Table is ubiquitous in web documents. Turning the web-table information into ontology requires automatic approaches. In this paper, we discuss how to learn ontology from web-table. In order to obtain table schemata for learning of ontology, we first present general layout structure of the table, then, we propose ontology extraction method according to the five table group. The learned ontology includes the following relationships: is-a relationships, class-instance relationships, RDF triples, property domains and property ranges.