Persian Web Document Retrieval Corpus

In this paper, we present a Persian dataset designed specifically for the task of text information retrieval through the web. Unlike most available datasets, we incorporate real-world user queries in this dataset distinguishing it from the existing ones in the field. We believe this dataset serves as a valuable resource for researchers and practitioners involved in the development and evaluation of information retrieval systems tailored for the Persian language. Besides the 951 anonymized Persian queries collected by the Zarebin search engine, this dataset consists of documents from the web retrieved based on the queries by both Google and Zarebin search engines. Another interesting feature is that the relevance of each retrieved document to the original query is human annotated. Overall, this corpus consists of 18568 pairs of query and document, each with a relevant/irrelevant tag¹.