This paper focuses on creating a dataset in an agricultural domain to understand and analyze sentiments related to pesticides, pests and crop disease. Collecting data allows us to gain insights into the opinions, attitudes, and experiences of farmers, agricultural experts, and consumers. Therefore, we collected data from social media and online platforms based on a list of keywords extracted from the AGROVOC multilingual controlled thesaurus. We have obtained two datasets: the first comprises 1617 tweets regarding pests and crop diseases from Twitter, while the second consists of 10 181 reviews on pesticide products from Amazon. Aspect extraction has been employed, identifying and extracting specific features or aspects discussed in reviews or tweets. This process enables a more granular and detailed data analysis by allowing us to go deeper into the data and understand the sentiment associated with each aspect rather than the overall sentiment. We identified eight aspects within the Amazon dataset, including “price”, “package”, “toxicity”, etc. Similarly, we extracted relevant aspects from our Twitter dataset, resulting in five aspects such as: “pests”, “safety”, “toxicity”, etc. Then, the data will be pre-annotated in collaboration with agricultural domain experts from INAT. This data will be used for future analyses and will inform the purchasing decisions of various decision-makers.