BIGBISC: Fueling intelligence to business processes with soft computing in Big Data Scenarios

In the last two decades, a number of systems have been developed for the management of business processes, whose objective is to automate, monitor, analyze and optimize the processes implemented in an organization. However, it can be said that the application of analysis techniques based on Artificial Intelligence is still quite limited at the moment in the field of business processes.

The BIGBISC project aims to investigate new soft computing techniques for two fundamental areas within business processes:

  • In the first place, in the automatic inference of the processes in complex and unstructured scenarios, and with large volumes of data.
  • Secondly, developing mechanisms to describe in natural language the processes and their relevant information, with the aim to communicate effectively and understandably the process information to the human agents responsible of making decisions.

The project has defined six relevant real use cases, provided by companies and organizations from different productive sectors, which proves the usefulness and transversality of the proposal.

In addition, a model will be proposed to provide the algorithms and techniques developed in the project following the "as a service" paradigm. With this aim, layers of middleware will be developed to facilitate the execution of algorithmic solutions in multi-provider cloud environments.

Objectives

The objective of the project is the development of a set of algorithms of process mining and data-to-text, based on soft computing techniques, which will be integrated into a Bigmining as-a-service infrastructure to support business intelligence in Big Data scenarios. Both the algorithms and the infrastructure will be validated with several data sets corresponding to different scenarios and real application domains, in order to demonstrate the applicability of the proposed solutions.
This general objective can be described in the following specific objectives:

  1. Develop process mining algorithms for the discovery, simplification and detection of the change of complex processes in scenarios where large volumes of traces are generated, using clustering techniques to group the large volume of traces, in order to reduce the size of the problems and make them treatable.
  2. Develop algorithms based on data-to-text techniques for the description in natural language of business indicators and models of structured and unstructured processes in scenarios with large volumes of data, which will require a new model (Big) data- to-text with the flexibility to deal with these processes and scenarios.
  3. Design a Big-mining as-a-service infrastructure for the integration of business intelligence algorithms, in particular algorithms based on process mining and data-to-text, in order to guarantee its availability and provide the necessary computational resources for its efficient and low cost execution.
  4. Demonstrate the effectiveness and efficiency of the solutions in real scenarios, in different types of application domains, with structured and unstructured processes, and with very clear needs of the business analysis and of the specification of the processes.