The first project year for CASPAR 02 (Combined Automated Semantic Processing Array) has passed successfully and the team is happy to announce the achievements of their work!
Throughout the project year the team not only designed the software selection methodology with the help of external experts, but also acquired all the necessary infrastructural components (API, UI and the respective hardware), needed for the successful implementation of CASPAR.
Another crucial part of the work was conducted by the team of media analysts, which successfully created golden standards of annotated content. The golden standards were later used for machine-training purposes, and neural networks were trained in several different domains (Pharmaceuticals, Automotive and Cosmetics). The results were analyzed and contributed to the identification of new features, improvements of the used algorithms and code, and the implementation of those during the refactoring and system optimizations tasks.
GeoImaging is the partner company, mainly responsible for handling image data tasks. Four project meetings were organized between A Data Pro and GeoImaging to present both partners’ achievements, discuss the work progress, and distribute further tasks and deliverables in accordance to each Work Package.
As a result of the first project year the team reached one of the major milestones of the project – a fully functional and trained Neural Network – ANNITA.
We are now looking forward to the second year, during which the development of SPAS (Symbolic-based Processing & Analysis Systems) will be realized. The project team will focus on ontologies and search strings, and develop non-NN-related methods of semantically analyzing content. Furthermore, ANNITA will be trained into new domains. Lastly, A Data Pro will integrate the two ANNITA modules together with the SPAS range of methods into a common platform accessible by independent users (CASPAR).