are already in progress.

LISA – Q&A research assistant of Kadi4Mat

LISA is the Q&A research assistant of Kadi4Mat for efficient knowledge management not only in battery science. It is based on RAG (retrieval augmented generation) architecture and powered by large language models (LLMs). Users can upload up to three own documents in PDF or XML format into the service and ask questions about these documents via a chat interface to gain insights into the content of the documents.