Computational Model of Sense Making – A High Level Approach

Sense making refers to how we structure the unknown so as to be able to act in it [1]Weick, Karl E.; Sutcliffe, Kathleen M., and Obstfeld, David. Organizing and the process of sense making. Organization Science, 16(4):409{421, 2005. doi:
10.1287/orsc.1050.0133.
. Sense making involves coming up with a plausible understanding, a map, of a shifting world; testing this map with other explanations of the same situation through data collection, action, and conversation; and then refining, or abandoning, the map depending on how credible it is. Sense Making is about actionable intelligence. Action is not a separate or later step in sense making, but an integral part of it [2]Ancona, D. SENSEMAKING Framing and Acting in the Unknown. SAGE,
2012
.

In the context of applied artificial intelligence, we propose to develop a model of sense making where computing capabilities of the machines can be exploited to carry out sense making in automated fashion and at scale. Before unfortunate dropping out of my research degree due to personal reasons, I had done some work in this field under the very capable supervision and guidance of Dr Tom Osborn and Dr Paul Kennedy at UTS. We were able to develop a high level architecture of a computational sense making model which I plan to further expand and finetune in due course as part of my ongoing research.

Computational sense making is a relatively new however active research
area. Computational Sense Making employs computational models and formal methodologies to develop a framework replacing the manual processes involved in traditional sense making. Field has gained a lot of traction in recent times and researchers are realizing the potential of employing semantic methodologies over manual sense making process to get better results. The work is being pursued in the field under different and some times seemingly unconnected titles and terminologies such as data mining using ontologies, unstructured data analysis, formal methods for decision making etc. Our model provides an end to end pipeline involving all the steps from data pre-processing, semantic enrichment, reasoning and querying. A conceptual view of our computational sense making process is given below –

To develop a formal computational model, we propose to use ontologies as knowledge base. Computational ontologies are a means to formally model the structure of a system, i.e., the relevant entities and relations that emerge from its observation, and which are useful to our purposes. 

One of the main use of formal ontologies is that intelligent agents and system components can reason on them to obtain desired knowledge. However, with increasing usage of ontologies for complex knowledge management tasks, it is desirable to have automated and scalable
mechanisms for reasoning. Since most of the formal ontologies are expressed in Description Logic, our approach to scalable reasoning is focused on Description Logic based reasoning (more specifically OWL based reasoning).

As part of this computational sense making architecture, we also propose to fuse the ontologies to be able to interoperate. As ontologies model specific domain knowledge and complex real world reasoning can not be done effectively without referring to more than one domains. Hence we need a mechanism to be able to refer to more than one ontology at a given time. These ontologies should be in alignment and our architecture should be able to handle any conflicts. For ontology fusion, we had extensively explored some current methods such as Graph Rewriting for Ontology Merging (GROM) Method [3]Mahfoudh, M.; Forestier, G., and Hassenforder, M. A Benchmark for Ontologies Merging Assessment, pages 555–566. 2016. and Formal Concept Analysis (FCA) Based Ontology Merging [4]Stumme, G. and Maedche, A. FCA-Merge: Bottom-up merging of ontologies, pages 225–230. 2001. .

A generic architecture of the entire sense making pipeline is given below –

 

I will try to publish more details of computational sense making concept and its implementation in subsequent blogs if time permits. In the meantime, if you want to get into the topic in more depth, please refer to 

Sense Making Draft Book

References

References
1 Weick, Karl E.; Sutcliffe, Kathleen M., and Obstfeld, David. Organizing and the process of sense making. Organization Science, 16(4):409{421, 2005. doi:
10.1287/orsc.1050.0133.
2 Ancona, D. SENSEMAKING Framing and Acting in the Unknown. SAGE,
2012
3 Mahfoudh, M.; Forestier, G., and Hassenforder, M. A Benchmark for Ontologies Merging Assessment, pages 555–566. 2016.
4 Stumme, G. and Maedche, A. FCA-Merge: Bottom-up merging of ontologies, pages 225–230. 2001.