Europe is different from other large media markets such as the US or China in that information is being generated in different languages and distributed via diverse streams of localised media channels. Automatic analysis is complicated further by different content types (audio, video, text) and different channels (mainstream, social media). Thus, information can only be analysed independently for each dimension. This restricts the extractable knowledge and keeps it fragmented, which ultimately constrains the exchange of information.xLiMe proposes to extract knowledge from different media channels and languages and relate it to cross-lingual, cross-media knowledge bases. By doing this in near real-time we will provide a continuously updated and comprehensive view on knowledge diffusion across media, e.g., from European communities like Catalonia to worldwide content in English.Tools and methods developed in xLiMe will be applied in three complementary case studies and evaluated by several business clients and up to 10mio end users . We will1.\taugment more than 250 TV channels in different languages with up-to-date information from social media and news in near real-time,2.\tmonitor brands and the diffusion of opinions across languages and media, and3.\tanalyse online shop performance with regard to external cross-lingual, cross-media factors, like campaigns for brands and the emergence of public opinions.By combining speech recognition, natural language processing, machine learning and semantic technologies we will advance key open research problems, by1.\textracting machine-readable knowledge (entities, sentiment, events and opinions) from multilingual, multimedia and social media content and integrate it with cross-lingual, cross-media knowledge bases,2.\tsearching this knowledge with structured and unstructured queries in near real-time,3.\tmonitoring its provenance, consumption and diffusion and4.\tanalysing the interdependency between media exposure and behavioural patterns.