16:45 〜 17:00
▲ [17p-A401-14] Discussing domain adaptation in TDM: from superconductors to large magnets research
キーワード：magnetic materials, tdm, machine learning
We have started working on new TDM processes for extracting materials and related properties in permanent magnetic materials publications. In particular, it aims to extract data related to Nd-Fe-B system for samples under various conditions, such as composition, additives, and grain size, and the main properties to be extracted are coercivity (Hc), remanence (Br), saturation magnetization (Hs) and BHmax.
The magnetic materials research domain has some overlapping with the superconductors research domain on which we have worked previously.
We have inherited 112 automatically annotated documents, of which 42 were manually corrected.
In this presentation, after sharing our analysis of the two domains from a data science perspective, we discuss the annotated data, and the details of our proposed new pipeline.
The magnetic materials research domain has some overlapping with the superconductors research domain on which we have worked previously.
We have inherited 112 automatically annotated documents, of which 42 were manually corrected.
In this presentation, after sharing our analysis of the two domains from a data science perspective, we discuss the annotated data, and the details of our proposed new pipeline.