About

Overview the project

CASPAR 02 is intended as a platform utilizing the potential of Artificial Neural Networks (ANNs) for text and image analytics. We will offer manually trained ANNs capable of processing content in multiple domains, that are available for non-tech users and do not require structured data or ontologies.

CASPAR will eliminate the need of data structuring or ontology creation, while alleviating the costs of content categorization and analysis. SMEs will benefit from automated entity extraction, topic classification, fact-linking, search and annotation to make better use of their information and big data without the complexity of existing tools. ANN technology will inevitably enter semantic processing and we want to utilize our potential and knowledge to lead this change in the data value chain.

The SaaS solution will uniquely combine text and image content analytic technologies, utilizing both ANNs (digital brains) and other existing semantic methods.

Project Goal

The project goal is to develop the conceptual base of a product, based on Artificial Neural Network (ANN) technology, in order to automate content analytics in a number of different domains. In order to operate efficiently, the system must be trained, both initially and regularly, within the domains that it is intended to be used in. Each domain is defined by the combination of: the industry or topic within which the meaning is intended; the language in use; and the information source that the data derives from. Each pre-trained domain-specific network will form a “digital brain”. Combined brains will form the core technology, called ANNITA (ANN for Image and Text Analytics). We will also identify and complement ANNITA with alternative content processing methods (search-string and ontology-based NLP), that we call SPAS (Symbolic-based Processing and Analysis Systems). By developing and integrating SPAS and ANNITA into a common framework we will create a Combined Automated Semantic Processing Array v.02 (CASPAR 02).

CASPAR 02 will utilize an API to receive image or text content sourced from an unstructured client data universe and process it utilizing ANNITA, SPAS or both methods, according to the needs and preferences of the client. The project will result in a web platform for fact extraction, topic classification, data-linking and annotation of image and textual documents in different languages. The CASPAR 02 platform is to bring a conceptual market change into automated content processing. Implemented as a SaaS and running on a distributed cloud infrastructure, it will transform the power of neural networks and semantic technologies into a commodity. It will simplify technology use for clients around the world and will not require extensive technological expertise. It will also have the scalability to process large-scale calls with millions of documents, typical for heavy content users and leaders in the media and information services industry.

Project duration: 24 months