Web-based Semantic Similarity Checker using Sentence-BERT

Abstract The increasing need to have effective tools to evaluate the semantic similarity of sentences has been driven by the ever-increasing rate of information in the digital world in the form of text. This paper proposes the design and implementation of an efficient and effective web-based tool to calculate the semantic similarity of sentences in real-time using the power of transformer models on the web. This tool has been built using the Sentence-BERT model, which is specifically the all-MiniLM-L6-v2 model, to calculate the semantic similarity of sentences using the cosine similarity measure. This tool has been built using the Flask web development framework to allow users to input two sentences and calculate the similarity in real-time. The contribution of this work is the demonstration of the efficient and effective implementation of pre-trained transformer models on the web without the need to fine-tune the model on the GPU to calculate the semantic similarity of sentences, which has been demonstrated to be effective in the results to calculate the semantic similarity of sentences in the context of plagiarism detection and content analysis.