Рефлексивные процессы и управление. Сборник материалов XI Международного симпозиума 16-17 октября 2017 г., Москва - стр. 16

3.2 The generation and use of tacit knowledge is redefined:

• Thenumber of issues analysed is dramatically increased: from maybe hundreds, analysed by an individual, to millions, analysed with the wisdom of the crowds.

• Theamount of data analysed per issue is increased, compared to standard statistical analysis, even when experts suggest that the variety of data they analyse in a situation, cannot be analysed with an algorithm.

• Theissue analysis can clearly explain the reasoning behind the outcome prediction, the risk of ignoring important attributes is reduced.

• The prescriptive analytics can support or even replace, within limits, the expert decision taking. In the decision situations, where the Big Data based reasoning capacity matches the complexity of the environmental situations. A clear example of such development are the traffic/ routing/ logistics management systems, where drivers are relying on the navigation systems.

3.3 The generation and use of theoretic knowledge is redefined:

• The theoretic knowledge generation can become data driven. Instead of confronting tacit knowledge of multiple experts, the models of behaviour can be extrapolated directly from the data of the issues themselves. Standardisation and the relevance of the issues would no longer be matter of expert perception, but data based. An example: in financial institutions operational risk management directives are based on the risk probability and consequences perception of the risk management experts. Based on the experiences in the financial sector, it is easy to conclude, that they often put emphasis on the wrong risks.

• Storing the theoretic knowledge: the big data theoretic knowledge is stored in the form of predictive or prescriptive models or is based upon reports, produced by the models. It upgrades and complements the existing theoretic body of knowledge.

• The theoretic knowledge use is instance based. The experience, stored in models is applied upon the instance data. The elaborations, predictions and prescriptions are used to support decision taking, communication and automation. The main advantage of using big data based models is, that users do not need to analyse the complete theoretic backgrounds, but can focus on instances.

The backdraft of implementing BDA is, that it cannot imply all of the knowledge on complex issues, especially if the data quality or the number of cases recorded is not sufficient to provide reliable models. Therefore Big Data stored knowledge should be focussed on supporting processes, related to relatively simple instances, with multiple repetitions.