When recommending videos, the current context of the user plays an important role. The first task in the context of the project is to identify those recommendation approaches that best help to support contextual recommendation of videos. A context is composed of different data sources, such as current search strings of the user, search strings in similar situations, search strings of the "Nearest Neighbors" (users with similar interaction behavior) etc. Analysis and design are carried out in the work package "On-Demand-Clipping" , Implementation and evaluation of a context-based recommendation approach. The solutions developed for "on-demand clipping" form the basis for further developments with the aim of supporting the synthesis of videos ("on-demand clip merging"). On the basis of information such as available time, keywords describing the context and other information (so-called company-related information parameters in the context of the use of knowlede graphs), the system should identify (synthesize) a sequence of partial videos, which the user sees as a "solution video "represents. A simple application is eLearning: here, for example, a corresponding video can be generated by the system on the basis of the learning requirements of a user.
|Effective start/end date||1/10/21 → 30/09/24|
Explore the research topics touched on by this project. These labels are generated based on the underlying awards/grants. Together they form a unique fingerprint.