Technology has infiltrated most of our daily personal and professional endeavors. Accordingly, Human-Computer Interaction investigates how to improve the user experience and reach technology's full potential by focusing on different human factors, e.g., affect (Affective Computing) or personality (Personality Computing). Many affect-, personality, and context-aware applications have emerged in recent years. Many of them rely on computational models representing the involved human factors to enable scaling, sharing, and reusing these applications. Due to the complex interactions and multimodality of human factors like affect or personality, ontologies have often been used to model them. There are many existing computational ontologies for representing and predicting behavior based on personality and emotions. However, personality, emotions, and context alone do not accurately compensate for perceived behavior differences in different situations. Based on Character Computing, behavior is driven by character, i.e., trait (e.g., personality, socio-economic embedding, culture) and state markers (e.g., affect, motivation, cognition, well-being) and the markers of the current context or situation. Thus, to achieve character-aware computing, we need a computational ontology-based model of situation-specific behavior based on state and trait markers. Existing ontologies are usually limited to specific character markers or a set thereof, certain models representing the chosen markers, and particular use-cases. The aim of this thesis is to develop an ontology of character-based situation-specific behavior in a psychologically driven manner to be leveraged for interactive applications. The main contribution of the proposed ontology, CCOnto, is overcoming the lack of flexibility of existing ontology-based models when choosing the state, trait, and behavior markers to be included and their models. Accordingly, the ontology is developed in a modular, model-agnostic approach. The ontology consists of a core tier modeling character-based interactions, which is extensible using use-case- and domain-specific tiers. The modularity of the ontology extends to the knowledge level, where any less stable knowledge instances (from the user's perspective) are defined as rules on top of the ontology. Computer scientists and psychologists can use the ontology to develop character-aware interactive applications and conduct research into human factors. The application possibilities of CCOnto are validated through application development and data-centered approaches. We developed two ontology-based applications, one for predicting sleep quality and personality traits and the other for predicting anxiety. Each application relies on a different rule input mode, i.e., literature- and data-based. CCOnto can be used for data-based psychological research. We validate collecting a student behavior dataset during the first COVID-19 lockdown. We show how the CCOnto can categorize and organize heterogeneous data, infer missing data, and test hypotheses against datasets. Finally, we propose three platforms supporting the research intended by CCOnto. As enabling flexibility and model-agnosticism leads to having huge complex ontologies, we propose, VisCOnto, a module-based ontology visualization approach. To ease the rule-based knowledge input, we provide a wizard-based visual rule authoring platform. Finally, future scalability is contingent on having enough data. Thus, we propose AppGen, a wizard-based platform for generating technology-assisted data collection applications.