The xLiMe system is a “cross-lingual and cross-modal semantic annotation, search, and recommendation” system. It can support data from social media, news, and TV with up to 13 different languages, although currently it works with news in all 13 languages; social media in English, German, and Spanish; and TV channels in German, English, Spanish, Italian, and French.
The system works by taking in data initially processed from multiple external sources. The processed data follow the data models of xLiMe. The xLiMe system integrates this data and provides users with an interface of querying and recommendation. With multimedia data, the audio files are first converted into text using voice recognition software. Then the search or annotation for TV and other multimedia data is done with the converted text. The recommendations can work across different applications (multimodal). The system is able to recommend social media or news for TV streams, or recommend TV streams for news.
The authors have done two types of tests for the system, a user study and a cross-comparison of features with other systems. In the user study, TV recommendation data was compared to Zattoo’s recommendation system. Sixteen TV shows of different genres were used. The results show that 60 percent of the tests were voted better for xLiMe, 25 percent were better for Zattoo, and the other 15 percent were equal. In the cross-comparison study, the authors compared six features against nine other products. These features, which xLiMe provides, include: “cross-lingual keyword search,” “semantic search,” “possibility to use complex queries,” “cross-modal search,” “interactive query refinement,” and “live-updates.” None of the nine other systems provide all six features as xLiMe does. In this sense, xLiMe is a clear winner.
The xLiMe system provides a very impressive list of services that are very practical to users of all kinds. It is especially well poised in today’s society where many forms of information sources exist, such as news, social media, and TV, in particular for areas such as Europe where multiple languages are often used in daily life at the same time. The authors conducted comprehensive studies to compare the performance of xLiMe, showing the system reaches its design goal “to break the barriers in between languages and modalities for a seamless semantic access to media streams.”