From OntologPSMW

Jump to: navigation, search
[ ]

Contents

Processes involving Learning, Reasoning and Ontologies     (2)

It should be obvious that learning, reasoning and ontologies occur within larger processes where they, and the relationships between them, form steps of the process. John Sowa proposed that “For intelligent systems, the cognitive cycle is more fundamental than any particular notation or algorithm.” Then he concluded that “By integrating perception, learning, reasoning, and action, the cycle can reinvigorate AI research and development.” [1] There are many examples of cognitive cycles where learning, reasoning and ontologies all occur. The scientific discovery process is an example with a long history. A great many other activities can be regarded as decision making cycles in which each iteration improves understanding and awareness. In the past, a single iteration of a cognitive cycle, such as the scientific discovery process, could take decades. Today, cognitive cycles occur more quickly, much more data must be processed and the data is more complex. Learning, reasoning and ontologies, and the relationships between them that are the subject of this Ontology Summit, can play important roles in cognitive cycles. Several previous summits are also relevant to the cognitive cycle, including the need to deal with massive amounts of data OntologySummit2014 which come from large collections of sensors Ontology Summit 2015 and require many steps that must interoperate OntologySummit2016.     (2A)

Opportunities     (2B)

Combining learning, reasoning and ontologies within cognitive cycles introduces new opportunities.     (2B1)

  1. The cognitive cycle is a process made up of interrelated steps. This can be modeled using process and workflow ontologies. For scientific discovery, ontologies for scientific materials and methods were developed in [2], and the process was represented using a workflow ontology in [3]. In [4], decision making processes are represented using an ontology. Capturing scientific knowledge about data and analytic processes can assist scientists in analyzing data systematically and efficiently. There are similar advantages for capturing knowledge about decision making processes. In both cases, formalization can help ensure that the questions being asked and conclusions being made are precise. In addition, by modeling the processes and workflows one can automate discovery and decision making processes.     (2B2)
  2. By developing a domain-independent ontology of functions, structures and behavior, one can reuse the same techniques for other problems and in other domains. Indeed, one can use the cognitive cycle recursively, such as a cognitive cycle where the steps in the cycle are themselves cognitive cycles or a a higher-level cognitive cycle that manages another cognitive cycle [5].     (2B3)
  3. Because a cognitive cycle has multiple steps as well as many iterations of the cycle, it is important to keep of record of what has been done. This has long been a fundamental requirement of scientific discovery, but it has been expressed in natural language in lab notebooks. Capturing the provenance of cognitive cycles in a more formal manner would have significant advantages, such as unambiguously determining priority and responsibility. Other advantages are to monitor and prevent misconduct and to ensure that inferences are properly performed. However, the most important advantage is the need to explain why steps were performed. Humans remain important in any cognitive cycle whose ultimate goal is to improve understanding and awareness. This is only possible if the cognitive cycle can explain itself.     (2B4)

Challenges     (2C)

While combining learning, reasoning and ontologies within cognitive cycles has potential advantages, it is not commonly practiced. To the extent that such processes are automated at all, they are generally ad hoc and informal. For the scientific discovery process, it is necessary to use NLP to extract the materials and methods used by experiments [2] and the scientific hypotheses that are generated [3]. The challenge is to develop the required ontologies, to standardize them, to formulate best practices, and to convince communities to use them. In some cases, such as PROV-O for provenance [6], the ontology has been standardized, but other requirements of the cognitive cycle are less advanced. None of them are frequently used, and best practices are only starting to emerge.     (2C1)

Future Prospects     (2D)

There are many roles that learning, reasoning and ontologies can play in the cognitive cycle. These include:     (2D1)

Reasoning is fundamental for each of the steps in the cognitive cycle as well as the transition from one step to another step. Reasoning determines which facts produced by the sensors are relevant to the goal, whether scientific discovery or decision making. Reasoning is responsible for determining the best hypothesis or hypotheses that are supported by the relevant facts. Reasoning is used for making the decision about which hypothesis is the best and what should be done next.     (2D2)

Learning can be used for processing the relevant sensor data. Learning can also be used for developing the ontology that organizes the data and gives it meaning. Learning can be used at a meta-level to optimize the cognitive cycle, to detect problems with the cycle, and to help correct problems. Still another use for learning is to develop a library of optimized modules that are available for constructing new cognitive processes. Ontologies and reasoning can be used to ensure that the modules interoperate with one another correctly in a new cognitive process.     (2D3)

Ontologies play many roles in the cognitive cycle. They can formalize the whole process in a domain-independent manner which promotes reuse. Ontologies are closely connected with reasoning and interoperability. Ontologies can be the basis for maintaining provenance and for explaining the process, both of which promote human understanding.     (2D4)

References     (2E)

[1] John Sowa, Why Has AI Failed? And How Can It Succeed?     (2E1)

[2] Baclawski, K. et al, Database techniques for biological materials & methods, First Int. Conf. Intell. Sys. Molecular Biology, pp 21-28. 1993.     (2E2)

[3] Gil, Y., [1]     (2E3)

[4] Baclawski, K. et al, Framework for Ontology-Driven Decision Making, Journal of Applied Ontology, to appear.     (2E4)

[5] K. Baclawski et al, Self-Adaptive Dynamic Decision Making Processes. In CogSIMA 2017. [2]     (2E5)

[6] The PROV ontology at [3]     (2E6)