Current Multimedia Information Retrieval (MMIR) needs to be extended in a smart way to provide semantic, explainable, human-understandable, effective, efficient, interoperable, and integrated solutions.
Multimedia Information Retrieval is a core technology for almost any modern application. Broadcast, Healthcare, Social Media, Content Producers, End-Users, Automotive, and many others. Therefore, any improvement of this core technology enormously pushes applications, frameworks, apps, or tools in numerous areas.S.W.
According to the Oxford English Dictionary, the definition of the term smart is: “intelligent, able to learn and think quickly. Showing good judge- ment. Done quickly with a lot of force. In a clever and effective way”. For MMIR, this means, that
- MMIR querying and result presentation needs to be made human-understandable.
- MMIR needs to be enabled, to automatically explain the MMIR pro- cessing steps and results.
- MMIR needs to be highly scalable in any means (e.g., level of detail, features, media types, collections, servers).
- MMIR needs to fully support semantics, semantic integration, and ex-planation.
- The effectiveness and efficiency of MMIR retrieval needs to increase.
MMIR solutions need to understand right and wrong information. Given the definition of the term “smart” and considering this list of corresponding challenges, it is clear, that a literally “smart” solution for MMIR is needed to solve the overall goal or providing human-understandable, expressive, semantic, and explainable MMIR.
Therefore, the term “Smart Multimedia Information Retrieval” has been introduce describing MMIR solutions, that fulfill the above mentioned criteria.