Semantic Visions runs an automatic sensor which understands the meaning of text. This sensor is powered by a semantic cross-language ontology, which is the key part of all our solutions, but it can work also outside our system – it can be integrated into other platforms. We have developed the ontology from scratch and it represents the most valuable part of our technology.
Our ontology covers the entire information spectrum (business, science & technology, politics, defense & security, health, entertainment, sports, geographic segmentation, etc.). Semantic Visions' ontology structure is both hierarchic and network. In each supported language, the ontology is structured across 90,000 nodes, of which each contains its own definition. The cross-language equivalence of the structure is currently available in eleven languages, with the number of nodes in excess of 1.25 million. We employ both human developed and automatically tested semantic definitions. Our node definitions are based on complex sets of criteria which include semantic representations and functions (synonyms, hyponyms, specific terminology, count, distance, part of speech, relevance etc.)
Semantic Visions has developed semantic structures and definitions which are able to capture the sentiment of written text and custom defined events with the highest precision possible. Our approach differentiates between the sentiment of the whole article and multiple entities included in it (e.g. names of companies or persons), it is multilevel and applicable for both longer documents as well as short wall posts and comments.
Semantic Visions research takes document similarity clustering to a new level. Because we comprehend the "meaning" of text, we are able to cluster or group similar news items, events and a diverse range of documents - across various languages. In addition our approach enables de-duplication of content. With comparisons of source efficiency, Semantic Visions applies research to capture and consolidate world events and globally discussed themes in many different languages. This combination of unique and timely acquisition with multi-language similarity clustering enables our solution to detect emerging stories as they break and thus provide an "early warning" detector.
Semantic Visions’ set of sources is independent on Google or any other search engines. We acquire data from various types of sources including Online news, Blogs, Press Releases, Password protected sites, Proprietary digitized print content, Social media, Customer internal content and Texts converted from TV & Radio. We apply advanced filtering options to adjust the data input to your needs.
Our platform is a functional system which integrates components for discovery of data sources, data acquisition, semantic processing, statistic processing, similarity clustering, web user interface and application programming interface. All components are service based and can be implemented as separate modules. The world's most powerful semantic engine - taking care of semantic data processing - is the holy grail.