Dr. Corrado Mencar is Assistant Professor at the Department of Informatics, University of Bari (Italy). He received his laurea (Ms.Sci degree) in Informatics in 2000 at the University of Bari and his Ph.D. in 2005 at the same university, defending a thesis entitled "Theory of Fuzzy Information Granulation: Contributions to Interpretability Issues".
The scientific activity of Corrado Mencar is mainly focused on three fields. In the field of neuro-fuzzy systems, the research unfolds to the study of learning processes of neuro-fuzzy entworks, the cost functions used in such processes, the validity and interpretability of the resulting rule bases. In the field of Granular Computing, Corrado Mencar has studied and developed algorithms for extracting interpretable information granules from data and he is currently investigating methods for assessing co-intesion of fuzzy information granules with human knowledge. Recently, Corrado Mencar has also been investingating the field of user modeling, with special emphasis on user clustering based on preference similarities and user profiling and filtering based on fuzzy operators.
Corrado Mencar is author of more than 50 international, peer-reviewed scientific publications; he is reviewer for high impact-factor journals, he serves as associate editor for the International Journal of Artificial Intelligence and attended to several international conferences, where he has been speaker or session chair. Finally, Corrado Mencar has been involved in several national-wide research projects and is teacher in several academic undergraduate courses.
Abstract: This paper presents a user profile modelling approach based on fuzzy logic techniques. The proposed approach is conceived to find application in the context of e-learning processes, with the aim of providing personalised contents to different categories of users. Several concepts are introduced and formalised within a peculiar mathematical framework which bases its working engine on an innovative scheme of metadata exploiting the expressive strength of fuzzy sets. Along with the formal presentation of the profile modelling approach, suitable examples of application are provided for the sake of illustration.
Notes: Computational Science and Its Applications – ICCSA 2008
Abstract: The research activity described in this paper concerns the personalisation process in e-learning contexts. Particular emphasis is laid on the mechanisms of user profiling and association between user profiles and pedagogical resources. A particular profiling model is proposed where both the pedagogical resources and the user profiles are described in terms of a fuzzy valued metadata specification. The adoption of specific fuzzy operators enables the proposed model to perform associations with a high degree of flexibility, yielding a customised resource allocation for each user.
Notes: Knowledge-Based Intelligent Information and Engineering Systems
Abstract: Information granules are complex entities that arise in the process of abstraction of data and derivation of knowledge. The automatic generation of information granules from data is an important task, since it gives to machines the ability of acquiring knowledge that can be communicated to users. For this purpose, knowledge acquisition should provide for fuzzy information granules that can be naturally labelled by linguistic terms, i.e. symbols that belong to the natural language. Information granules of this type are called interpretable. However, interpretability cannot be guaranteed until a set of constraints is imposed on the granulation process. In literature, several interpretability constraints have been proposed, but, due to the subjective interpretation of interpretability, there is no agreement on which constraints should be adopted. This survey is an attempt to provide for a complete presentation of interpretability constraints adopted in literature with the following objectives: (i) to give a homogeneous description of all interpretability constraints; (ii) to provide for a critical review of such constraints; (iii) to identify potentially different meanings of interpretability. Hopefully, this survey may serve as a guidance for designing interpretable fuzzy models as well as for identifying new methods of interpretable information granulation.
Abstract: Distinguishability is a semantic property of fuzzy sets that has a great relevance in the design of interpretable fuzzy models. Distinguishability has been mathematically defined through different measures, which are addressed in this paper. Special emphasis is given to similarity, which exhibits sound theoretical properties but its calculation is usually computationally intensive, and possibility, whose calculation can be very efficient but it does not exhibit the same properties of similarity. It is shown that under mild conditions – usually met in interpretable fuzzy modelling – possibility can be used as a valid measure for assessing distinguishability, thus overcoming the computational inefficiencies of similarity measures. Moreover, procedures that minimize possibility also minimize similarity and, consequently, improve distinguishability. In this sense, the use of possibility is fully justified in interpretable fuzzy modeling.
Abstract: Data Mining, a central step in the broader overall process of Knowledge Discovery from Databases, concerns with discovering useful properties, called patterns, from data. Understandability is an essential — yet rarely tackled — feature that makes resulting patterns accessible by end users. In this paper we argue that the adoption of Fuzzy Logic for Data Mining can improve understandability of derived patterns. Indeed, Fuzzy Logic is able to represent concepts in a “human-centric” way. Hence, Data Mining methods based on Fuzzy Logic may potentially meet the so-called “Comprehensibility Postulate”, which characterizes the blurry notion of understandability. However, the mere adoption of Fuzzy Logic for Data Mining is not enough to achieve understandability. This paper describes and comments a number of issues that need to be addressed to provide for understandable patterns. A careful consideration of all such issues may end up in a systematic methodology todiscover comprehensible knowledge from data.
Abstract: In this paper we address the issue of designing optimal fuzzy interfaces, which are fundamental components of a fuzzy inference system. Due to the different roles of input and output interfaces, optimality conditions are analyzed separately for the two types of interface. We prove that input interfaces are optimal when based on a particular class of fuzzy sets called â??bi-monotonicâ??, provided that mild conditions hold. The class of bi-monotonic fuzzy sets covers a broad range of fuzzy sets shapes, including convex fuzzy sets, so that the provided theoretical results can be applied to several fuzzy models. Such theoretical results are not applicable to output interfaces, for which a different optimality criterion is proposed. Such criterion leads to the definition of an optimality degree that measures the quality of a fuzzy output interface. Illustrative examples are presented to highlight the features of the proposed optimality degree in assessing the quality of output interfaces.
Abstract: In this paper, we describe a method to derive prediction intervals for neuro-fuzzy networks used as predictive systems. The method also enables the definition of prediction intervals for the fuzzy rules that constitute the rule base of the neuro-fuzzy network, resulting in a more readable and robust knowledge base. Moreover, the method does not depend on a specific architecture and can be applied to a variety of neuro-fuzzy models. An illustrative example and a real-world case study are reported to show the effectiveness of the proposed method.
Abstract: In this paper a neuro-fuzzy modeling framework is proposed, which is devoted to discover knowledge from data and represent it in the form of fuzzy rules. The core of the framework is a knowledge extraction procedure that is aimed to identify the structure and the parameters of a fuzzy rule base, through a two-phase learning of a neuro-fuzzy network. In order to obtain reliable and readable knowledge, two further stages are integrated with the knowledge extraction procedure: a pre-processing stage, performing variable selection on the available data to obtain simpler and more reliable fuzzy rules, and a post-processing stage, that granulates outputs of the extracted fuzzy rules so as to provide a validity range of estimated outputs. Moreover, the framework can address complex multi-input multi-output problems. In such case, two distinct modeling strategies can be followed with the opportunity of producing both a single MIMO model or a collection of MISO models. The proposed framework is verified on a real-world case study, involving prediction of chemical composition of ashes produced by combustion processes carried out in thermo-electric generators located in Italy.
Notes: Special Issue on Fuzzy Sets in Knowledge Discovery
Abstract: Human-Centric Information Processing requires tight communication processes between users and computers. These two actors, however, traditionally use different paradigms for representing and manipulating information. Users are more inclined in managing perceptual information, usually expressed in natural language, whilst computers are formidable number-crunching systems, capable of manipulating information expressed in precise form. Fuzzy information granules could be used as a common interface for communicating information and knowledge, because of their ability of representing perceptual information in a computer manageable form. Nonetheless, this connection could be established only if information granules are interpretable, i.e. they are semantically co-intensive with human knowledge. Interpretable information granulation opens several methodological issues, regarding the representation and manipulation of information granules, the interpretability constraints and the granulation processes. By taking into account all such issues, effective Information Processing systems could be designed with a strong Human-Centric imprint.
Abstract: This contribution presents a user profile modelling approach based on fuzzy logic techniques. The proposed approach is conceived to find application in various contexts, with the aim of providing personalised contents to different categories of users. Both contents and users are described by metadata, so a description language is introduced along with a formal model defining their association mechanism. The strength of the model is the use of the expressive flexibility of fuzzy sets exploited by an innovative scheme of metadata. Along with the formal presentation of the profile modelling approach, the design of a software system based on a Service Oriented Architecture is presented. The system exposes a number of services to be consumed by information systems for personalized content access. In this way the system can be used in different application contexts.
Abstract: DC* (Double Clustering by A*) is an algorithm for interpretable fuzzy information granulation of data. It is mainly based on two clustering steps. The first step applies the LVQ1 algorithm to find a suitable representation of data relationships. The second clustering step is based on the A* search strategy and is aimed at finding an optimal number of fuzzy granules that can be labeled with linguistic terms. As a result, DC* is able to linguistically describe hidden relationships among available data. In this paper we propose an extension of the DC* algorithm, called DC , which improves the generalization ability of the original DC* by modifying the A* search procedure. This variation, inspired by Support Vector Machines, results empirically effective as reported in experimental results.
Abstract: Adaptive e-learning systems are growing in popularity in recent years. These systems can offer personalized learning experiences to learners, by supplying each learner with
learning contents that meet his/her specific interests and needs. The efficacy of such systems is strictly related to the possibility of automatically deriving models encoding the preferences of each learner, analyzing their navigational behavior during their interactions with the system. Since learner preferences may change over time, there is the need to define mechanisms of dynamic adaptation of the learner models so as to capture the changing learner interests. Moreover, learner preferences are characterized by imprecision and gradedness. Fuzzy Set Theory provides useful tools to deal with these characteristics. In this paper a novel strategy is presented to derive and update learner
models by encoding preferences of each individual learner in terms of fuzzy sets. Based on this strategy, adaptation is continuously performed, but in earlier stages it is more sensitive to updates (plastic phase) while in later stages it is less sensitive (stable phase) to allow Learning-Object suggestion. Simulation results are reported to show the effectiveness of the proposed approach.
Abstract: We describe an automatic approach for evaluating interpretability of fuzzy rule-based classifiers. The approach is based on the logical view of fuzzy rules, which are interpreted as rows in truth tables. These truth tables are subject of a minimization procedure based on a variant of the Quine-McCluskey algorithm. The minimized truth tables are used to build new fuzzy rules, which are compared with the original ones in terms of classification accuracy. If the two sets of rules have similar performances, we deduce that the logical view of rules is applicable to the fuzzy classifier, which is hence considered interpretable. On the other hand, a significant difference in classification ability shows that fuzzy rules cannot be interpreted in logical terms, hence linguistic labelling may not be significant. Two illustrative examples show both the cases.
Abstract: The paper shows an application of neural networks for the prediction of water levels in artesian wells. The design of the neural network follows a systematic methodology, which can be used for a variety of prediction problems. A part of the design methodology is based on cross validation, which helped us in finding and correcting data anomalies due to different methods used for generating data. The final network is able to predict water level within the required tolerance, thus resulting in an effective decision support system to help managers in programming the exploitation of artesian wells in the short-term.
Abstract: Questo lavoro presenta uno studio sulla personalizzazione dei contenuti erogati in modalitàe-learning, soffermandosi sulla attivitàrelativa alla profilazione degli utenti e alla associazione tra profili utente e risorse didattiche. In particolare, Ú proposto un modello di profilazione in cui sia le risorse didattiche sia i profili degli utenti sono descritti attraverso metadati specificati da valori fuzzy. L’uso di valori e operatori fuzzy consente la creazione di associazioni flessibili tra profili e risorse sulla base di un grado di compatibilità, consentendo una erogazione altamente personalizzata dei contenuti per ciascun utente
Abstract: In this paper we present an approach for extracting interpretable information granules for classification. The approach, called DC¿ (Double Clustering with Genetic Algorithms) is based on two clustering steps. The first step uses LVQ1 to identify cluster prototypes in the multidimensional data space so as to represent hidden relationships among data. In the second step a genetic algorithm is applied to the projections of these prototypes with the objective of finding a minimal number of fuzzy information granules that verify some interpretability constraints. The key feature of DC¿ is the efficiency of the minimization process carried out in the second step. Experimental results on two medical diagnosis problems show the effectiveness of the proposed approach in terms of accuracy, interpretability and efficiency.
Abstract: The World Wide Web is continuously growing, being today one of the biggest repositories ever built. As a consequence, there is an increasing need of making Web sites adaptive, i.e. capable of providing different content/structure to users according to their needs and preferences. Analyzing the visitor behavior is a fundamental part in the design of an adaptive Web site. A common source of information about the visitor browsing behavior is represented by Web server log files. Processing these huge files of raw web log data in order to retrieve significant information about the navigational behavior of users is not a trivial task. In this paper, we present LODAP, a web LOg DAta Preprocessor which is able to identify actions (and hence preferences) of individual users during navigation through a Web site. To derive a model of the browsing behavior, LODAP processes a log file by activating several modules in sequence. Firstly, the data cleaning module cleans the log file from all irrelevant entries corresponding to resources not explicitly requested by the user. Then, the data structuration module groups relevant requests into user sessions, by using a time-based method. Once user sessions have been identified, LODAP uses time information about resources accessed in each session to evaluate the degree of interest for each resource requested by the user during navigation through the Web site. Finally, the data filtering module reduces the size of data concerning the extracted user sessions by removing very low support requests, i.e. requests to those resources which appear in a low number of sessions. Tests on the log files of a specific Web site are reported to show how the proposed log data preprocessor can condense raw log data into a collection of significant user sessions from which a behavior model can be easily derived.
Abstract: User profiling is a fundamental task in Web personalization. Fuzzy clustering is a valid approach to derive user profiles by capturing similar user interests from web usage data available in log files. Often, fuzzy clustering is based on the assumption that data lay on an Euclidean space; however, clustering based on Euclidean distance can lead the clustering process to find user representations that do not capture the semantic information incorporated in the original Web usage data. In this paper, we propose a different approach to express similarity between Web users. The measure is based on the evaluation of similarity between fuzzy sets. The proposed measure is employed in a relational fuzzy clustering algorithm to discover clusters embedded in the Web usage data and derive profiles modeling the real user preferences. An application example on usage data extracted from log files of a sample Web site is reported and a comparison with the results obtained using the cosine measure is shown to demonstrate the effectiveness of the proposed similarity measure.
Abstract: In this paper we present a multi-level approach for extracting well-defined and semantically sound information granules from numerical data. The approach is based on the Double Clustering framework (DCf), which performs two main clustering steps on the data space in order to extract granules qualitatively described in terms of fuzzy sets that meet a number of interpretability constraints. While DCf can extract information granules with a fixed level of granulation, its multi-level extension, called ML-DC (Multi-Level Double Clustering), can perform granulation of data at different levels, in a hierarchical fashion. At the first level, the whole dataset is granulated. At the second level, data embraced in each firstlevel granule are further granulated taking into account the context generated by that granule. The hierarchical collection of granules derived via ML-DC is then used to construct a committee of fuzzy inference systems that can approximate any I/O mapping with a good balance between accuracy and interpretability.
Abstract: In this paper, we present a framework for extracting well-defined and semantically sound information granules. The framework is mainly centered on a double clustering process, hence, it is called DCf (double clustering framework). A first clustering process identifies cluster prototypes in the multidimensional data space, then the projections of these prototypes are further clustered along each dimension to provide a granulation of data. Finally, the extracted granules are described in terms of fuzzy sets that meet interpretability constraints so as to provide a qualitative description of the information granules. Different implementations of DCf are presented and compared on a medical diagnosis problem to show the utility of the proposed framework.
