Datasets are the part and parcel of machine learning models, and Google has invested millions in artificial intelligence and its applications. One such initiative is Dataset Search, a specialized search engine tailored for combing through millions of datasets.
Today, after spending approximately a year in testing, Dataset Search is finally out of the beta phase. And based on the feedback it received from early testers, it has a bunch of new features to add to it as well.
First, Dataset Search is now available on mobile, just like the majority of traditional search engines. Second, you can now filter search results according to the type of data you want. Results can be filtered to show only images, text, tables, price, etcetera. Third, the quality of the search results has also been improved to show more information. For example, if a dataset in the search result pertains to a geographic area, a map will be shown in the description.
Regardless of Dataset Search's development stage, Google has claimed that it will continue to take feedback and improve the product. So far, Dataset Search has indexed around 25 million datasets. Those who publish data and wish for their datasets to appear in the search results can do so by using an open standard (schema.org) to describe the properties of their dataset on their particular web page.