"Marathi-English Dataset for Query Refinement using word sense disambiguation"

"The standard dataset needed for this study is currently unavailable due to various ambiguities in the queries regarding the Marathi language. The dataset employed by the proposed model was gathered from public websites. (Source: https:\/\/marathivishwakosh.org , Source: https:\/\/vishwakosh.marathi.gov.in , Source: https:\/\/www.wikipedia.org , https:\/\/vikaspedia.in\/) as well as from different news website to facilitate conversational language retrieval, query refinement, query classification, word sense disambiguation in natural language. The meaning of the ambiguous words is taken from the Marathi Wordnet which is developed by IIT Mumbai.  The dataset is designed in Marathi-English languagecontaining ambiguous word, query and respective domain of that query and expected results in Marathi-English language. The number of records in Query part is 237 and results parts contains 1913 records approximately for Marathi-English language. We categorize this dataset as 70 percent for training purpose and 30 percent for testing purpose.But further as our research work progressed, we decided to create the data for both the Marathi and English languages. The number of records in Query part is 237 for English, Marathi Query and results parts contains 1913 records approximately for Marathi, English language. We categorize this dataset as 70 percent for training purpose and 30 percent for testing purpose."