13 research outputs found
Health Hazard Evaluation Report: HETA-84-029-1427: Jewish Family and Childrens Agency; Ardmore, Pennsylvania
No full textEnvironmental and breathing zone samples were analyzed for asbestos (1332214) at the Jewish Family and Children's Agency Office (SIC- 8999), Ardmore, Pennsylvania in March, 1984. The evaluation was requested by the district director of the agency because of concern about possible exposure to airborne asbestos from the ceiling in the reception area. The ceiling, previously encapsulated, had recently cracked due to water leakage. The asbestos concentrations were less than the analytical detection limit of 0.01 fibers per cubic centimeter. Samples taken from the ceiling and inside a duct contained approximately 1 to 2 and 5 percent chrysotile (12001295) asbestos, respectively. The author concludes that the employees at the Agency were not exposed to asbestos at the time of the survey. He recommends that the duct, not presently encapsulated, be treated so any loose fibers will not enter the reception room and periodic inspection of the encapsulation should be instituted
Human evaluation of Kea, an automatic keyphrasing system.
No full textThis paper describes an evaluation of the Kea automatic keyphrase extraction algorithm. Tools that automatically identify keyphrases are desirable because document keyphrases have numerous applications in digital library systems, but are costly and time consuming to manually assign. Keyphrase extraction algorithms are usually evaluated by comparison to author-specified keywords, but this methodology has several well-known shortcomings. The results presented in this paper are based on subjective evaluations of the quality and appropriateness of keyphrases by human assessors, and make a number of contributions. First, they validate previous evaluations of Kea that rely on author keywords. Second, they show Kea's performance is comparable to that of similar systems that have been evaluated by human assessors. Finally, they justify the use of author keyphrases as a performance metric by showing that authors generally choose good keywords
Mining housekeeping genes with a Naive Bayes classifier
No full textThe first author was supported by the Student Awards Agency for Scotland. The second author is supported by BBSRC grant BBS RC BB/D006473/1, and under the Advanced Knowledge Technologies (AKT) Interdisciplinary Research Collaboration (IRC), which is sponsored by the UK Engineering and Physical Sciences Research Council under grant number GR/N15764/01.Background: Traditionally, housekeeping and tissue specific genes have been classified using direct assay of mRNA presence across different tissues, but these experiments are costly and the results not easy to compare and reproduce. Results: In this work, a Naive Bayes classifier based only on physical and functional characteristics of genes already available in databases, like exon length and measures of chromatin compactness, has achieved a 97% success rate in classification of human housekeeping genes ( 93% for mouse and 90% for fruit fly). Conclusion: The newly obtained lists of housekeeping and tissue specific genes adhere to the expected functions and tissue expression patterns for the two classes. Overall, the classifier shows promise, and in the future additional attributes might be included to improve its discriminating power.Peer reviewe
A user evaluation of hierarchical phrase browsing
No full textPhrase browsing interfaces based on hierarchies of phrases extracted automatically from document collections offer a useful compromise between automatic full-text searching and manually-created subject indexes. The literature contains descriptions of such systems that many find compelling and persuasive. However, evaluation studies have either been anecdotal, or focused on objective measures of the quality of automatically-extracted index terms, or restricted to questions of computational efficiency and feasibility. This paper reports on an empirical, controlled user study that compares hierarchical phrase browsing with full-text searching over a range of information seeking tasks. Users found the results located via phrase browsing to be relevant and useful but preferred keyword searching for certain types of queries. Users experiences were marred by interface details, including inconsistencies between the phrase browser and the surrounding digital library interface
Large-Scale Assessment of Mobile Notifications
No full textNotifications are a core feature of mobile phones. They in-form users about a variety of events. Users may take imme-diate action or ignore them depending on the importance of a notification as well as their current context. The nature of notifications is manifold, applications use them both sparsely and frequently. In this paper we present the first large-scale analysis of mobile notifications with a focus on users ’ subjec-tive perceptions. We derive a holistic picture of notifications on mobile phones by collecting close to 200 million notifica-tions from more than 40,000 users. Using a data-driven ap-proach, we break down what users like and dislike about noti-fications. Our results reveal differences in importance of no-tifications and how users value notifications from messaging apps as well as notifications that include information about people and events. Based on these results we derive a number of findings about the nature of notifications and guidelines to effectively use them. Author Keywords notification; mobile phone; large-scale; in the wild; apps
The knowledge standard for ISP copyright and trademark secondary liability: A comparative study on the analysis of US and EU laws
No full textThis thesis was submitted for the degree of Doctor of Philosophy and awarded by Brunel UniversityHolders of rights sue ISPs for copyright and trademark infringement: specifically, for contributory liability through the ISP’s knowledge of user infringement. Knowledge about user infringement has been prevalently recognised as a crucial element of ISPs’ secondary liability, but the approaches concerning the knowledge standard are different in US copyright case law (traditional tort), the US Digital Millennium Copyright Act, the US Lanham Act, US trademark case law, and the EU Electronic Commerce Directive. Their differences have posed questions on the efficacy of the current knowledge standards and case law interpretations to omit legal ambiguities and offer appropriate guidance for tackling issues. This research presents that the US knowledge standards and the ECD knowledge standard apply broad knowledge standards to evaluate ISPs’ knowledge but they differ in terms of their elements and conditions for permitting ISPs and copyright holders to co-exist and combat copyright infringement. US copyright case law, the InWood knowledge standard, and the EU knowledge standard are deficient in terms of offering a suitable notice and take-down regime to reduce the duties of ISPs and to tackle the high risk of an ISP being held liable without knowledge. This is in contrast to the DMCA, which is free from such legal concerns because of its specified notice and take-down regime. Consequently, to fulfil the aims of this research, the following recommendations are made: the US copyright knowledge standard should preserve the broad knowledge standard of the DMCA, subject to implementing a compulsory notice and take-down regime, establishing a special body regarding the notification in section 512, and designing technical criteria for the ‘red flag’ test. In addition, it is recommended that the Lanham Act codify the InWood knowledge standard and the DMCA’s notice and take-down procedures. Besides, it is recommended that the ECD establish a notice and take-down regime similar to that applied by the US DMCA (subject to the above amendments)
High performance latent dirichlet allocation for text mining
No full textThis thesis was submitted for the degree of Doctor of Philosophy and awarded by Brunel University.Latent Dirichlet Allocation (LDA), a total probability generative model, is a three-tier Bayesian model. LDA computes the latent topic structure of the data and obtains the significant information of documents. However, traditional LDA has several limitations in practical applications. LDA cannot be directly used in classification because it is a non-supervised learning model. It needs to be embedded into appropriate classification algorithms. LDA is a generative model as it normally generates the latent topics in the categories where the target documents do not belong to, producing the deviation in computation and reducing the classification accuracy. The number of topics in LDA influences the learning process of model parameters greatly. Noise samples in the training data also affect the final text classification result. And, the quality of LDA based classifiers depends on the quality of the training samples to a great extent. Although parallel LDA algorithms are proposed to deal with huge amounts of data, balancing computing loads in a computer cluster poses another challenge. This thesis presents a text classification method which combines the LDA model and Support Vector Machine (SVM) classification algorithm for an improved accuracy in classification when reducing the dimension of datasets. Based on Density-Based Spatial Clustering of Applications with Noise (DBSCAN), the algorithm automatically optimizes the number of topics to be selected which reduces the number of iterations in computation. Furthermore, this thesis presents a noise data reduction scheme to process noise data. When the noise ratio is large in the training data set, the noise reduction scheme can always produce a high level of accuracy in classification. Finally, the thesis parallelizes LDA using the MapReduce model which is the de facto computing standard in supporting data intensive applications. A genetic algorithm based load balancing algorithm is designed to balance the workloads among computers in a heterogeneous MapReduce cluster where the computers have a variety of computing resources in terms of CPU speed, memory space and hard disk space
A modular architecture for systematic text categorisation
No full textThis work examines and attempts to overcome issues caused by the lack of formal standardisation when defining text categorisation techniques and detailing how they might be appropriately integrated with each other. Despite text categorisation’s long history the concept of automation is relatively new, coinciding with the evolution of computing technology and subsequent increase in quantity and availability of electronic textual data. Nevertheless insufficient descriptions of the diverse algorithms discovered have lead to an acknowledged ambiguity when trying to accurately replicate methods, which has made reliable comparative evaluations impossible.
Existing interpretations of general data mining and text categorisation methodologies are analysed in the first half of the thesis and common elements are extracted to create a distinct set of significant stages. Their possible interactions are logically determined and a unique universal architecture is generated that encapsulates all complexities and highlights the critical components. A variety of text related algorithms are also comprehensively surveyed and grouped according to which stage they belong in order to demonstrate how they can be mapped.
The second part reviews several open-source data mining applications, placing an emphasis on their ability to handle the proposed architecture, potential for expansion and text processing capabilities. Finding these inflexible and too elaborate to be readily adapted, designs for a novel framework are introduced that focus on rapid prototyping through lightweight customisations and reusable atomic components.
Being a consequence of inadequacies with existing options, a rudimentary implementation is realised along with a selection of text categorisation modules. Finally a series of experiments are conducted that validate the feasibility of the outlined methodology and importance of its composition, whilst also establishing the practicality of the framework for research purposes. The simplicity of experiments and results gathered clearly indicate the potential benefits that can be gained when a formalised approach is utilised
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
