The HIBU Platform is the result of years of information retrieval and computational linguistics experience.
The architecture allows the application of a sequence of analysis steps performed to an input document. After a first NLP pipeline application, the document text goes through a list of so-called «annotators», able to enhance the original information content with semantic tags. These tags are added and indexed with the document itself and will be used to characterize and further describe it, also allowing semantic based retrieval, intelligent filtering (faceting), computation of document similarity, etc.
The annotators are the core of the analysis. There are different kinds of them, from list and rule-based to machine learning based. They are able to recognize locations, person and company names, perform statistics on the content, classify, recognize a document type, extract information, and more. New annotators can be added and existing ones can be customized, as well as switched on and off, in order to better suite the needs of the specific application use cases.
This text describes in words what is depicted in the following diagram:
