Dejing Dou

Ontology-based information extraction: An introduction and a survey of current approaches

Information extraction (IE) aims to retrieve certain types of information from natural language text by processing them automatically. For example, an IE system might retrieve information about geopolitical indicators of countries from a set of web pages while ignoring other types of information. Ontology-based information extraction (OBIE) has recently emerged as a subfield of information extraction. Here, ontologies - which provide formal and explicit specifications of conceptualizations - play a crucial role in the IE process. Because of the use of ontologies, this field is related to knowledge representation and has the potential to assist the development of the Semantic Web. In this paper, we provide an introduction to ontology-based information extraction and review the details of different OBIE systems developed so far. We attempt to identify a common architecture among these systems and classify them based on different factors, which leads to a better understanding on their operation. We also discuss the implementation details of these systems including the tools used by them and the metrics used to measure their performance. In addition, we attempt to identify the possible future directions for this field.

Ontology translation on the semantic web

Ontologies are a crucial tool for formally specifying the vocabulary and relationship of concepts used on the Semantic Web. In order to share information, agents that use different vocabularies must be able to translate data from one ontological framework to another. Ontology translation is required when translating datasets, generating ontology extensions, and querying through different ontologies. OntoMerge, an online system for ontology merging and automated reasoning, can implement ontology translation with inputs and outputs in OWL or other web languages. Ontology translation can be thought of in terms of formal inference in a merged ontology. The merge of two related ontologies is obtained by taking the union of the concepts and the axioms defining them, and then adding bridging axioms that relate their concepts. The resulting merged ontology then serves as an inferential medium within which translation can occur. Our internal representation, Web-PDDL, is a strong typed first-order logic language for web application. Using a uniform notation for all problems allows us to factor out syntactic and semantic translation problems, and focus on the latter. Syntactic translation is done by an automatic translator between Web-PDDL and OWL or other web languages. Semantic translation is implemented using an inference engine (OntoEngine) which processes assertions and queries in Web-PDDL syntax, running in either a data-driven (forward chaining) or demand-driven (backward chaining) way.

Ontology translation by ontology merging and automated reasoning

Ontologies are a crucial tool for formally specifying the vocabulary and relationship of concepts used on the Semantic Web. In order to share information, web-based agents that use different vocabularies must be able to translate data from one ontological framework to another, both syntactically and semantically. This dissertation describes ontology translation in three categories: dataset translation, ontology extension generation, and querying through different ontologies. The approach that has been proposed in this work is: ontology translation by ontology merging and automated reasoning. Ontology translation can be thought of in terms of formal inference. The merge of two related ontologies is obtained by taking the union of the concepts and the axioms defining them, and then adding bridging axioms that relate their concepts. The resulting merged ontology serves as an inferential medium for ontology translation. Web-PDDL, a strongly typed first-order logic language for the Semantic Web, can work as the internal representation for ontology merging and ontology translation. The syntactic translation can be done by an automatic syntax translator between Web-PDDL and other web agent languages. The semantic translation can be implemented by an inference engine, OntoEngine, a first order theorem prover with equality substitutions, running in both forward and backward chaining ways. The approach has also been extended to handle conditional fact translation, axiom derivation, ontology composition and data integration for web-based databases.

Ontology-based integration for relational databases

In this paper, we show that representation and reasoning techniques used in traditional knowledge engineering and the emerging Semantic Web can play an important role for heterogeneous database integration. Our OntoGrate architecture combines ontology-based schema representation, first order logic inference, and some SQL wrappers to integrate two sample relational databases. We define inferential data integration as the theoretical framework for our approach. The performance evaluation for query answering shows that OntoGrate reformulates conjunctive queries and retrieves over 100,000 answers from a target database in under 30 seconds. In addition to query answering, the system translates 40,000 database facts from source to target in under 30 seconds.

Calculating Feature Weights in Naive Bayes with Kullback-Leibler Measure

Naive Bayesian learning has been popular in data mining applications. However, the performance of naive Bayesian learning is sometimes poor due to the unrealistic assumption that all features are equally important and independent given the class value. Therefore, it is widely known that the performance of naive Bayesian learning can be improved by mitigating this assumption, and many enhancements to the basic naive Bayesian learning have been proposed to resolve this problem including feature selection and feature weighting. In this paper, we propose a new method for calculating the weights of features in naive Bayesian learning using Kullback-Leibler measure. Empirical results are presented comparing this new feature weighting method with some other methods for a number of datasets.

Integrating Databases into the Semantic Web through an Ontology-Based Framework

To realize the Semantic Web, it will be necessary to make existing database content available for emerging Semantic Web applications, such as web agents and services, which use ontologies to formally define the semantics of their data. Our research in the design and implementation of an ontology-based system, OntoGrate, addresses the critical and challenging problem of supporting human experts in multiple domains to interactively integrate information that is heterogenous in both structure and semantics. Databases, knowledge bases, the World Wide Web, and the emerging Semantic Web are some of the resources for which scalable integration remains a challenge. To integrate databases into the Semantic Web, we use Semantic Web ontologies to incorporate database schemas. An expressive first order ontology language, Web-PDDL, is used to define the structure, semantics, and mappings of data resources. A powerful inference engine, OntoEngine, can be used for query answering and data translation. In this paper, besides introducing new ideas in the OntoGrate system, we will elaborate on two case studies for which our system works well.

Development of NeuroElectroMagnetic ontologies(NEMO): a framework for mining brainwave ontologies

Event-related potentials (ERP) are brain electrophysiological patterns created by averaging electroencephalographic (EEG) data, time-locking to events of interest (e.g., stimulus or response onset). In this paper, we propose a generic framework for mining anddeveloping domain ontologies and apply it to mine brainwave (ERP) ontologies. The concepts and relationships in ERP ontologies can be mined according to the following steps: pattern decomposition, extraction of summary metrics for concept candidates, hierarchical clustering of patterns for classes and class taxonomies, and clustering-based classification and association rules mining for relationships (axioms) of concepts. We have applied this process to several dense-array (128-channel) ERP datasets. Results suggest good correspondence between mined concepts and rules, on the one hand, and patterns and rules that were independently formulated by domain experts, on the other. Data mining results also suggest ways in which expert-defined rules might be refined to improve ontologyrepresentation and classification results. The next goal of our ERP ontology mining framework is to address some long-standing challenges in conducting large-scale comparison and integration of results across ERP paradigms and laboratories. In a more general context, this work illustrates the promise of an interdisciplinary research program, which combines data mining, neuroinformatics andontology engineering to address real-world problems.

Semantic data mining: A survey of ontology-based approaches

Semantic Data Mining refers to the data mining tasks that systematically incorporate domain knowledge, especially formal semantics, into the process. In the past, many research efforts have attested the benefits of incorporating domain knowledge in data mining. At the same time, the proliferation of knowledge engineering has enriched the family of domain knowledge, especially formal semantics and Semantic Web ontologies. Ontology is an explicit specification of conceptualization and a formal way to define the semantics of knowledge and data. The formal structure of ontology makes it a nature way to encode domain knowledge for the data mining use. In this survey paper, we introduce general concepts of semantic data mining. We investigate why ontology has the potential to help semantic data mining and how formal semantics in ontologies can be incorporated into the data mining process. We provide detail discussions for the advances and state of art of ontology-based approaches and an introduction of approaches that are based on other form of knowledge representations.

Discovering executable semantic mappings between ontologies

Creating executable semantic mappings is an important task for ontology-based information integration. Although it is argued that mapping tools may require interaction from humans (domain experts) for best accuracy, in general, automatic ontology mapping is an AI-Complete problem. Finding matchings (correspondences) between the concepts of two ontologies is the first step towards solving this problem but matchings are normally not directly executable for data exchange or query translation. This paper presents an systematic approach to combining ontology matching, object reconciliation and multi-relational data mining to find the executable mapping rules in a highly automatic manner. Our approach starts from an iterative process to search the matchings and do object reconciliation for the ontologies with data instances. Then the result of this iterative process is used for mining frequent queries. Finally the semantic mapping rules can be generated from the frequent queries. The results show our approach is highly automatic without losing much accuracy compared with human-specified mappings.

An internet routing forensics framework for discovering rules of abnormal BGP events

Abnormal BGP events such as attacks, misconfigurations, electricity failures, can cause anomalous or pathological routing behavior at either global level or prefix level, and thus must be detected in their early stages. Instead of using ad hoc methods to analyze BGP data, in this paper we introduce an Internet Routing Forensics framework to systematically process BGP routing data, discover rules of abnormal BGP events, and apply these rules to detect the occurrences of these events. In particular, we leverage data mining techniques to train the framework to learn rules of abnormal BGP events, and our results from two case studies show that these rules are effective. In one case study, rules of worm events discovered from the BGP data during the outbreaks of the CodeRed and Nimda worms were able to successfully detect worm impact on BGP when an independent worm, the Slammer, subsequently occurred. Similarly, in another case study, rules of electricity blackout events obtained using BGP data from the 2003 East Coast blackout were able to detect the BGP impact from the Florida blackout caused by Hurricane Frances in 2004.