Technical advances in agricultural machinery are expected to improve farmers’ working conditions and meet society’s and consumers’ demand for more efficient and healthier use of environmental resources. However, these advances may reduce farmers’ mastery of equipment. In this study, we aim to understand how farmers master agricultural equipment and adapt it to their farming needs. Aside from the design and development of new equipment, farmers can find ways to adapt their equipment through what we will call 'bricolage'. Bricolage refers here to the non-structural changes that go beyond the settings and options proposed by the equipment manufacturers and whose purpose is to adapt the equipment functioning to the farmer’s specific needs. Although bricolage could provide relevant insights into farmers’ needs in terms of new functions and equipment, the related information is difficult to collect because it is very specific to each individual farmer. To gather exploratory information on the agricultural equipment bricolages, we addressed the use of social media as a source, focusing on YouTube for two main reasons. First, bricolages are 3D equipment modifications that cannot be simply described by text or static photos. In this sense, videos are the best medium to transfer this type of information. Second, video production and streaming are more accessible thanks to the spread of smartphones. A new generation of passionate, super-connected farmers (the so-called “agri-YouTubers”) is conquering this and other social media. Their videos cover different styles, from the educational presentation of agriculture to the transmission of specialized knowledge. Altogether, YouTube videos are the best medium for farmers to share their experiences, especially on how to master and adapt equipment to their farming requirements. In this context, the study was structured into two parts. In the first part, we created a database of agricultural content creators on social media, to define a baseline of the number of people creating farmer-related content and to describe their goals and agricultural background. In the second part, we directly exploited the YouTube content using a scraping technique to create a collection of links via a query based on keywords related to bricolage, agricultural equipment and major agricultural operations (tillage, sowing, weeding, harvesting, post-harvesting). A preliminary set of 160 videos was retrieved for the period from 2005 to May 2022. Next, we applied automatic transcription and text-cleaning algorithms on all these videos to build a corpus that was then analyzed with the natural language pre-processing (NLP) technique. Such a process helps to classify the themes and the bricolages related to each practice. Then, we evaluated the deep-learning models to explore the relationships between the agricultural themes and the equipment type-specific bricolages. In perspective, the obtained results will contribute to enhancing an in-depth survey with a sample of farmers to identify the reasons underpinning the bricolages, so as to improve the understanding of farmers’ needs by equipment manufacturers.

This work is part of S.K.'s PhD thesis. It contributes to the activities of the Agro-Machinisme et Nouvelles Technologies Chair, UniLaSalle, with the financial support of the Michelin Corporate Foundation, AGCO – Massey-Ferguson, the Kuhn Group, and the Hauts-de-France Regional Council (FEDER funds). S.K. benefits also from a research allocation from the Hauts-de-France Region for the years 2020–2023.