With storage cost dropping and even large SSD devices coming into the price range for private consumers, it has become easy to collect and accumulate information - and never throw it away again. Capacities in the range of Petabytes have become a new standard for large organizations, not simplifying the task of employees to quickly locate information or to investigate what is available on a certain topic. Digital transformation processes digitize legacy information into electronic form and further contribute to volume and complexity.
Often, relying on colleagues and partners is the only way to reliably locate pieces of information needed. However, as in agile organizations the roles and functions of employees may change unexpectedly, and as skills evolve over time without being formally described, alone the task of identifying which colleagues may be suitable to ask has become a major challenge.
Plain text search through large collections of documents is neither efficient, nor is it guaranteed to find what one is looking for. Either there are far too many results or none. Tagging is an approach to attach semantic information to documents, however, this practice is often neglected and nobody wants to re-tag documents on a large scale because of new developments. As an example: documents describing certain explosives may have to be re-tagged with "safety" because of the invention of the airbag requiring these fast-acting, gas-producing explosives.
Consequently, what is called for is a more automated approach to reach a higher semantic coverage of information in an organization. Intelligent search is a key requirement needed in any knowledge-intensive enterprise to rapidly and accurately explore and find information in today's large piles of data from the Internet, corporate file servers and applications.
We create solutions to collect, process and index pieces of any type of information on a potentially large scale. Processing includes semantic extraction and enrichment, i.e., steps necessary to automatically derive meta-data, categorizations and relationships - according to rules, linguistic heuristics, or machine learning components. This will help later with search and exploration processes but also to facilitates active alerting for users who want to be informed if there is new information matching their profiles of interest. The semantic extraction of features enables a point-and-click metaphor to easily navigate through even large unknown bodies of information. Mobile devices with their limited keyboard input capabilities greatly benefit from this approach.