Semi-structured Information Sources Research Articles

Integrating Concepts and Knowledge in Large Content Networks

Large content networks like the World Wide Web contain huge amounts of information that have the potential of being integrated because their components fit within common concepts and/or are connected through hidden, implicit relationships. One attempt at such an integration is the program called the “Web of Data,” which is an evolution of the Semantic Web. It targets semi-structured information sources such as Wikipedia and turns them into fully structured ones in the form of Web-based databases like DBpedia and then integrates them with other public databases such as Geonames. On the other hand, the vast majority of the information residing on the Web is still totally unstructured, which is the starting point for our approach that aims to integrate unstructured information sources. For this purpose, we exploit techniques from Probabilistic Topic Modeling, in order to cluster Web pages into concepts (topics), which are then related through higher-level concept networks; we also make implicit semantic relationships emerge between single Web pages. The approach has been tested through a number of case studies that are here described. While the applicative focus of the research reported here is on knowledge integration on the specific and relevant case of the WWW, the wider aim is to provide a framework for integration generally applicable to all complex content networks where information propagates from multiple sources.

Building Large Collections of Chinese and English Medical Terms from Semi-Structured and Encyclopedia Websites

To build large collections of medical terms from semi-structured information sources (e.g. tables, lists, etc.) and encyclopedia sites on the web. The terms are classified into the three semantic categories, Medical Problems, Medications, and Medical Tests, which were used in i2b2 challenge tasks. We developed two systems, one for Chinese and another for English terms. The two systems share the same methodology and use the same software with minimum language dependent parts. We produced large collections of terms by exploiting billions of semi-structured information sources and encyclopedia sites on the Web. The standard performance metric of recall (R) is extended to three different types of Recall to take the surface variability of terms into consideration. They are Surface Recall (), Object Recall (), and Surface Head recall (). We use two test sets for Chinese. For English, we use a collection of terms in the 2010 i2b2 text. Two collections of terms, one for English and the other for Chinese, have been created. The terms in these collections are classified as either of Medical Problems, Medications, or Medical Tests in the i2b2 challenge tasks. The English collection contains 49,249 (Problems), 89,591 (Medications) and 25,107 (Tests) terms, while the Chinese one contains 66,780 (Problems), 101,025 (Medications), and 15,032 (Tests) terms. The proposed method of constructing a large collection of medical terms is both efficient and effective, and, most of all, independent of language. The collections will be made publicly available.

A Framework for Extracting, Classifying, Analyzing, and Presenting Information from Semi-Structured Web Data Sources

Extracting information from the web data sources becomes very important because the massive and increasing amount of diverse semi-structured information sources in the Internet that are available to users, and the variety of web pages making the process of information extraction from web a challenging problem. This paper proposes a framework for extracting, classifying, analyzing, and presenting semi-structured web data sources. The framework is able to extract relevant information from different web data sources, and classify the extracted information based on the standard classification scheme of Nokia products, which has been chosen as the case study.

A Technique for Extracting Sub-source Similarities from Information Sources Having Different Formats

In this paper we propose a semi-automatic technique for deriving the similarity degree between two portions of heterogeneous information sources (hereafter, sub-sources). The proposed technique consists in two phases: the first one selects the most promising pairs of sub-sources, whereas the second one computes the similarity degree relative to each promising pair. We show that the detection of sub-source similarities is a special case (and a very interesting one, for semi-structured information sources) of the more general problem of Scheme Match. In addition, we present a real example case to clarify the proposed technique, a set of experiments we have conducted to verify the quality of its results, a discussion about its computational complexity and its classification in the context of related literature. Finally, we discuss some possible applications which can benefit by derived similarities.

LESSONS LEARNED FROM APPLYING AI TO THE WEB

Ontobroker applies Artificial Intelligence techniques to improve access to heterogeneous, distributed and semi-structured information sources as they are presented in the World Wide Web or organization-wide intranets. It relies on the use of ontologies to annotate web pages, formulate queries and derive answers. In this paper we will briefly sketch Ontobroker. Then we will discuss its main shortcomings, i.e. we will share the lessons we learned from our exercise. We will also show how On2broker overcomes these limitations. Most important is the separation of the query and inference engines and the integration of new web standards like XML and RDF.

Managing historical semistructured data

Semistructured data may be irregular and incomplete and does not necessarily conform to a fixed schema. As with structured data, it is often desirable to maintain a history of changes to data, and to query over both the data and the changes. Representing and querying changes in semistructured data is more difficult than in structured data due to the irregularity and lack of schema. We present a model for representing changes in semistructured data and a language for querying over these changes. An important feature of our approach is that we represent and query changes directly as annotations on the affected data, instead of indirectly as the difference between database states. We describe the implementation of our model and query language. We present extensions that permit convenient snapshot-based access in our change-based data model. We also describe our design and implementation of a wind query subscription service that permits users to subscribe to changes in semistructured information sources. © 1999 John Wiley & Sons, Inc.