Abstract Clustering algorithms are essential in data analysis, but evaluating their performance is challenging when the true labels are not available, especially for non-convex clusters. Traditional performance evaluation metrics struggle to identify clustering quality, often assigning higher scores for linearly separated clusters than the true clusters. We propose an original approach to distance computation that accounts for the data structure, thus improving the clustering quality evaluation for non-convex clusters without affecting other shapes of clusters. We also showcase the applicability of this method through a modified version of K-Means using the proposed method that is capable of correctly separating non-convex clusters. The validation included the analysis of performance and time complexity of 3 traditional clustering quality evaluation metrics and the K-Means clustering algorithm against their augmented versions with the proposed approach. This analysis conducted on 7 benchmark synthetic datasets and 6 real datasets with various numbers of examples and features of diverse characteristics and joint complexities: simple convex clusters, overlapped and imbalanced clusters, and non-convex clusters. Through these analyses, we show the ineffectiveness of traditional methods and that the proposed approach overcomes the weaknesses of traditional methods.

In this paper we address the problem of identifying contradictions by opinion mining across documents. Our approach involves opinion extraction and storage by processing natural language documents such as reviews, news etc. and aims the identification of contradictory opinions related to the same target expressed by the same holder or by different holders. By matching the structured representations of opinions we identify a potential inconsistency occurring in two documents that is signaled and further analysis is applied to confirm/infirm the contradiction. Moreover, communities are detected both on individual opinions and social data. Thus, the (in)consistency might be tracked for the holder as a member of a community, as well as for the holder as an individual. We addressed scalability by designing a cloud-based storage infrastructure and an efficient indexing system that allows for fast retrieval and matching of structured representations.

Nowadays, the volume of information increases exponentially, forcing the corporations to keep their business information distributed under several heterogeneous sources such as relational databases, spread sheets, XML documents and Web pages, and stored under different structures and formats. Integrating heterogeneous sources is recently acknowledged as an important vision on semantic web research. The concept of heterogeneity arises at different levels: from the lexical level to the semantic or structural level. For discovering and consolidating the semantic relationships among the semantically related data present in different types of databases and files, this paper presents the enhancements obtained due to the use of available online large lexical databases, combined with lexical and structural similarity models and the available source metadata. Finally, we reveal the experimental results that demonstrate the applicability and usability of our approach.

SUMMARY The clustering and sorting behavior of ants, as well as the foraging behavior of birds in nature represented sources of inspiration for designing clustering methods applicable in computer science. This paper investigates how biologically‐inspired clustering methods can be adapted to cluster Semantic Web services aiming at the efficiency of the discovery process. The methods consider the semantic similarity between services as the main clustering criterion. To measure the semantic similarity between two services, we propose a matching method that evaluates the degree of match between the semantic description of the two services. We have tested the biologically‐inspired clustering methods on the SAWSDL service retrieval test collection (SAWSDL‐TC) benchmark, and we have comparatively evaluated their performance using the Dunn index and the Average‐Item Cluster Similarity metric, the latter being introduced in this paper. Copyright © 2011 John Wiley & Sons, Ltd.

This paper presents a method for Web service clustering based on Particle Swarm Optimization aiming at the efficiency of the discovery process. The proposed method clusters services based on the similarity between their semantic descriptions. To evaluate the semantic similarity we have defined a set of metrics which compute the degree of match between two services. The proposed metrics take into consideration the hierarchical and property-based non-hierarchical relations between the concepts that semantically describe the input and output service parameters. These metrics can be applied to the exact, subsume and sibling match. To test our method for service clustering we have used the SAWSDL-TC service collection. The performance of the clustering method has been evaluated using the Dunn Index, Intra-Cluster Variance and Average-Item Cluster Similarity metrics.

This paper presents a Tabu search-based method for selecting the optimal or a near-optimal solution in semantic Web service composition. The proposed method is applied on an Enhanced Planning Graph structure which encodes all the composition solutions that satisfy a user request. The criteria for selecting the optimal solution include the QoS attributes and the semantic similarity between the services involved in composition. The Tabu search-based method was evaluated on scenarios from the trip planning domain.

In this paper we present a bee-inspired method for selecting the optimal composition solution. The proposed method uses a composition graph model and a matrix of semantic links to search for the optimal composition solution. For improving the performance of the traditional bee colony optimization algorithm a 1-OPT heuristic is defined. This makes the composition solutions more diverse so as to avoid the stagnation on local optimal solutions. The optimal composition solution is identified by using a multi-criteria fitness function. The fitness function evaluates a composition solution according to QoS attributes and the semantic quality between the services involved in a composition solution.

In this paper we approach the context adaptation problem by defining a self-healing model that uses a policy-driven reinforcement learning mechanism to take run-time decisions. The self-healing property is enforced by monitoring the system's execution environment for evaluating the degree of fulfilling the context policies in the current context situation, and selecting the healing actions that may be executed. For action selection a reinforcement learning based approach is used to generate the best sequence of actions to be taken for keeping the context resources as close as possible to a policy compliant state.

This paper addresses the problem of traceability in the context of a complex supply chain by proposing a model that captures the main elements necessary to follow a product throughout its lifecycle, from manufacturing to the end consumer. The model is general enough to be applied for several types of supply chains and focuses on external traceability, but also enables the retrieval of internal traceability data. We also present a broker-based service oriented architecture that provides logistics and traceability support according to the model. The traceability operations are performed on the products that travel along supply chains. The architecture targets a distributed environment, with the broker gathering and processing traceability data stored at different sites. We describe two approaches for traceability data retrieval, one for the logistic chains created and managed by the broker, and another one for the supply chains discovered in the EPCglobal Network.