Semantic modeling and knowledge representation in multimedia databases

Abstract

Abstract-In this paper, we present the current state of the art in semantic data modeling of multimedia data. Semantic conceptualization can be performed at several levels of information granularity, leading to multilevel indexing and searching mechanisms. Various models at different levels of granularity are compared. At the finest level of granularity, multimedia data can be indexed based on image contents, such as identification of objects and faces. At a coarser level of granularity, indexing of multimedia data can be focused on events and episodes, which are higher level abstractions. In light of the above, we also examine modeling and indexing techniques of multimedia documents.

Video Analysis and Indexing

“…or temporal (e.g. motion) features or are related to semantic objects (Li & Özsu, 1997) (Dağtas et al, 2000) (Al-Khatib et al, 1999. More sophisticated approaches are either a hierarchical representation of video objects (Analyti & Christodoulakis, 1995) (Yeo & Yeung, 1997) (Kyriakaki, 2000) based on their structure, or an event-based approach that represents a video object as a set of (non-contiguous, even overlapping) video segments called strata or temporal cohesions (Hacid et al, 2000) that correspond to individual events.…”

Constructing Knowledge Representation from Lecture Videos through Multimodal Analysis

“…Reference [5] presents a knowledge representation on American football videos. The trajectories of the America football are used as highly abstracted entities in the representation.…”

“…Instead of concentrating on efficient reasoning between different entities, video knowledge representations focus on facilitating the search and retrieval of videos, such as answering the query of "find a shot of a dog chasing a ball". Reference [5] gives a broad overview on knowledge representation of videos, including entities of different levels of abstraction and different models on representing the entities.…”

Object Segmentation and Ontologies for MPEG-2 Video Indexing and Retrieval

“…In all cases, preprocessing of video data is necessary as the basis on which indices are extracted. The preprocessing is of coarse granularity if it involves processing of video frames as a whole, whereas it is of fine granularity if it involves detection of objects within a video frame [1]. In this work, a fine granularity approach is adopted.…”