The concept of bisociation

The concept of association is at the heart of many of today's powerful ICT technologies such as information retrieval and data mining. These technologies typically employ association by similarity or co-occurrence to discover new information relevant to the evidence already known to the user. However, association techniques fail to discover relevant information that is not related in obvious associative ways, in particular information that is related across different contexts. It is these kinds of context-crossing “associations” that are often needed in innovative domains.

Domains that are characterized by the need to develop innovative solutions require a form of creative information discovery from increasingly complex, heterogeneous and geographically distributed information sources. These domains, including design and engineering (drugs, materials, processes, devices), areas involving art (fashion and entertainment), and scientific discovery disciplines, require a different ICT paradigm that can help users to uncover, select, re-shuffle, and combine diverse contents to synthesize new features and properties leading to creative solutions. People working in these areas employ creative thinking to connect seemingly unrelated information, for example, by using metaphors or analogical reasoning. These modes of thinking allow the mixing of conceptual categories and contexts, which are normally separated. The functional basis for these modes is a mechanism called bisociation (see Arthur Koestler - The Act of Creation).

Bisociation refers to the mixture of concepts from two contexts or categories of objects that are normally considered separate by the literal processes of the mind. Koestler coined the term bisociation to distinguish the type of metaphoric thinking that leads to the acts of great creativity from familiar associative thinking. Bisociation, according to Koestler, “means to join unrelated, often conflicting, information in a new way.

The history of engineering and science is full of serendipitous discoveries, which are based on bisociative processes. Arguably, the most well-known ones are the anecdotal falling apple that gave Newton the inspiration for his theory of gravitation, and Archimedes’ bisociations between the concept of specific weight and his body displacing water in the bathtub. Other serendipitous discoveries include penicillin, Viagra, aspartame (artificial sweeteners), and many more.

Objectives

While current ICT approaches provide methodologies and tools for association-based search and processing of information, there is currently no comprehensive ICT methodology or tool which facilitates the bisociative exploration for discovery and design tasks. The overall aim of the BISON project is to develop and validate a computational methodology, which facilitates bisociative information discovery in large-scale heterogeneous information environments.

In particular, the main objectives of the BISON project are the following:

• Objective 1: To develop a conceptual and theoretical framework for bisociative information discovery in multi-source and distributed computing environments.
• Objective 2: To design and implement a comprehensive, computational realisation of the bisociative information discovery framework. Main elements of this development include
  • (i) a meta-structure called bisociative information network,
  • (ii) novel graph-processing algorithms facilitating bisociational reasoning within such networks,
  • (iii) methods that extract and populate these networks based on heterogeneous and distributed sources, and
  • (iv) user interfaces that facilitate effective and efficient exploration of such networks.
• Objective 3: To demonstrate the validity of the BISON framework and to evaluate and assess its properties on the basis of
  • (i) a diverse set of demonstration scenarios drawn from different domains, and
  • (ii) a range of hypothetical test scenarios with known characteristics, that will be proposed in the project.

  The chosen problem scenarios will have in common that they are difficult to solve (effectiveness, efficiency) with known ICT methods.

• Objective 4: To assess whether the methodology and its proof-of-concept implementation could lead to a new ICT paradigm.
• Objective 5: To widely disseminate the results of the project, raise awareness in the relevant communities and explore ways of exploiting the project results.

Vision and challenges

While frequently discussed in cognitive science, psychology and related literature, there does not seem to exist a serious attempt at trying to formalize and computerize the concept of bisociation. Discovering bisociative relationships requires sufficiently large and diverse underlying information bases, demanding dedicated data integration, storage, distribution and processing strategies.

We anticipate that the BISON system will provide truly creative solutions in an interactive environment that implements novel knowledge integration, network visualisation and machine learning methods to aid creative discovery. BISON builds on widely researched methodologies such as association rule learning, analogical, metaphoric and case-based reasoning.

The BISON implementation of bisociation will be a unique concept, driven by the need for bisociative discovery in innovative domains, by the effective and efficient integration of very large and highly heterogeneous information sources, and the confluence of human and machine capabilities.

Project information

The BISON project is funded by the European Commission under the Framework 7 programme.

 See also:

• FP7 project page

References

For further reading please refer to these books and publications:

1. Arthur Koestler (1990): The Act of Creation. Penguin. ISBN 0140191917. (currently out of print)