CHALLENGES

Semasphere technology targets application domains and problems where the goals are to uncover novel information in large volumes of data, or to confirm the presence of complex phenomena and behaviours through information extracted from such data. We distinguish such problems from challenges consisting simply in detecting recurring patterns or correlations in data.

Data is an asset, and the way in which we are able to search and interpret it determines its value. Intelligently sifting through large volumes of data and being able to discover and extract task-relevant knowledge is a powerful decision-support capability. Finding information nuggets in data, i.e. new correlations which yield insight into behaviours and processes, has become a source of valuable technological, commercial and/or competitive advantage.

From this perspective 'data analysis' is taking on an entirely new meaning. Instead of finding patterns, clusters or individual correlations, tools for complex domains need to identify complex conditions embedded in large data volumes, which together have relevance in the semantic context of a given problem domain. To use a medical metaphor, it is the difference from identifying symptomps to generating characterizations of diseases.

Semasphere technology is intended to support the transition to higher complexity data search and analysis, by allowing analysts to extensively use and apply domain knowledge and semantics in order to uncover and/or confirm complex phenomena in data. Data analysis anchored in a problem context, supported by domain knowledge and relying on domain semantics has the potential to yield high returns in terms of the value of the information/knowledge it is capable to discover and extract.

Below are examples of problem domains standing to benefit from complex data analysis capabilities:

• Anomalies in the behaviour of technical systems can be characterised by conducting data analyses in sizable repositories of test and usage data to identify the wider context of such anomalies as well as their defining elements.
• Key observations in financial and economic data can be extracted and placed in a wider context via an intelligent exploration of large data spaces made possible by data analysis capabilities which can be dynamically re-configured by the analyst during the exploration process.
• Clinical trials of new drugs need to determine the context in which the drugs are effective or counterindicated. This can be determined by seeking to correlate effectiveness parameters with a multitude of background factors which may play a role through an in-dpeth analysis of clinic trial data sets.