Extracting Semantic-preserved Ontology from Relational Database

Ontology provides a shared and reusable piece of knowledge about a specific domain, and has been applied in many fields, such as semantic Web, e-commerce and information retrieval, etc. However, building ontology by hand is a very hard and error-prone task. Learning ontology from existing resources is a good solution. Because relational database is widely used for storing data and OWL is the latest standard recommended by W3C, this paper proposes an approach of learning OWL ontology from data in relational database. Compared with existing methods, the approach can acquire ontology from relational database automatically by using a group of learning rules instead of using a middle model. In addition, it can obtain OWL ontology, including the classes, properties, properties characteristics, cardinality and instances, while none of existing methods can acquire all of them. The proposed learning rules have been proven to be correct by practice.

Relational databases (RDB) are widely used as a backend for information systems, and contain interesting structured data (schema and data). In the case of ontology learning, RDB can be used as knowledge source. Multiple approaches exist for building ontologies from RDB. They mainly use schema mapping to transform RDB components to ontologies. Most existing approaches do not deal with recursive relationships that can encapsulate good semantics. In this paper, two technics are proposed for transforming recursive relationships to OWL2 components: (1) Transitivity mechanism and (2) Concept Hierarchy. The main objective of this work is to build richer ontologies with deep taxonomies from RDB.

Relational databases are valuable sources for ontology learning. Previous work showed how precise ontologies can be learned and be fruitfully exploited to solve practical problems, such as ensuring integration and interoperation of heterogeneous databases. However, a major persisting limitation of the existing approaches is the derivation of ontologies with flat structure that simply mirror the schema of the source databases. In this paper, we present the RTAXON learning method that shows how the content of the databases can be exploited to identify categorization patterns from which class hierarchies can be generated. This fully formalized method combines a classical database schema analysis with hierarchy mining in the stored data. RTAXON is one of the methods implemented in RDBToOnto, a comprehensive tool that support the transitioning process from access to the data to generation of populated ontologies.

Ontology is a crucial building block of semantic web, which is accepted as the most advanced knowledge representation model. But ontology learning is a big obstacle for its complexity and labor-denseness. We use rule-based information extraction (IE) to get instances from text. There is great challenge for the adaptivity of IE for ontology learning, so we put forward RGA-CIE - a rule generation algorithm which applies supervised learning with bottom-up strategy. RGA-CIE is a rule generalization process with a heuristic method to decide rule generalization path and laplacian* formula to evaluate the performance of rules. Empirical results show that our approach does be of use in learning of ontology instances.

There is a surge of interests in distributed computing thanks to advances in clustered computing and big data technology. My research explores topics on machine learning and big data technologies related to learning under decentralized resources. One topic of distributed learning is to distribute large scale centralized computation to clustered or multi-core computers. We propose a method for fast computation of kNN search, random projection forests (rpForests). RpForests finds nearest neighbors by combining multiple kNN-sensitive trees with each constructed recursively through a series of random projections. RpForests has a very low computational complexity as a tree-based methodology and achieves a remarkable accuracy in terms of fast decaying missing rate of kNNs and that of discrepancy in the k-th nearest neighbor distances, as demonstrated on many datasets. The ensemble nature of rpForests makes it easily parallelized to run on clustered or multi-core computers; the running time is shown to be nearly inversely proportional to the number of cores or machines. Another two topics treats the data in machine learning as a computing resource. Existing learning algorithms typically assume all the data to be in one centralized place while it is increasingly often that the data are located at a number of distributed sites, and we wish to learn over data from all the sites with low communication overhead. Also, it is often that the data of interest has features shared by some other datasets from multiple sources. It is desirable to take advantage of such auxiliary datasets. We proposed two approaches under this topic—fast communication-efficient spectral clustering overdistributed data and fuzzy join of data with shared features. A novel framework is proposed that enables computation over data from all the physical nodes, with minimal communications overhead while a major speedup in computation for spectral clustering. The loss in accuracy is negligible compared to the non-distributed setting. The proposed approach allows local parallel computing at where the data are located and the speedup is most substantial when the data are evenly distributed across sites. Experiments show almost no loss in accuracy with our approach while a 2x speedup under various settings with two distributed sites. As the transmitted data does not need to be in their original form, the framework readily addresses the privacy concern for data sharing in distributed computing. We propose another efficient algorithm fuzzy join that enhances the learning from the provided data by leveraging the auxiliary data through shared features. Fuzzy join enables the extraction of additional information along the dimension implied by features in the auxiliary data that are not in the given data. Our implementation based on random projection forests is efficient with log linear computational complexity, and is resistant to noises in the data. Experiments demonstrate the practicality of our approach. Fuzzy join extends the scope of the join operation in relational databases by performing join on non-index key columns and allowing non-exact matches between rows from different datasets.

The main claim of this paper is that machine learning can help integrate the construction of ontologies and extraction grammars and lead us closer to the Semantic Web vision. The proposed approach is a bootstrapping process that combines ontology and grammar learning, in order to semi-automate the knowledge acquisition process. After providing a survey of the most relevant work towards this goal, recent research of the Software and Knowledge Engineering Laboratory (SKEL) of NCSR “Demokritos” in the areas of Web information integration, information extraction, grammar induction and ontology enrichment is presented. The paper concludes with a number of interesting issues that need to be addressed in order to realize the advocated bootstrapping process.

The explosive growth of textual data on the web coupled with the increase on demand for ontologies to promote the semantic web, have made the automatic ontology construction from the text a very promising research area. Ontology learning (OL) from text is a process that aims to (semi-) automatically extract and represent the knowledge from text in machine-readable form. Ontology is considered one of the main cornerstones of representing the knowledge in a more meaningful way on the semantic web. Usage of ontologies has proven to be beneficial and efficient in different applications (e.g. information retrieval, information extraction, and question answering). Nevertheless, manually construction of ontologies is time-consuming as well extremely laborious and costly process. In recent years, many approaches and systems that try to automate the construction of ontologies have been developed. This paper reviews various approaches, systems, and challenges of automatic ontology construction from the text. In addition, it also discusses ways the ontology construction process could be enhanced in the future by presenting techniques from shallow learning to deep learning (DL).

Smart cities are designed to overcome problems that traditional cities face like the exponential increase of the population and related needs. Smart cities design is a relatively new research field that appeared with the concept of smart city. Unfortunately, it is a laborious and error prone task which suggests designers' expertise and adequate skills. To overcome this issue, we attempt to acquire expertise from existing strategies and projects about smart cities. For this end, we develop strategy ontology. The strategy ontology allows not only to formalise and conceptualise the strategy related concepts but also to analyse existing projects and strategies. We attempt thereafter to enrich it via automatic learning techniques. We use therefore automatic annotation of smart cities related documents. In addition to automatic annotation, we also use the ontology-based information extraction (OBIE) for ontology population and enrichment with new concepts and instances. The followed process is composed of five tasks: corpus construction, ontology population, ontology enrichment, ontology's consistency verification and ontology learning evaluation. The resulting ontology will allow sharing the gathered knowledge between people participating in the activities of smart cities design and thus learning from smart cities previous projects and strategies in order to create new ones.

This paper covers the first research activity in the field of automatic processing of business documents in Turkish. In contrast to traditional information extraction systems which process input text as a linear sequence of words and focus on semantic aspects, proposed approach doesn't ignore document layout information and benefits hints provided by layout analysis. In addition, approach not only checks relations of entities across document for verifying its integrity, but also verifies extracted information against real word data (e.g. customer database). This rule-based approach uses a morphological analyzer for Turkish, a lexicon integrated domain ontology, a document layout analyzer, an extraction ontology and a template mining module. Based on extraction ontology, conceptual sentence analysis increases portability which requires only domain concepts when compared to information extraction systems that rely on large set of linguistic patterns.

A significant portion of information that is nowa-days collected in enterprises and organizations resides in text documents, and thus is inherently unstructured. Turning it into a structured form is the aim of information extraction (IE). Depending on the approach followed, the output of an IE process can fill forms or populate relational tables, or can be presented through an ontology. This last approach is particularly interesting, since ontologies may facilitate the integration with other corporate and external data, and enable data management and governance at an abstract, conceptual level, as in Ontology-based Data Access (OBDA). To this aim, OBDA uses declarative mappings that specify the relation between the ontology and the database to be accessed. In OBDA, however, only mappings towards relational databases have been so far considered, and how to declaratively relate the ontology to unstructured sources is still unexplored. By leveraging the study on document spanners for IE, in this paper we propose a new framework that allows to map text documents to ontologies, in the spirit of the OBDA approach. We then investigate the problem of answering conjunctive queries (CQs) in our framework, and show that, if the ontology is specified in the lightweight Description Logic DL-LiteR, the problem can be solved by reformulating the user query into a new spanner. Interestingly, both the spanners used in the mapping and the one computed by the rewriting algorithm have the same expressiveness, and CQ answering in this case is polynomial in data complexity.

Many types of maritime incident databases have been established that allow people to learn from past incidents and develop corresponding mitigation measures. However, our investigation of multiple international and national databases shows that most existing databases only record basic information regarding incidents in a single table. Lots of useful information is not included in the database (i.e., limited extension of the database). Meanwhile, some basic information is recorded tautologically (i.e., data redundancy). In this paper, two widely used databases are taken as examples, the Global Integrated Shipping Information System and the Lloyd's List Intelligence, to explain these common problems of existing databases. To overcome these limitations and improve the efficiency of data maintenance, this paper develops a relational maritime safety management database. The entity-relationship model is first used to depict the inter-related semantic information surrounding maritime incidents, and a relational database model is subsequently formed. Microsoft Access is employed to implement the proposed database, and a database application is also designed to demonstrate the utility of the database. Our preliminary study shows that the proposed database is implementable and has potential usage for both industry and academic research.

The Internet presents large amount of useful information which is usually formatted for its users, which makes it hard to extract relevant data from diverse sources. Therefore, there is a significant need of robust, flexible Information Extraction (IE) systems that transform the web pages into program friendly structures such as a relational database will become essential. IE produces structured data ready for post processing. Roadrunner will be used to extract information from template web pages. In this paper, we present novel algorithm for extracting templates from a large number of web documents which are generated from heterogeneous templates. The proposed system focuses on information extraction from heterogeneous web pages. We cluster the web documents based on the common template structures so that the template for each cluster is extracted simultaneously. The resultant clusters will be given as input to the Roadrunner system.

In this paper Semantic-Enhanced Entity Relationship (SEER) and Semantic – Relational Database (SRD) diagrams are proposed. SEER and SRD are extensions of Enhanced Entity Relationship (EER) and Relational Database (RD) diagrams respectively that are based on dealing with Semantic Integrity Constraints. EER and RD do not take semantic constraints in matter. A lots of information are lost because of the shortage of documentations. The proposed SEER and SRD models embed the semantic constraints using a new notations and pseudo code to give a full diagram for a database system. The proposed models handle the documentation shortage problem.

Identifying protein complexes in protein-protein interaction (ppi) networks is often handled as a community detection problem, with algorithms generally relying exclusively on the network topology for discovering a solution. The advancement of experimental techniques on ppi has motivated the generation of many Gene Ontology (go) databases. Incorporating the functionality extracted from go with the topological properties from the underlying ppi network yield a novel approach to identify protein complexes. Additionally, most of the existing algorithms use global measures that operate on the entire network to identify communities. The result of using global metrics are large communities that are often not correlated with the functionality of the proteins. Moreover, ppi network analysis shows that most of the biological functions possibly lie between local neighbours in ppi networks, which are not identifiable with global metrics. In this paper, we propose a local community detection algorithm, (lcda-go), that uniquely exploits information of functionality from go combined with the network topology. lcda-go identifies the community of each protein based on the topological and functional knowledge acquired solely from the local neighbour proteins within the ppi network. Experimental results using the Krogan dataset demonstrate that our algorithm outperforms in most cases state-of-the-art approaches in assessment based on Precision, Sensitivity, and particularly Composite Score. We also deployed lcda, the local-topology based precursor of lcda-go, to compare with a similar state-of-the-art approach that exclusively incorporates topological information of ppi networks for community detection. In addition to the high quality of the results, one main advantage of lcda-go is its low computational time complexity.

the massive volume of data in databases, web pages, and document files usually causes information to be disorganized and unclear for the user. Therefore, information in such an environment can be classified into three forms: structured, semistructured, or unstructured. Structured information is the best form of information because it facilitates the acquisition and comprehension of knowledge. Relational Database Management System (RDBMS) has a robust structure that manages, organizes and retrieves data. There are many attempts have been made in order to deal with such data. These attempts can be categorized into three groups: within a database schema, by a developed data model within the database, or by query-based techniques in database. Nonetheless, RDBMS contain massive amount of unstructured data such as textual data.. This paper proposed Textual Virtual Schema Model (TVSM). TVSM is conducted to perform semantic textual data linking and clustering and is embedded in the relational database structure (schema). In addition, linking and converting the unstructured information to structured data. Quality improvement of textual data clusters. Achievement of high query processing efficiency in retrieving data clusters. TVSM initially developed to assist researchers, developers, and database administrators who are concerned on unstructured information management, information extraction, multi-document clustering, information retrieval, query processing efficiency, personal information management, question answering, information integration, news tracking, and news summarization