7 research outputs found
Extraction of lignin from lignocellulose at atmospheric pressure using alkylbenzenesulfonate ionic liquid
No full textRadon toxicity
No full textPrepared by DeLima Associates ... under contract no. 205-88-0636.Includes bibliographical references (p. 15-16).1992205-88-0636833
Distributed data management for large scale applications
No full textImprovements in data storage and network technologies, the emergence of new highresolution scientific instruments, the widespread use of the Internet and the World Wide Web and even globalisation have contributed to the emergence of new large scale dataintensive applications. These applications require new systems that allow users to store, share and process data across computing centres around the world. Worldwide distributed data management is particularly important when there is a lot of data, more than can fit in a single computer or even in a single data centre. Designing systems to cope with the demanding requirements of these applications is the focus of the present work.This thesis presents four contributions. First, it introduces a set of design principles that can be used to create distributed data management systems for data-intensive applications. Second, it describes an architecture and implementation that follows the proposed design principles, and which results in a scalable, fault tolerant and secure system. Third, it presents the system evaluation, which occurred under real operational conditions using close to one hundred computing sites and with more than 14 petabytes of data. Fourth, it proposes novel algorithms to model the behaviour of file transfers on a wide-area network.This work also presents a detailed description of the problem of managing distributed data, ranging from the collection of requirements to the identification of the uncertainty that underlies a large distributed environment. This includes a critique of existing work and the identification of practical limits to the development of transfer algorithms on a shared distributed environment. The motivation for this work has been the ATLAS Experiment for the Large Hadron Collider (LHC) at CERN, where the author was responsible for the development of the data management middleware
Semi-Supervised Image Classification based on a Multi-Feature Image Query Language
No full textThe area of Content-Based Image Retrieval (CBIR) deals with a wide range of research disciplines. Being closely related to text retrieval and pattern recognition, the probably most serious issue to be solved is the so-called \semantic gap". Except for very restricted use-cases, machines are not able to recognize the semantic content of digital images as well as humans.
This thesis identifies the requirements for a crucial part of CBIR user interfaces, a multimedia-enabled query language. Such a language must be able to capture the user's
intentions and translate them into a machine-understandable format. An approach to tackle this translation problem is to express high-level semantics by merging low-level image features. Two related methods are improved for either fast (retrieval) or accurate(categorization) merging.
A query language has previously been developed by the author of this thesis. It allows the formation of nested Boolean queries. Each query term may be text- or content-based and the system merges them into a single result set. The language is extensible by arbitrary new feature vector plug-ins and thus use-case independent.
This query language should be capable of mapping semantics to features by applying machine learning techniques; this capability is explored. A supervised learning algorithm based on decision trees is used to build category descriptors from a training set. Each resulting \query descriptor" is a feature-based description of a concept which is comprehensible and modifiable. These descriptors could be used as a normal query and return a result set with a high CBIR based precision/recall of the desired category. Additionally, a method for normalizing the similarity profiles of feature vectors has been
developed which is essential to perform categorization tasks.
To prove the capabilities of such queries, the outcome of a semi-supervised training session with \leave-one-object-out" cross validation is compared to a reference system. Recent work indicates that the discriminative power of the query-based descriptors is similar and is likely to be improved further by implementing more recent feature vectors
Scalable Text Mining with Sparse Generative Models
No full textThe information age has brought a deluge of data. Much of this is in text form, insurmountable in scope for humans and incomprehensible in structure for computers. Text mining is an expanding field of research that seeks to utilize the information contained in vast document collections. General data mining methods based on machine learning face challenges with the scale of text data, posing a need for scalable text mining methods.
This thesis proposes a solution to scalable text mining: generative models combined with sparse computation. A unifying formalization for generative text models is defined, bringing together research traditions that have used formally equivalent models, but ignored parallel developments. This framework allows the use of methods developed in different processing tasks such as retrieval and classification, yielding effective solutions across different text mining tasks. Sparse computation using inverted indices is proposed for inference on probabilistic models. This reduces the computational complexity of the common text mining operations according to sparsity, yielding probabilistic models with the scalability of modern search engines.
The proposed combination provides sparse generative models: a solution for text mining that is general, effective, and scalable. Extensive experimentation on text classification and ranked retrieval datasets are conducted, showing that the proposed solution matches or outperforms the leading task-specific methods in effectiveness, with a order of magnitude decrease in classification times for Wikipedia article categorization with a million classes. The developed methods were further applied in two 2014 Kaggle data mining prize competitions with over a hundred competing teams, earning first and second places
From user behaviours to collective semantics
No full textThe World Wide Web has developed into an important platform for social interactions with the rise of social networking applications of different kinds. Collaborative tagging systems, as prominent examples of these applications, allow users to share their resources and to interact with each other. By assigning tags to resources on the Web in a collaborative manner, users contribute to the emergence of complex networks now commonly known as folksonomies, in which users, documents and tags are interconnected with each other. To reveal the implicit semantics of entities involved in a folksonomy, one requires an understanding of the characteristics of the collective behaviours that create these interconnections. This thesis studies how user behaviours in collaborative tagging systems can be analysed to acquire a better understanding of the collective semantics of entities in folksonomies. We approach this problem from three different but closely related perspectives. Firstly, we study how tags are used by users and how their different intended meanings can be identified. Secondly, we develop a method for assessing the expertise of users and quality of documents in folksonomies by introducing the notion of implicit endorsement. Finally, we study the relations between documents induced from collaborative tagging and compare them with existing hyperlinks between Web documents. We show that, in each of these scenarios, it is crucial to consider the collective behaviours of the users and the social contexts in order to understand the characteristics of the entities. This project can be considered as a case study of the Social Web, the research outcomes of which can be easily generalised to many other social networking applications. It also fits into the larger framework for understanding the Web set out by the emerging interdisciplinary field of Web Science, as the work involves analyses of the interactions and behaviour of Web users in order to understand how we can improve existing systems and facilitate information sharing and retrieval on the Web
Automatisches Klassifizieren : Verfahren zur Erschliessung elektronischer Dokumente
No full textAutomatic classification of text documents refers to the computerized allocation of class numbers from existing classification schemes to natural language texts by means of suitable algorithms. Based upon a comprehensive literature review, this thesis establishes an informed and up-to-date view of the applicability of automatic classification for the subject approach to electronic documents, particularly to Web resources. Both methodological aspects and the experiences drawn from relevant projects and applications are covered. Concerning methodology, the present state-of-the-art comprises a number of statistical approaches that rely on machine learning; these methods use pre-classified example documents for establishing a model - the "classifier" - which is then used for classifying new documents. However, the four large-scale projects conducted in the 1990s by the Universities of Lund, Wolverhampton and Oldenburg, and by OCLC (Dublin, OH), still used rather simple and more traditional methodological approaches. These projects are described and analyzed in detail. As they made use of traditional library classifications their results are significant for LIS, even if no permanent quality services have resulted from these endeavours. The analysis of other relevant applications and projects reveals a number of attempts to use automatic classification for document processing in the fields of patent and media documentation. Here, semi-automatic solutions that support human classifiers are preferred, due to the yet unsatisfactory classification results obtained by fully automated systems. Other interesting implementations include Web portals, search engines and (commercial) information services, whereas only little interest has been shown in the automatic classification of books and bibliographic records. In the concluding part of the study the author discusses the most significant applications and projects, and also addresses several problems and issues in the context of automatic classification
