With Dataset Search, records can now be found in the search engine and filtered by categories such as publication date and usage rights.
Google is making the dataset search freely available after one year of the beta phase. This allows you to find studies, statistics and other data bundles that are of particular interest to scientists and data journalists in order to find information more easily, process it further or use it as a basis for your own research. Google wants to create an open data ecosystem in this way.
In the search interface, which is separate from the usual search engine, categories can be selected, such as the publication period, the rights of use, and whether access is free or paid for. It is also possible to differentiate between images, tables or text documents. Google then scans the metadata of thousands of sources on the Internet. Machine learning models are used to evaluate the data. Experience from the Knowledge Graph, which populates the information panels on the right in the normal Google search, also flows into the evaluation.
Search requests can be made via datasetresearch.research.google.com. Results are listed with the headline, source and publication date in a column on the left. If you click on it, a large preview opens to the right of the column. Here you can find further information and the link to the original source. Above the results, the categories can be selected. Google also shows how often a source has already been cited in scientific articles, also including links to the scientific literature search Google Scholar.
Google had already introduced dataset search in 2018, but initially in a beta version. Categories have been introduced since its launch, and Google Maps has also been included for searches on geographical topics. New is the availability of mobile devices. According to Google, a total of 25 million datasets can be found so far, and the collection is naturally set to grow further.
Markups for finding the data records
In order to make your own data findable in the search, metadata describing the content must be added to the corresponding pages, for example with the open standard Schema.org. Such markups help Google to find and classify records.