A Natural Language Processing-like Method for Streamline Similarity Analysis

This paper introduces a method for streamline similarity analysis, named NLSQ (Natural Language Streamline Query), which is akin to NLP. This method involves segmenting streamlines and constructing geometric feature vectors for the segments. A vocabulary is established based on the similarity of streamline segments. Each streamline is modeled as a sentence, with the context of the segments preserved through the sentence’s word relationships. The Word Mover’s Distance is introduced to calculate the similarity between sentences, enabling the analysis of streamline similarity. The proposed method is compared with two other streamline query methods using four sets of 3D flow field data. The experiments demonstrate that NLSQ exhibits superior performance in in the streamline query task.