Agriculture and allied sectors are crucial for any economy, specifically in the Indian economy it provides a livelihood to about two-thirds of the population. India has a diverse agricultural climate across different areas of the country, enabling the cultivation of a wide variety of crops. Government-approved agriculture web pages host a diverse bunch of handy information to farming enthusiasts, such as the latest farming methods, pest management options, recommended use of fertilizers and pesticides. Besides, agriculture forums, social networks, blogs, and other web resources provide relevant details on soil conservation, crop-friendly soils, and proper use of pesticides. Prudent employment of information extraction tools enables retrieval of required information pertinent to farmer’s queries from government agriculture websites and agriculture search cache. This necessity has engendered Knowledge Bases(KB) or on to logies in the agricultural domain. The major sub-tasks for KB construction include Named Entity Recognition (NER) and Relation Extraction (RE). In this work, we study weighted distributional semantic model for unsupervised Named Entity Recognition (NER) in domain specific texts, specifically focusing on agricultural domain. Developing accurate agriculture NER models requires over coming several challenges, including the lack of annotated data, domain-specific vocabulary, entity ambiguity, and contextual variation. The proposed approach is completely unsupervised and utilizes an extended BERT model with LDA topic modeling (exBERT_LDA+) for NER. The proposed Agricultural Named Entity Recognition (AGRONER)model, focuses on identifying six major entities, disease, soil, pathogen, pesticide, crops, and place. The existing four entities are recognized using the proposed algorithm while we utilize the AGROVOC dictionary for crops and Geo coding APIs for Place entities.
