Q2. How many sets of construction-related cases were available?
In total, 1105 sets of construction-related cases were available, of which 550 from years 1991 to 1995 were used for training, 275 from years 1996 to 1997 were used for testing, and 280 from years 1998 to 2000 were used to validate the network results with the observations.
Q3. What are the main factors that can affect the construction industry?
Keywords: particle swarm optimization, artificial neural networks, construction claimsinterrelation with a multitude of factors, the construction industry is particularly vulnerable to litigation.
Q4. What is the key contribution of the presented research and the unique works done by the author?
A key contribution of the presented research and the unique works done by the author is the adoption of the PSO-based AI techniques tailoring for the prediction of construction litigation outcomes, which is a field where new technological aids are rarely applied.
Q5. What is the concept of fitness in evolutionary computation paradigms?
As in evolutionary computation paradigms, the concept of fitness is employed and candidate solutions to the problem are termed particles or sometimes individuals, each of which adjusts its flying based on the flying experiences of both itself and its companions.
Q6. What is the use of the back-propagation algorithm?
The back-propagation with Levenberg-Marquardt (LM) algorithm under the neural network toolbox in MATLAB software (MATLAB, 2001) is employed as the benchmarking tool for comparison.
Q7. What is the current setting of disputes resolution in Hong Kong?
In Hong Kong, the current setting of disputes resolution is such that the processes of mediation, arbitration, and the courts should be followed successively (Chau, 1992).
Q8. How many neurons are used in the input layer?
a perceptron with an input layer with thirty neurons, a hidden layer with fifteen neurons, and output layer with six neurons, is adopted.
Q9. What is the potential for honest misunderstanding?
since a project usually involves thousands of separate pieces of work items to be integrated together to constitute a complete functioning structure, the potential for honest misunderstanding is extremely high.
Q10. What can be used to predict the outcome of construction claims?
Recent artificial intelligence techniques can be used to identify the hidden relationships among various interrelated factors and to predict decisions that will be made by the court, based on characteristics of cases and the corresponding past court decisions.
Q11. What is the paradigm of the PSO algorithm?
Its paradigm can be implemented in simple form of computer codes and is computationally inexpensive in terms of both memory requirements and speed.
Q12. What is the paradigm of the PSO algorithm?
Its paradigm can be implemented in simple form of computer codes and is computationally inexpensive in terms of both memory requirements and speed.
Q13. What is the main purpose of this paper?
This paper presents a PSO-based neural network approach for prediction of the outcome of construction litigation, based on court decisions in the last 10 years in Hong Kong.
Q14. What is the fitness value of the i-th particle?
the vector of the position of the previous best fitness value of any particle is represented by},{ ]2[]1[ iii PPP = (2)where Pi[1] and Pi[2] represent the position of the previous best fitness value of the i-th particle, between the input layer and the hidden layer, and that between the hidden layer and the output layer, respectively.
Q15. What are the characteristics of the case application that make PSO more suitable than traditional BP?
Special characteristics of the case application that make PSO more suitable than traditional BP include the sufficient amount of the data during the 10 years and the subtle inter-relationships among various principal parameters in determining the outcomes of construction litigation.