A knowledge based architecture for the interpretation of aerial images is presented. The Image Recognition System (IRS) utilises a multiresolution perceptual clustering methodology as a robust alternative to the more traditional edge or region based approaches. Initially, data driven feature generation and primary perceptual clustering is performed independently for two or more reduced resolution versions of the image. A Rule Based Frame System (RBFS) is then used to instantiate more complex geometrical structures from symbolic multiresolution feature representations. Final interpretation is achieved by using knowledge of contextual relations between objects in the domain.

IRS: a hierarchical knowledge based system for aerial image interpretation

A computer-based configuration and planning system

An approach toward improving the accessbility of the knowledge and information structures of expert systems is described; it is based upon a foundation development environment called the Rule-Based Frame System (RBFS), which forms the kernel of a larger system, IDEAS. RBFS is a knowledge representation language, within which a distinction is drawn between information which represents the world or domain, and knowledge which states how to make conclusions based upon the domain. Information takes the form of frames, for system processing, but is presented to the user/developer as an associative network via a Visual Editor for the Generation of Associative Networks (VEGAN). Knowledge takes the form of production rules, which are connected at suitable points in the domain model, but again it is presented to the user via a graphical interface known as the Knowledge Encoding Tool (KET). KET is designed to assist in knowledge acquisition in expert systems. It uses a combination of decision support trees and associative networks as its representation. A combined use of VEGAN and KET will enable domain experts to interactively create and test their knowledge base with minimum involvement on behalf of a knowledge engineer. An inclusion of learning features in VEGAN/KET is desirable for this purpose. The main objective of these tools, therefore, is to encourage rapid prototyping by the domain expert. VEGAN and KET are implemented in the Poplog environment on SUN 3/50 workstations.

Integrated graphical approach to knowledge representation and acquisition

The paper presents a practical evaluation of a particular type of expert planning system, known as XPERT. This has been designed to support interactive knowledge-based reasoning of expedition strategies that automatically apply preferred levels of investment to restore projects to meet their original (or improved) deadlines. The evaluation was performed in structured fashion by experienced project managers, who used XPERT in parallel with traditional computing methods on a variety of practical projects. The paper gives conclusions as to the efficacy of the system and recommendations for the system's future development. The authors comment on how XPERT might become part of the vanguard for third-generation computer-aided project-management tools.

http://ieeexplore.ieee.org/iel1/2202/1627/00042756.pdf

Practical evaluation of an intelligent knowledge-based project control system

Analytics and analysis from a massive database using various approaches and techniques are experimented, and ongoing research brings its main focus toward the domain such as big data. Economic growth and technological growth combined with its data production are also highlighted in big data approaches. Data are analyzed from social media, online stock markets, healthcare data, etc., which can be collaborated with artificial intelligence by developing the automated learning algorithms and development in cloud computing as well. Data can be either discrete or continuous, which are independent of the various processes for understanding the decision-making that relies on knowledge engineering. The proposed work converges in transforming the observed sequential data analysis from weather forecasting dataset. These systems can perform the cognitive task in improving the performance along with integrity of data using the enhanced framework. The prediction of natural disasters is a challenge for customers accessing forecast data, since fluctuations in data occur frequently, which fail to update the localization, that are identified as sensor latitude and longitude that are updated as a sequence on regular intervals from various directions. These four hidden states are the features that differentiate the probability of distributions for calculating the best cognitive tasks. Improved Bayesian Hidden Markov Frameworks (IBHMFs) have been proposed to identify the exact flow of state and detect the high congestion, which leads to earthquakes, tremor, etc. As the data from the analysis are unsupervised and features are converted to discrete and sequential data (independent variables), IBHMF can utilize in increasing the performance and produce the accuracy results in state estimation. 

John T. Boardman

Papers

A methodology for knowledge engineering using an interactive graphical tool for knowledge modelling