On the Semantic Web, data will inevitably come from many different ontologies, and information processing across ontologies is not possible without knowing the semantic mappings between them. Manually finding such mappings is tedious, error-prone, and clearly not possible at the Web scale. Hence, the development of tools to assist in the ontology mapping process is crucial to the success of the Semantic Web. We describe GLUE, a system that employs machine learning techniques to find such mappings. Given two ontologies, for each concept in one ontology GLUE finds the most similar concept in the other ontology. We give well-founded probabilistic definitions to several practical similarity measures, and show that GLUE can work with all of them. Another key feature of GLUE is that it uses multiple learning strategies, each of which exploits well a different type of information either in the data instances or in the taxonomic structure of the ontologies. To further improve matching accuracy, we extend GLUE to incorporate commonsense knowledge and domain constraints into the matching process. Our approach is thus distinguished in that it works with a variety of well-defined similarity notions and that it efficiently incorporates multiple types of knowledge. We describe a set of experiments on several real-world domains, and show that GLUE proposes highly accurate semantic mappings. Finally, we extend GLUE to find complex mappings between ontologies, and describe experiments that show the promise of the approach.

Integrating heterogeneous resources of the web will require finding agreement between the underlying ontologies. A variety of methods from the literature may be used for this task, basically they perform pair-wise comparison of entities from each of the ontologies and select the most similar pairs. We introduce a similarity measure that takes advantage of most of the features of OWL-Lite ontologies and integrates many ontology comparison techniques in a common framework. Moreover, we put forth a computation technique to deal with one-to-many relations and circularities in the similarity definitions.

There is a great deal of research on ontology integration which makes use of rich logical constraints to reason about the structural and logical alignment of ontologies. There is also considerable work on matching data instances from heterogeneous schema or ontologies. However, little work exploits the fact that ontologies include both data and structure. We aim to close this gap by presenting a new algorithm (ILIADS) that tightly integrates both data matching and logical reasoning to achieve better matching of ontologies. We evaluate our algorithm on a set of 30 pairs of OWL Lite ontologies with the schema and data matchings found by human reviewers. We compare against two systems-the ontology matching tool FCA-merge [28] and the schema matching tool COMA++ [1]. ILIADS shows an average improvement of 25 % in quality over FCA-merge and a 11% improvement in recall over COMA++.

Due to the recent explosion of the amount of on-line accessible biomedical data and tools, finding and retrieving the relevant information is not an easy task. The vision of a Semantic Web for life sciences alleviates these difficulties. A key technology for the Semantic Web are ontologies. In recent years many biomedical ontologies have been developed and many of these ontologies contain overlapping information. To be able to use multiple ontologies they have to be aligned or merged. In this paper we propose a framework for aligning and merging ontologies. Further, we developed a system for aligning and merging biomedical ontologies (SAMBO) based on this framework. The framework is also a first step towards a general framework that can be used for comparative evaluations of alignment strategies and their combinations. In this paper we evaluated different strategies and their combinations in terms of quality and processing time and compared SAMBO with two other systems.

We develop a Bayesian framework for tackling the supervised clustering problem, the generic problem encountered in tasks such as reference matching, coreference resolution, identity uncertainty and record linkage. Our clustering model is based on the Dirichlet process prior, which enables us to define distributions over the countably infinite sets that naturally arise in this problem. We add supervision to our model by positing the existence of a set of unobserved random variables (we call these “reference types”) that are generic across all clusters. Inference in our framework, which requires integrating over infinitely many parameters, is solved using Markov chain Monte Carlo techniques. We present algorithms for both conjugate and non-conjugate priors. We present a simple—but general—parameterization of our model based on a Gaussian assumption. We evaluate this model on one artificial task and three real-world tasks, comparing it against both unsupervised and state-of-the-art supervised algorithms. Our results show that our model is able to outperform other models across a variety of tasks and performance metrics.

The performance of search engines crucially depends on their ability to capture the meaning of a query most likely intended by the user. We study the problem of mapping a search engine query to those nodes of a given subject taxonomy that characterize its most likely meanings. We describe the architecture of a classification system that uses a web directory to identify the subject context that the query terms are frequently used in. Based on its performance on the classification of 800,000 example queries recorded from MSN search, the system received the Runner-Up Award for Query Categorization Performance of the KDD Cup 2005. 1.

This paper presents a hybrid, extensional and asymmetric matching approach designed to find out semantic relations (equivalence and subsumption) between entities issued from two textual taxonomies (web directories or OWL ontologies). By using the association rule paradigm and a statistical measure developed in this context, this method relies on the following idea: “An entity A will be more specific than or equivalent to an entity B if the vocabulary (i.e. terms and data) used to describe A and its instances tends to be included in that of B and its instances”. This matching approach is divided into two parts: (1) The representation of each entity by a set of relevant terms and data; (2) The discovery of binary association rules between entities. The selection of rules uses two criteria. The first one permits to assess the implication quality by using implication intensity measure. The second criterion verifies the generativity of the rule and then permits to reduce redundancy. Finally, the proposed method is evaluated on two benchmarks. The first contains two conceptual hierarchies containing textual documents and the second one is composed of OWL ontologies. The experimentations show that the method obtains good precision values and also permits to discover meaningful subsumptions that are not taken into account by similarity-based approaches. 1

Abstract. Most of the schema matching tools are assembled from multiple match algorithms, each employing a particular technique to improve matching accuracy and making matching systems extensible and customizable to a particular domain. Recently, it has been pointed out that the main issue is how to select the most suitable match algorithms to execute for a given domain and how to adjust the multiple knobs (e.g. threshold, performance, quality, etc.). The solutions provided by current schema matching tools consist in aggregating the results obtained by several match algorithms to improve the quality of the discovered matches. However, aggregation entails several drawbacks. In this article, we present a novel method for combining schema matching algorithms. The matching engine makes use of a decision tree to combine the most appropriate match algorithms. As a first consequence of using the decision tree, the performance of the system is improved since the complexity is bounded by the height of the decision tree. Thus, only a subset of these match algorithms is used during the matching process. The second advantage is the improvement of the quality of matches. Indeed, for a given domain, only the most suitable match algorithms are used. The experiments show the effectiveness of our approach w.r.t. other matching tools. 1

Abstract: Schema matching is the process of identifying semantic correspondences between schemas. COMA++ is a matching prototype which uses several characteristics of schemas to determine similarities between them, for example the names and data types of the schema elements and structural information. In this paper we propose two instance-based matchers for COMA++ to gain a further quality improvement. The features of the matchers and first results are described. 1

Ontology mediation is a broad field of research which is concerned with determining and overcoming differences between ontologies in order to allow the reuse of such ontologies, and the data annotated using these ontologies, throughout different heterogeneous applications. Ontology mediation can be subdivided into three areas: ontology mapping, which is mostly concerned with the representation of correspondences between ontologies; ontology alignment, which is concerned with the (semi-)automatic discovery of correspondences between ontologies; and ontology merging, which is concerned with creating a single new ontology, based on a number of source ontologies. This chapter reviews the work which has been done in the three mentioned areas and proposes an integrated approach to ontology mediation in the area of knowledge management. A language is developed for the representation of correspondences between ontologies. An algorithm, which generalizes current state-of-the-art alignment algorithms, is developed for the (semi-)automated discovery of such mappings. A tool is presented for browsing and editing ontology mappings. An ontology mapping can be used for a variety of different tasks, such as transforming data between different representations and querying different heterogeneous knowledge bases. 1