The semantics of rich multimedia presentations in the web such as SMIL, SVG and Flash cannot or only to a very limited extend be understood by search engines today. This hampers the retrieval of such presentations and makes their archival and management a difficult task. Existing metadata models and metadata standards are either conceptually too narrow, focus on a specific media type only, cannot be used and combined together, or are not practically applicable for the semantic description of rich multimedia presentations. In this paper, we propose the Multimedia Metadata Ontology (M3O) for annotating rich, structured multimedia presentations. The M3O provides a generic modeling framework for representing sophisticated multimedia metadata. It allows for integrating the features provided by the existing metadata models and metadata standards. Our approach bases on Semantic Web technologies and can be easily integrated with multimedia formats such as the W3C standards SMIL and SVG. With the M3O, we unlock the semantics of rich multimedia presentations in the web by making the semantics machine-readable and machine-understandable. The M3O is used with our SemanticMM4U framework for the multi-channel generation of semantically-rich multimedia presentations.

Topic models are a popular tool to extract concepts of large text corpora. These text corpora tend to contain hidden meta groups. The size relation of these groups is frequently imbalanced. Their presence is often ignored when applying a topic model. Therefore, this thesis explores the influence of such imbalanced corpora on topic models. The influence is tested by training LDA on samples with varying size relations. The samples are generated from data sets containing a large group differences i.e language difference and small group differences i.e. political orientation. The predictive performance on those imbalanced corpora is judged using perplexity. The experiments show that the presence of groups in training corpora can influence the prediction performance of LDA. The impact varies due to various factors, including language-specific perplexity scores. The group-related prediction performance changes for groups when varying the relative group sizes. The actual change varies between data sets. LDA is able to distinguish between different latent groups in document corpora if differences between groups are large enough, e.g. for groups with different languages. The proportion of group-specific topics is under-proportional to the share of the group in the corpus and relatively smaller for minorities.

Data visualization is an effective way to explore data. It helps people to get a valuable insight of the data by placing it in a visual context. However, choosing a good chart without prior knowledge in the area is not a trivial job. Users have to manually explore all possible visualizations and decide upon ones that reflect relevant and desired trend in the data, are insightful and easy to decode, have a clear focus and appealing appearance. To address these challenges we developed a Tool for Automatic Generation of Good viSualizations using Scoring (TAG²S²). The approach tackles the problem of identifying an appropriate metric for judging visualizations as good or bad. It consists of two modules: visualization detection: given a data-set it creates a list of combination of data attributes for scoring and visualization ranking: scores each chart and decides which ones are good or bad. For the later, an utility metric of ten criteria was developed and each visualization detected in the first module is evaluated on these criteria. Only those visualizations that received enough scores are then presented to the user. Additionally to these data parameters, the tool considers user perception regarding the choice of visual encoding when selecting a visualization. To evaluate the utility of the metric and the importance of each criteria, test cases were developed, executed and the results presented.

This thesis presents novel approaches for integrating context information into probabilistic models. Data from social media is typically associated with metadata, which includes context information such as timestamps, geographical coordinates or links to user profiles. Previous studies showed the benefits of using such context information in probabilistic models, e.g.\ improved predictive performance. In practice, probabilistic models which account for context information still play a minor role in data analysis. There are multiple reasons for this. Existing probabilistic models often are complex, the implementation is difficult, implementations are not publicly available, or the parameter estimation is computationally too expensive for large datasets. Additionally, existing models are typically created for a specific type of content and context and lack the flexibility to be applied to other data. This thesis addresses these problems by introducing a general approach for modelling multiple, arbitrary context variables in probabilistic models and by providing efficient inference schemes and implementations. In the first half of this thesis, the importance of context and the potential of context information for probabilistic modelling is shown theoretically and in practical examples. In the second half, the example of topic models is employed for introducing a novel approach to context modelling based on document clusters and adjacency relations in the context space. They can cope with areas of sparse observations and These models allow for the first time the efficient, explicit modelling of arbitrary context variables including cyclic and spherical context (such as temporal cycles or geographical coordinates). Using the novel three-level hierarchical multi-Dirichlet process presented in this thesis, the adjacency of ontext clusters can be exploited and multiple contexts can be modelled and weighted at the same time. Efficient inference schemes are derived which yield interpretable model parameters that allow analyse the relation between observations and context.

Through the increasing availability of access to the web, more and more interactions between people take place in online social networks, such as Twitter or Facebook, or sites where opinions can be exchanged. At the same time, knowledge is made openly available for many people, such as by the biggest collaborative encyclopedia Wikipedia and diverse information in Internet forums and on websites. These two kinds of networks - social networks and knowledge networks - are highly dynamic in the sense that the links that contain the important information about the relationships between people or the relations between knowledge items are frequently updated or changed. These changes follow particular structural patterns and characteristics that are far less random than expected. The goal of this thesis is to predict three characteristic link patterns for the two network types of interest: the addition of new links, the removal of existing links and the presence of latent negative links. First, we show that the prediction of link removal is indeed a new and challenging problem. Even if the sociological literature suggests that reasons for the formation and resolution of ties are often complementary, we show that the two respective prediction problems are not. In particular, we show that the dynamics of new links and unlinks lead to the four link states of growth, decay, stability and instability. For knowledge networks we show that the prediction of link changes greatly benefits from the usage of temporal information; the timestamp of link creation and deletion events improves the prediction of future link changes. For that, we present and evaluate four temporal models that resemble different exploitation strategies. Focusing on directed social networks, we conceptualize and evaluate sociological constructs that explain the formation and dissolution of relationships between users. Measures based on information about past relationships are extremely valuable for predicting the dissolution of social ties. Hence, consistent for knowledge networks and social networks, temporal information in a network greatly improves the prediction quality. Turning again to social networks, we show that negative relationship information such as distrust or enmity can be predicted from positive known relationships in the network. This is particularly interesting in networks where users cannot label their relationships to other users as negative. For this scenario we show how latent negative relationships can be predicted.