Evaluation can be defined as a process of determining the significance of a research output. This is usually done by devising a well-structured study on this output using one or more evaluation measures in which a careful inspection is performed. This paper presents a review of evaluation techniques for Conversational Agents (CAs) and Natural Language Interfaces to Databases (NLIDBs). It then introduces the developed customized evaluation methodology for Conversation-Based Interface to Relational Databases (C-BIRDs). The evaluation methodology created has been divided into two groups of measures. The first is based on quantitative measures, including two measures: task success and dialogue length. The second group is based on a number of qualitative measures, including: prototype ease of use, naturalness of system responses, positive/negative emotion, appearance, text on screen, organization of information, and error message clarity. Then an elaboration is carried out on the devised methodology by adding a discussion and recommendations on the sample size, the experimental setup and the scaling in order to provide a comprehensive evaluation methodology for C-BIRDs. In conclusion the evaluation methodology created is better way for identifying the strengths and weaknesses of C-BIRDs in comparison to the usage of single measure evaluations.

One of the challenges in data mining practices is that the datasets vary in complexity and often have different characteristics such as number of attributes, dependent variables characteristics etc. In terms of regression problems, the features that describe the dataset will vary in their complexity, sparseness verses coverage in relation to the decision space, and the number of outcome classes. Fuzzy Decision trees are well-established classifiers in terms of building robust, representative models of the domain. In order to represent different perspectives of the same domain, fuzzy trees can be used to construct fuzzy decision forests to enhance the predictive ability of singular trees. This paper describes an empirical study which examines the applicability of fuzzy tree regression forests to seven different datasets which have complex properties. The relationship between dataset characteristics and the performance of fuzzy regression tree forests is debated.

Fuzzy decision forests aim to improve the predictive power of single fuzzy decision trees by allowing multiple views of the same domain to be modelled. Such forests have been successfully created for classification problems where the outcome field is discrete; however predicting a continuous output value is more challenging in combining the output from multiple fuzzy decision trees. This paper presents a new approach to creating fuzzy regression tree forests based upon the induction of multiple fuzzy regression decision trees from one training sample, where each tree will represent a different view of the data domain. The singular fuzzy regression trees are induced using a proven algorithm known as Elgasir which fuzzifies crisp CHAID decision trees using trapezoidal membership functions for fuzzification and applies Takagi-Sugeno inference to obtain the final predicted values. A modified version of Artificial Immune System Network model (opt-aiNet) is then used for the simultaneous optimization of the membership functions across all trees within the forest. A strength of the proposed method is that data does not require fuzzification before forest induction this reducing pre-processing time and the need for subjective human experts. Five problem sets from the UCI repository and KEEL repository are used to evaluate the approach. The experimental results have shown that fuzzy regression tree forests reduce the error rate compared with single fuzzy regression trees.