In this article, we explain how the Semantic Web works and, in particular, how content providers can “discretise” information in order to correlate them within ontologies. We compare three syntaxes they can use: Resource Description Framework, microdata, and microformats, and we compare the normalization processes of each of these syntaxes. We show how different ‘politics of meaning’ are confronted on the Web and supported by companies whose objectives and interests differ. We analyse not only the mechanism of description and processing operated by hypertext markups, but also the mechanism of interaction between the people who describe and publish content on the Web and the machines that process and editorialise data.