Searching for evidence
A step-by-step guide to searching for interventions using the Ceres2030 dashboard
Photo: Adapted from image by Tomás Guardia Bencomo, via iStock.
Introducing the Ceres2030 dashboard
We used an open-source dashboard to make all this information accessible, visualizable, and shareable. It is also easy to add new sources of information—and the potential to automate updating in the future brings us as close as possible to real-time analysis of research for policy relevance.
We see this as a way forward to help non-research audiences make better use of scientific information to aid decision-making for the Sustainable Development Goals.
Our database currently contains a sample dataset of 49,910 articles about agriculture, 31,900 from science journals, 18,010 from non-science journals (“grey literature). This was compiled from a search for research on small scale food producers, a target of SDG 2.
Articles have been analyzed using machine learning. We used a Word2Vec model to help train our dataset to recognize interventions and outcomes. Next, we used a model booster approach to discover and classify specific interventions within our dataset.
What does the dashboard allow you to do? It goes beyond keyword searching. Articles have been analyzed using machine learning and classified into categories that include interventions and outcomes.
Let’s take you through a sample search. First, we’ll select an outcome from the “outcomes found” search box, in this case “Greenhouse gas emissions.”
We can build a search using different filters to select for specific information we want to find—in this case, we also want to know about greenhouse gas emissions in terms of rural infrastructure and socioeconomic interventions.
Our search turns up 201 articles.
Now, we can analyze the search data through Kibana’s visualizations, or with custom visualizations.
We can quickly visualize topic trends, authors, countries covered.
Here’s an example of a custom visualization.
We also built an app so that data can be downloaded as a CSV and shared through the open source reference management software, Zotero.