Extending WHIRL with background knowledge for improved text classification

Intelligent use of the many diverse forms of data available on the Internet requires new tools for managing and manipulating heterogeneous forms of information. This paper uses WHIRL, an extension of relational databases that can manipulate textual data using statistical similarity measures developed by the information retrieval community. We show that although WHIRL is designed for more general similarity-based reasoning tasks, it is competitive with mature systems designed explicitly for inductive classification. In particular, WHIRL is well suited for combining different sources of knowledge in the classification process. We show on a diverse set of tasks that the use of appropriate sets of unlabeled background knowledge often decreases error rates, particularly if the number of examples or the size of the strings in the training set is small. This is especially useful when labeling text is a labor-intensive job and when there is a large amount of information available about a particular problem on the World Wide Web.

Result merging methods in distributed information retrieval with overlapping databases

In distributed information retrieval systems, document overlaps occur frequently among different component databases. This paper presents an experimental investigation and evaluation of a group of result merging methods including the shadow document method and the multi-evidence method in the environment of overlapping databases. We assume, with the exception of resultant document lists (either with rankings or scores), no extra information about retrieval servers and text databases is available, which is the usual case for many applications on the Internet and the Web. The experimental results show that the shadow document method and the multi-evidence method are the two best methods when overlap is high, while Round-robin is the best for low overlap. The experiments also show that [0,1] linear normalization is a better option than linear regression normalization for result merging in a heterogeneous environment.

“This is the Whiteman’s Law”: Aboriginal resistance,bureaucratic change and the Census of Canada, 1830–2006

Census information of some form has been collected in Canada since the 1611 census of New France. Aboriginal people, identified or not, have been included in these enumerations. The collection of this information has had a profound impact on Aboriginal people and has been an element that has shaped their relationship with the dominant society. In response, Canadian Aboriginal people have often resisted and refused to co-operate with census takers and their masters. This article is an examination of this phenomenon focused on the censuses conducted in the post-Confederation period to the present. A census is made to collect information on populations and individuals that can then be used to configure and shape social and political relations between those being enumerated and the creators of the census. However, the human objects of the census are not just passive integers and they have resisted its creation in a number of ways, including being “missing” when the census is taken, refusing to answer the questions posed by enumerators or even driving them off Aboriginal territory. A census identifies elements of the social order and attempts to set them in their “proper” place and those who do not wish to be part of that order may refuse to take part. Archivists and historians must understand that the knowledge gained in a census is bound with the conditions of own creation. This has been noted by contemporary Aboriginal researchers who often state that the archival record of their people often distorts history and reflects the ideas and superficial observations of their Euro-Canadian creators. Changes to the Census of Canada since 1981, have increased the participation rate and therefore changed the nature of the record.

A world-class archival achievement: the People’s Republic of China archivists’ success in opening the Ming-Qing central-government archives, 1949–1998

This article describes the first half century of the Communist government’s supervision and management of the central-government archives of the last two dynasties. Immediately with the Communist ascent to power in 1949, the new government took great interest in assembling and protecting the country’s archival documents, readying the Ming-Qing archives for access to scholars, and preparing for publication of selected materials. By the 1980s Beijing’s Number One Historical Archives, in charge of the largest holding of Ming-Qing documents, had become the first Chinese authority to complete a full sorting and preliminary catalogues for such a collection. Moreover, to facilitate searches, an attempt has recently begun to create a subject-heading system for these and other holdings in the country. In the first half century’s final decades, foreign researchers were admitted for the first time and tours and international exchanges began to take place.

Hyperactive Grey Objects

Previous papers on grey literature by the authors have described (1) the need for formal metadata to allow machine understanding and therefore scalable operations; (2) the enhancement of repositories of grey (and other) e-publications by linking with CRIS (Current Research Information Systems); (3) the use of the research process to collect metadata incrementally reducing the threshold barrier for end-users and improving quality in an ambient GRIDs environment. This paper takes the development one step further and proposes “intelligent” grey objects. The hypothesis is in 2 parts: (1) that the use of passive catalogs of metadata does not scale (a) in a highly distributed environment with millions of nodes and (b) with vastly increased volumes of R&D output grey publications with associated metadata; (2) that a new paradigm is required that (a) integrates grey with white literature and other R&D outputs such as software, data, products and patents (b) in a self-managing, self-optimizing way and that this paradigm manages automatically curation, provenance digital rights, trust, security and privacy. Concerning (1) existing repositories provide catalogs; harvesting takes increasing time ensuring non-currency. The end-user expends much manual effort/intelligence to utilize the results. The elapsed time of (1) the network (2) the centralized (or centrally controlled distributed) catalog server searches (3) end-user intervention becomes unacceptable. Concerning (2) there is no paradigm currently known to the authors that satisfies the requirement. Our proposal is outlined below. Hyperactive combines both hyperlinking and active properties of a (grey) object. Hyperlinking implies multimedia components linked to form the object and also external links to other resources. The term active implies that objects do not lie passively in a repository to be retrieved by end-users. They “get a life” and the object moves through the network knowing where it is going. A hyperactive grey object is wrapped by its (incrementally recorded) formal metadata and an associated (software) agent. It moves through process steps such as initial concept, authoring, reviewing and depositing in a repository. The workflow is based on the rules and information in the corporate data repository with which the agent interacts. Once the object is deposited, the agent associated with it actively pushes the object to the end-users (or systems) whose metadata indicate interest or an obligation in a workflowed process. The agents check the object and user (or system) metadata for rights, privacy, security parameters, and for any charges and assure compatibility. Alternatively the object can be found passively by end-user or system agents. The object can also associate itself with other objects forming relationships utilising metadata or content. Declared relationships include references and citations; workflowed relationships include versions and also links to corporate information and research datasets and software; inferenced relationships are discovered relationships such as between documents by different authors developed from an earlier idea of a third author. Components of this paradigm have been implemented to some extent. The challenge is implementing—respecting part two of the hypothesis—the integration architecture. This surely is harnessing the power of grey.

Growth in biomedical publications and scientific institutions in the Emirates (1998–2004): an Arabian renaissance?

OBJECTIVES: The United Arab Emirates is witnessing a rapid development in population and infrastructure, leading to an expansion in the educational field and the health care system, including universities, hospitals and research institutes. One potential outcome of this development is an increase in biomedical research publications. This was examined in this study. METHOD: Searches were made of three databases; PubMed, Current Contents and embase. All records from these databases related to the Emirates were downloaded and cross-referenced. Duplications were removed as well as erroneous records. A total of 1369 publications during 1998-2004 were obtained. RESULTS: Analysis of these records revealed that the Faculty of Medicine and Health Sciences, located at the federal University, was responsible for 48% of all biomedical publications. All hospitals together accounted for 24%, while other universities and other research institutes accounted for 13 and 14% respectively. Further analysis revealed that there had been no further growth in the number of publications since 1998. This is in contrast to the increasing number of institutions that have been set up in the 1998-2004 period; from 29 to 112. CONCLUSION: The Emirates are witnessing a rapid development in universities, colleges, hospitals and research centres. The ratio of publications, as yet, has not kept up with this development. The enormous growth in institutes, however, may, in time, herald an Arabian renaissance in scientific activity and related publications.     

The long-term preservation of identifiable personal data: a comparative archival perspective on privacy regulatory models in the European Union,Australia, Canada and the United States

This article analyses the extent to which archival exemptions for historical, scientific and statistical research in privacy legislation support preservation in selected European Union countries, and comparable aspects of Australian, American and Canadian law within a legal, ethical and digital archival perspective. The authors recommend that the further processing of personal data under data protection law be given a wider scope of interpretation for archival preservation purposes in both the public and private sector, coupled with the use of researcher and archival codes in relation to access to personal data. They also recommend early appraisal and integration of privacy with freedom of information and archival regimes.

Beyond chip monks and paper tigers: towards a new culture of archival format specialists

The emergence of the new format of electronic/digital records provides the opportunity for archivists to reconsider the presumed format-neutrality of professional practice. As research in electronic records has served to re-emphasise, without an understanding of the needs and forms of material, then the work of archivists can have a profound impact on the evidential value and long-term research potential of the material. This paper attempts to broaden the debate about the requirements of all archival formats, and to build a new regime of 21st-century format specialists.