One of the components of natural language processing that has received a lot
of investigation recently is semantic textual similarity. In computational
linguistics and natural language processing, assessing the semantic similarity
of words, phrases, paragraphs, and texts is crucial. Calculating the degree of
semantic resemblance between two textual pieces, paragraphs, or phrases
provided in both monolingual and cross-lingual versions is known as semantic
similarity. Cross lingual semantic similarity requires corpora in which there
are sentence pairs in both the source and target languages with a degree of
semantic similarity between them. Many existing cross lingual semantic
similarity models use a machine translation due to the unavailability of cross
lingual semantic similarity dataset, which the propagation of the machine
translation error reduces the accuracy of the model. On the other hand, when we
want to use semantic similarity features for machine translation the same
machine translations should not be used for semantic similarity. For Persian,
which is one of the low resource languages, no effort has been made in this
regard and the need for a model that can understand the context of two
languages is felt more than ever. In this article, the corpus of semantic
textual similarity between sentences in Persian and English languages has been
produced for the first time by using linguistic experts. We named this dataset
PESTS (Persian English Semantic Textual Similarity). This corpus contains 5375
sentence pairs. Also, different models based on transformers have been
fine-tuned using this dataset. The results show that using the PESTS dataset,
the Pearson correlation of the XLM ROBERTa model increases from 85.87% to
95.62%.

本文为首次使用语言专家生成了一个波斯语和英语句子之间语义文本相似性语料库 (PESTS)，并使用此数据集微调不同的基于 transformers 的模型，并且结果表明，使用 PESTS 数据集，XLM ROBERTa 模型的 Pearson 相关性从 85.87％提高至 95.62％。