We propose a “smart” language learning system for students to acquire domain-specific vocabulary while taking an online course. F-Lingo, a browser plugin, works on top of the FutureLearn MOOC platform to provide learners with opportunities to study the words, phrases, and concepts that are important to the course topic. F-Lingo comprises three components. The Material Gathering component crawls the web pages of the MOOC course the student has chosen, collecting the entire textual content (with some exceptions). The Vocabulary Extraction component identifies domain-specific words, phrases, and concepts, and hyperlinks in the MOOC page to draw the student’s attention to them. Clicking a link displays a dialog window in which lexico-grammatical features, and definitions, of the extracted items can be studied, including illustrations in example sentences retrieved from external resources such as Wikipedia and FLAX. The Progress Tracking component records the clicks that students make on hyperlinks and the time spent in the dialog windows. This allows us to build the student’s vocabulary learning profile under the assumption that the more time the student pays attention to an item, the more worthy the item to be included in a follow-up language activity. These statistical data provide evidence and reasoning in our current and ongoing work on automatically generating personalized language activities and vocabulary tests at the end of the MOOC course. F-Lingo has been made available in three Data Mining courses on the FutureLearn MOOC platform and has been used by 109 learners. This research is ongoing. Future work focuses on automatically generating personalized vocabulary tests and activities based on the student’s click statistics.

The Greenstone Digital Library Software has helped spread the practical impact of digital library technology around the globe. As Greenstone enters its third decade, this article takes a renewed look at its development, the challenges that have been faced, and the lessons that have been learned in deploying a comprehensive open-source system for the construction of digital libraries internationally. In particular, we outline architectural changes in the software that have occurred over time, and highlight how the user interfaces have evolved to provide a more immersive, interactive user experience, made possible through advancements in the underlying web technologies.

Learning collocations is one of the most challenging aspects of language learning as there are literally hundreds of thousands of possibilities for combining words. Corpus consultation with concordancers has been recognized in the literature as an established way for language learners to study and explore collocations at their own pace and in their own time although not without technological and sometimes cost barriers. This paper describes the development and evaluation of a learner-friendly collocation consultation system called FlaxLC in a design departure away from the traditional concordancer interface. Two evaluation studies were conducted to assess the learner-friendliness of the system: a face-to-face user study to find out how international students in a New Zealand university used the system to collect collocations of their own interest and a user query analysis-based on an observable artefact of how online learners actually used the system over the course of one year-to examine how the system is used in real life to search and retrieve collocations.

F-Lingo is a chrome extension that works on top of the FutureLearn MOOC platform to support content-based language learning of domain-specific terminology for professional and academic purposes.

This chapter describes the automated FLAX language system (flax.nzdl.org) that extracts salient linguistic features from academic text and presents them in an interface designed for L2 students who are learning academic writing. Typical lexico-grammatical features of any word or phrase, collocations, and lexical bundles are automatically identified and extracted in a corpus; learners can explore them by searching and browsing, and inspect them along with contextual information. This chapter uses a single running example, the PhD abstracts corpus of 9.8 million words derived from the open access Electronic Theses Online Service (EThOS) at the British Library, but the approach is fully automated and can be applied to any collection of English writing. Implications for reusing open access publications for non-commercial educational and research purposes are presented for discussion. Design considerations for developing teaching and learning applications that focus on the rhetorical and lexico-grammatical patterns found in the abstract genre are also discussed.

This article reports on a language learning system and a program designed to help students with academic vocabulary in the New Zealand university computer science department. The system is a learner-friendly corpus-based tool that allows students to look up lexico-grammatical patterns of a given word in academic writing. The program, based on a data-driven learning approach, comprises tutorials, workshops, and follow-up exercises that help students learn useful formulaic patterns of academic words that are typical in computer science. The authors' results capture students' awareness of language patterns in academic text and their growing confidence in using academic words with the assistance of the tool. Not surprisingly, interpreting and transferring the corpus data into students' own writing requires training and practice. The effectiveness and limitations of the resources and tools used in this learning program are examined, and suggestions are made for further improvement and future research.

IEEE, Computer, Jun. 1984, pp. 8-19, Terry A. Welch, Sperry Research Center, "A Technique for High-Per formance Data Compression.” IEEE, Transactions On Information Theory, vol. II 23, No. 3, May 1977, pp. 337-343 J. Ziv, A. Lempel, "A Universal Algorithm for Sequential Data Compres sion.' IEEE, Transactions On Information Theory, vol. II 24, No. 5, Sep. 1978, pp. 530-536, J. Ziv, A. Lempel, "Compression of Individual Sequences via Variable Rate Coding.” IEEE Transactions on Information Theory vol. II, 21,

There are many transformations that can make real-world datasets more amenable to the learning algorithms discussed in the rest of the book. We first consider methods for attribute selection, which remove attributes that are not useful for the task at hand. Then we look at discretization methods: algorithms for turning numeric attributes into discrete ones. Next we discuss several techniques for projecting data into a space that is more suitable for learning: well-known methods for dimensionality reduction, including unsupervised approaches such as principal component analysis, independent component analysis, and random projections, as well as supervised approaches such as partial least squares regression and linear discriminant analysis. We consider how to turn textual data into numeric attribute vectors so that standard learning techniques can be applied, and present simple methods for approaching time series data. The last four sections deal with data sampling, data cleansing, generic approaches for multiclass classification, and calibration of class probabilities, respectively. Sampling is nontrivial when the data arrives as a stream, and we discuss the "reservoir" method for taking an unbiased sample in this case. Data cleansing can be performed by iteratively applying standard supervised learning algorithms to remove outliers, but there are also dedicated techniques for anomaly detection and so-called "one-class learning" that are applicable. For dealing with multiclass classification problems, we consider several ways of decomposing them into a set of two-class problems, e.g., by applying error-correcting output codes. Finally, we describe how to calibrate class probability estimates to improve their accuracy.

This chapter explains practical decision tree and rule learning methods, and also considers more advanced approaches for generating association rules. The basic algorithms for learning classification trees and rules presented in Chapter 4, Algorithms: the basic methods, are extended to make them applicable to real-world problems that contain numeric attributes, noise, and missing values. We discuss the seminal C4.5 algorithm for decision tree learning, consider an alternative pruning method implemented in the CART tree learning algorithm, and discuss the incremental reduced-error pruning method for growing and pruning classification rules, leading up to the RIPPER and PART algorithms for rule induction. We also briefly consider rule sets with exceptions. The last section of this chapter switches to unsupervised learning of rule sets by investigating how a special-purpose data structure can be constructed to accelerate the process of finding association rules. More specifically, we consider frequent-pattern trees and how they can be used to efficiently search for frequent item sets.

We begin by revisiting the basic instance-based learning method of nearest-neighbor classification and considering how it can be made more robust and storage efficient by generalizing both exemplars and distance functions. We then discuss two well-known approaches for generalizing linear models that go beyond modeling linear relationships between the inputs and the outputs. The first is based on the so-called kernel trick, which implicitly creates a high-dimensional feature space and models linear relationships in this extended space. We discuss support vector machines for classification and regression, kernel ridge regression, and kernel perceptrons. The second approach is based on applying simple linear models in a network structure that includes nonlinear transformations. This yields neural networks, and we discuss the classical multilayer perceptron. The final part of the chapter discusses an alternative method for tackling learning problems with complex relationships: building linear models that are local in the sense that they only apply to a small part of the input space. We consider model trees, which are decision trees with linear regression models at the leaf nodes, and locally weighted linear regression, which combines instance-based learning and linear regression.