Timeline summarization in a cross-lingual scenario

Timeline Summarization (TLS) is the main approach that has been used to automatically create timelines of news reports related to some long-running event of interest. These timelines present a good way to keep up-to-date to the topic and follow its development over time, while getting only the salient information related to it. Most of the previous work done on TLS is focused on creating timelines for resources in a monolingual scenario, typically in the English language, and although some of them can be portable to other languages, they are still unable to combine the knowledge that has been extracted from news articles in different languages. The main contribution of this Thesis work is the study and development of TLS strategies tailored to multilingual resources. It aims at combining the textual information from different languages in order to generate a news timeline in a given target language. Among the biggest challenges addressed in this study is the analysis of these multilingual resources, as well as exploring the possibility to use the additional knowledge provided among the languages. The motivation for this is based on the fact that news provided in different languages can demonstrate a different aspect of the same event, as they may portray different cultural, economical or political reflections. The use of this additional knowledge gained from the multilingual resources has been proved to be beneficial, enriching the resources of the various languages and increasing their summarization quality significantly.