In France and Europe, farmland represents a large fraction of land cover. The study and assessment of biodiversity in farmland is therefore a major challenge. To monitor biodiversity across wide areas, citizen science programs have demonstrated their effectiveness and relevance. The involvement of citizens in data collection offers a great opportunity to deploy extensive networks for biodiversity monitoring. But citizen science programs come with two issues: large amounts of data to manage and large numbers of participants with heterogeneous skills, needs and expectations about these data. In this article, we offer a solution to these issues, concretized by an information system. The study is based on a real life citizen science program tailored for farmers. This information system provides data and tools at several levels of complexity, to fit the needs and the skills of several users, from citizens with basic IT knowledge to scientists with strong statistical background. The proposed system is designed as follows. First, a data warehouse stores the data collected by citizens. This data warehouse is modelled depending on future data analysis. Secondly, associated with the data warehouse, a standard OLAP tool enables citizens and scientists to explore data. To complete the OLAP tool, we implement and compare four feature selection methods, in order to rank explanatory factors according to their relevance. Finally, for users with extended statistical skills, we use Generalized Linear Mixed Models to explore the temporal dynamics of invertebrate diversity in farmland ecosystems. The proposed system, a combination of business intelligence tools, data mining methods and advanced statistics, offers an example of complete exploitation of data by several user profiles. The proposition is supported by a real life citizen science program, and can be used as a guideline to design information systems in the same field.

在法国和欧洲，农田占土地覆盖的很大一部分。因此，对农田生物多样性的研究和评估是一项重大挑战。为了监测广泛领域的生物多样性，公民科学计划已经证明了其有效性和相关性。公民参与数据收集为部署广泛的网络进行生物多样性监测提供了绝佳的机会。但是，公民科学计划存在两个问题：要管理的大量数据以及对这些数据具有不同技能，需求和期望的大量参与者。在本文中，我们提供了针对这些问题的解决方案，信息系统对此予以了具体化。该研究基于为农民量身定制的现实生活公民科学计划。该信息系统提供了几种复杂程度的数据和工具，从具有基本IT知识的公民到具有强大统计背景的科学家，都可以满足几个用户的需求和技能。提出的系统设计如下。首先，数据仓库存储市民收集的数据。该数据仓库的建模取决于将来的数据分析。其次，与数据仓库相关联，标准的OLAP工具使公民和科学家能够探索数据。为了完成OLAP工具，我们实现并比较了四种特征选择方法，以便根据其相关性对说明性因素进行排名。最后，对于具有扩展的统计技能的用户，我们使用广义线性混合模型探索农田生态系统中无脊椎动物多样性的时间动态。拟议的系统，结合了商业智能工具，数据挖掘方法和高级统计信息提供了一个由多个用户配置文件完全利用数据的示例。该提议得到了现实生活中的公民科学计划的支持，并且可以用作设计同一领域的信息系统的指南。