Mihai Gabroveanu

An Atanassov's intuitionistic fuzzy reasoning model

The task of the standard Mamdani fuzzy logic controller is to find a crisp control action from the fuzzy rule-base and from a set of crisp inputs. In this paper we modify this controller in order to work with Atanassovs intuitionistic fuzzy sets and to activate a set of rules having the same conclusion. Usually, the inference rules used in a fuzzy logic controller are given by a domain expert; in our system, these rules are automatically induced as fuzzy association rules starting from a training set. The fuzzy confidence value associated with each rule is used to obtain the fuzzy set inferred by our system.

Intuitionistic Fuzzy Control Based on Association Rules

The task of the standard Mamdani fuzzy logic controller is to find a crisp control action from the fuzzy rule-base and from a set of crisp inputs. In this paper we modify this controller in order to work with intuitionistic fuzzy sets and to activate a set of rules having the same conclusion. Usually, the inference rules used in a fuzzy logic controller are given by a domain expert; in our system, these rules are automatically induced as fuzzy association rules starting from a training set. The fuzzy confidence value associated with each rule is used to obtain the fuzzy set inferred by our system.

Towards Using Grid Services for Mining Fuzzy Association Rules

The knowledge grid provides us with the means to represent, share and manage globally distributed knowledge resources over the Internet. The process of analyzing geographically distributed large data sets for extracting novel and interesting patterns or models has become a major challenge during the last years. The knowledge grid infrastructure allows developing more efficient data mining applications that analyze this distributed datasets. The paper presents the design of a distributed algorithm for mining fuzzy association rules from the distributed databases over the grid.

Mining fuzzy association rules using MapReduce technique

Mining association rules from large databases is one of the most important tasks from data mining. Nowadays, the majority of companies produce a significant amount of data stored in distributed databases. In this case, most of the traditional algorithms for mining association rules become ineffective because they require a lot of resources to extract the frequent patterns. The cloud computing technologies provide us the infrastructure for handling such massive datasets. In this paper, we propose an extension of the Count Distribution algorithm for mining fuzzy association rules from a distributed database. The algorithm uses the MapReduce programming model, which aims to distribute the mining process over many cluster nodes. Distributing the mining process allows handling very large databases and significantly improves the execution time.

HTML Pattern Generator--Automatic Data Extraction from Web Pages

Existing methods of information extraction from HTML documents include manual approach, supervised learning and automatic techniques. The manual method has high precision and recall values but it is difficult to apply it for large number of pages. Supervised learning involves human interaction to create positive and negative samples. Automatic techniques benefit from less human effort but they are not highly reliable regarding the information retrieved

Implication-Based Support Measures for Fuzzy Association Rules

Several approaches generalizing association rules to fuzzy association rules have been proposed, so far. The quality measures have also been generalized. Such a measure is based on the so-called S-implicator. In this paper we point out some other implicators which can substitute the S-implicator in order to define new support measures which evaluate the quality of a fuzzy association rule, taking into consideration the non-contradicting rule examples (non-negative examples), while the classical measures use only positive examples.

Computing a Similarity Coefficient for Mining Massive Data Sets

Large amounts of data can be found today in all areas as a result of various processes like e-commerce transactions, banking or credit card transactions, or web navigation user sessions (recorded into web server logs). The development and implementation of algorithms able to process huge amounts of data have become more affordable due to cloud computing and the MapReduce programming model, which, in turn, enabled the development of some open-source frameworks, such as Apache Hadoop. Based on the values obtained by computing the Jaccard similarity coefficients for two very large graphs, we have analysed in this paper the connections and influences that certain nodes have over other nodes. Also, we have illustrated how the Apache Hadoop framework and the MapReduce programming model can be used for a large amount of computations.

RuleStore: Towards a Standard API for Rule Bases

Any rule-based system uses a knowledge base consists of a set of rules. Identifying a standard for storing and handling of rules and rulesets became a challenge. RuleML family of languages provides the interoperability framework for rules but does not specify any standard solution for rule storage and retrieval. In this paper we propose an model for a persistent storage for rules and an API specification for rules management. Our contribution aims to be aligned with the OMG specification towards a standard submission.

Experiments with computing similarity coefficient over big data

Big data is a hot topic nowadays due to the huge amount of data resulted from various commercial processes and also due to every day data handled by social networks. The MapReduce programming model focuses on processing and generating large data sets. Using the values obtained by computing the Jaccard similarity coefficients for two very large graphs, we have analysed the connections and influences that some nodes have over the other nodes. Furthermore, we have shown how Apache Hadoop framework and MapReduce programming model can be used for high volume computations. All tests were performed on a distributed cluster in order to obtain the results described in the paper.

WebKM - Online Data Mining System

The problem of using legacy information systems by making their services publicly available from the Web has become a tedious one due to the diversity of platforms (operating systems, programming languages, database management systems) on which they were implemented. In this paper we present an online system, WebKM, following the client-server paradigm, that allows executing data mining tasks from a thin client (Web Browser). An user is thus released from the management care of local installed software.