Technology
The Ximilar data-management engine is partly based on technology studied within the MUFIN project. On the software level, Ximilar
utilizes open source
codes shared under the GPL license. This technology approach assumes a very universal
concept of similarity that is based on the mathematical notion of metric space. In this model, the
data collection is seen as objects together with a method to measure similarity between pairs of objects.
This concept of searching has been described, for instance, in book Similarity Search: The Metric Space Approach.
The technology follows these principles:
- Extensibility - the appraoch can be used on a wide variety of data types and forms of similarity including various aspects of image visual similarity (shapes, colors, textures), video sequences, face recognition, biometric data, music, various scientific data like time series, etc.
- Scalability - efficient even for very large data collections (billions of objects). Ximilar offers various database structures including distributed mechanisms, so that a suitable architecture can be chosen for specific data volumes.
- Infrastructure independence - the database technology is inherently very flexible and can be deployed on various centralized or distributed HW infrastructures. In this way, the overall performance can be tuned to requirements of the specific application.