Pattern-based approaches for Information Extraction (IE) typically apply a pattern learner to a set of domain-specific training documents to generate extraction patterns for the IE system. This restricts the coverage of the system primarily to the expressions and language constructs that appear within the limited training data. Our research looks to the vast quantities of readily available text...

We present a first version of a filter dictionary for use in a computer-telephony text-to-speech synthesis system. The aim of the filter dictionary was to provide a lexicon that was compact, fast and had broader coverage than the standard dictionary used to create it. Correct phonemic transcriptions and lexical stress assignment were both required for a transcription to be deemed accurate. The ...

The problem of extractive text summarization for a collection of documents is defined as the problem of selecting a small subset of sentences so that the contents and meaning of the original document set are preserved in the best possible way. In this paper we describe the linear programming-based global optimization model to rank and extract the most relevant sentences to a summary. We introdu...

For automatic speech recognition, the construction of an adequate language model may be difficult when only a limited amount of training text is available. Previous work has shown that in the case of small training sets statistical language models may outperform grammars on out-of-coverage utterances, while showing comparable performance on incoverage input. In this paper, we compare the perfor...

We present a new method for translating unrestricted natural language texts into predicate logic formulas. This relies on the semantic evaluation procedure of Scope Control Theory (SCT), a variant of Dynamic Semantic formalisms. The key benefit is that parsed syntactic structures are shown to form sufficient input for semantic evaluation, eliminating the need to build distinct semantic expressi...

This paper details the steps involved in scaling-up a lexicalised finite-state morphology transducer for use on unrestricted text. Our starting point was a base-line inflectional morphology engine [1], with 81% token coverage measured against a 15 million word corpus of Irish texts [2]. Manually scaling the FST lexicon component of a morphology transducer is time-consuming, expensive and rarely...

We investigate the problem of automatically labelling faces of characters in TV or movie material with their names, using only weak supervision from automaticallyaligned subtitle and script text. Our previous work (Everingham et al. [8]) demonstrated promising results on the task, but the coverage of the method (proportion of video labelled) and generalization was limited by a restriction to fr...

For automatic speech recognition, the construction of an adequate language model may be difficult when only a limited amount of training text is available. Previous work has shown that in the case of small training sets statistical language models may outperform grammars on out-of-coverage utterances, while showing comparable performance on incoverage input. In this paper, we compare the perfor...

The purpose of implementation-based testing is to gain a measure of confidence in the correctness of the software by providing adequate coverage of the code. One unit of testing in object-oriented software is a class. However, classes use other classes and if class interactions form a cycle of dependencies then, to test a client class that uses an untested supplier class, stubs must be construc...