Pyramid annotation makes it possible to evaluate quantitatively and qualitatively the content of machine-generated (or human) summaries. Evaluation methods must prove themselves against the same measuring stick – evaluation – as other research methods. First, a formal assessment of pyramid data from the 2003 Document Understanding Conference (DUC) is presented; this addresses whether the form of annotation is reliable and whether score results are consistent across annotators. A combination of interannotator reliability measures of the two manual annotation phases (pyramid creation and annotation of system peer summaries against pyramid models), and significance tests of the similarity of system scores from distinct annotations, produces highly reliable results. The most rigorous test consists of a comparison of peer system rankings produced from two independent sets of pyramid and peer annotations, which produce essentially the same rankings. Three years of DUC data (2003, 2005, 2006) are used to assess the reliability of the method across distinct evaluation settings: distinct systems, document sets, summary lengths, and numbers of model summaries. This functional assessment addresses the method's ability to discriminate systems across years. Results indicate that the statistical power of the method is more than sufficient to identify statistically significant differences among systems, and that the statistical power varies little across the 3 years.

The distributional hypothesis states that words with similar distributional properties have similar semantic properties (Harris 1968). This perspective on word semantics, was early discussed in linguistics (Firth 1957; Harris 1968), and then successfully applied to Information Retrieval (Salton, Wong and Yang 1975). In Information Retrieval, distributional notions (e.g. document frequency and word co-occurrence counts) have proved a key factor of success, as opposed to early logic-based approaches to relevance modeling (van Rijsbergen 1986; Chiaramella and Chevallet 1992; van Rijsbergen and Lalmas 1996).

This paper focuses on an important step in the creation of a system of meaning representation and the development of semantically annotated parallel corpora, for use in applications such as machine translation, question answering, text summarization, and information retrieval. The work described below constitutes the first effort of any kind to annotate multiple translations of foreign-language texts with interlingual content. Three levels of representation are introduced: deep syntactic dependencies (IL0), intermediate semantic representations (IL1), and a normalized representation that unifies conversives, nonliteral language, and paraphrase (IL2). The resulting annotated, multilingually induced, parallel corpora will be useful as an empirical basis for a wide range of research, including the development and evaluation of interlingual NLP systems and paraphrase-extraction systems as well as a host of other research and development efforts in theoretical and applied linguistics, foreign language pedagogy, translation studies, and other related disciplines.

Natural Language Engineering (NLE) enters the second decade of the twenty-first century having established itself as a leading forum for high-quality articles covering all aspects of applied Natural Language Processing research, including, but not limited to, the engineering of natural language methods and applications. It continues to promote first class original research and bridge the gap between traditional computational linguistics research and the implementation of practical applications with potential real-world use. The journal has responded in several ways to the ongoing interest in and growth of research in this area. In 2007 NLE increased its number of pages per issue, thus enabling the publication of more articles. As of January 2010, new publication types are also promoted. In addition to welcoming articles which report on original, unpublished research, the journal now invites surveys presenting the state of the art in important areas of Natural Language Engineering and Natural Language Processing (such as tasks, tools, resources or applications) as well as squibs discussing specific problems. Book reviews and reports on industrial applications will continue to have a prominent place in the Journal. Conference reports, comparative discussions of Natural Language Engineering products and policy-orientated papers examining, for example, funding programmes or market opportunities, are welcome too. Special issues will remain an important feature of the Journal. We envisage one special issue per year, on average. Special issues are selected on a competitive basis after regular calls for proposals.

We show how a quantitative context may be established for what is essentially qualitative in nature by topologically embedding a lexicon (here, WordNet) in a complete metric space. This novel transformation establishes a natural connection between the order relation in the lexicon (e.g., hyponymy) and the notion of distance in the metric space, giving rise to effective word-level and document-level lexical semantic distance measures. We provide a formal account of the topological transformation and demonstrate the value of our metrics on several experiments involving information retrieval and document clustering tasks.

We investigate the use of different machine learning methods in combination with feature selection techniques to explore human multimodal dialogue strategies and the use of those strategies for automated dialogue systems. We learn policies from data collected in a Wizard-of-Oz study where different human ‘wizards’ decide whether to ask a clarification request in a multimodal manner or else to use speech alone. We first describe the data collection, the coding scheme and annotated corpus, and the validation of the multimodal annotations. We then show that there is a uniform multimodal dialogue strategy across wizards, which is based on multiple features in the dialogue context. These are generic features, available at runtime, which can be implemented in dialogue systems. Our prediction models (for human wizard behaviour) achieve a weighted f-score of 88.6 per cent (which is a 25.6 per cent improvement over the majority baseline). We interpret and discuss the learned strategy. We conclude that human wizard behaviour is not optimal for automatic dialogue systems, and argue for the use of automatic optimization methods, such as Reinforcement Learning. Throughout the investigation we also discuss the issues arising from using small initial Wizard-of-Oz data sets, and we show that feature engineering is an essential step when learning dialogue strategies from such limited data.

Distributional word similarity is most commonly perceived as a symmetric relation. Yet, directional relations are abundant in lexical semantics and in many Natural Language Processing (NLP) settings that require lexical inference, making symmetric similarity measures less suitable for their identification. This paper investigates the nature of directional (asymmetric) similarity measures that aim to quantify distributional feature inclusion. We identify desired properties of such measures for lexical inference, specify a particular measure based on Average Precision that addresses these properties, and demonstrate the empirical benefit of directional measures for two different NLP datasets.

Electronic written texts used in computer-mediated interactions (emails, blogs, chats, and the like) contain significant deviations from the norm of the language. This paper presents the detail of a system aiming at normalizing the orthography of French SMS messages: after discussing the linguistic peculiarities of these messages and possible approaches to their automatic normalization, we present, compare, and evaluate various instanciations of a normalization device based on weighted finite-state transducers. These experiments show that using an intermediate phonemic representation and training, our system outperforms an alternative normalization system based on phrase-based statistical machine translation techniques.

In this article, we present a comprehensive study aimed at computing semantic relatedness of word pairs. We analyze the performance of a large number of semantic relatedness measures proposed in the literature with respect to different experimental conditions, such as (i) the datasets employed, (ii) the language (English or German), (iii) the underlying knowledge source, and (iv) the evaluation task (computing scores of semantic relatedness, ranking word pairs, solving word choice problems). To our knowledge, this study is the first to systematically analyze semantic relatedness on a large number of datasets with different properties, while emphasizing the role of the knowledge source compiled either by the ‘wisdom of linguists’ (i.e., classical wordnets) or by the ‘wisdom of crowds’ (i.e., collaboratively constructed knowledge sources like Wikipedia).

The article discusses benefits and drawbacks of different approaches to evaluating semantic relatedness. We show that results should be interpreted carefully to evaluate particular aspects of semantic relatedness. For the first time, we employ a vector based measure of semantic relatedness, relying on a concept space built from documents, to the first paragraph of Wikipedia articles, to English WordNet glosses, and to GermaNet based pseudo glosses. Contrary to previous research (Strube and Ponzetto 2006; Gabrilovich and Markovitch 2007; Zesch et al. 2007), we find that ‘wisdom of crowds’ based resources are not superior to ‘wisdom of linguists’ based resources. We also find that using the first paragraph of a Wikipedia article as opposed to the whole article leads to better precision, but decreases recall. Finally, we present two systems that were developed to aid the experiments presented herein and are freely available1 for research purposes: (i) DEXTRACT, a software to semi-automatically construct corpus-driven semantic relatedness datasets, and (ii) JWPL, a Java-based high-performance Wikipedia Application Programming Interface (API) for building natural language processing (NLP) applications.

We explore the use of independent component analysis (ICA) for the automatic extraction of linguistic roles or features of words. The extraction is based on the unsupervised analysis of text corpora. We contrast ICA with singular value decomposition (SVD), widely used in statistical text analysis, in general, and specifically in latent semantic analysis (LSA). However, the representations found using the SVD analysis cannot easily be interpreted by humans. In contrast, ICA applied on word context data gives distinct features which reflect linguistic categories. In this paper, we provide justification for our approach called WordICA, present the WordICA method in detail, compare the obtained results with traditional linguistic categories and with the results achieved using an SVD-based method, and discuss the use of the method in practical natural language engineering solutions such as machine translation systems. As the WordICA method is based on unsupervised learning and thus provides a general means for efficient knowledge acquisition, we foresee that the approach has a clear potential for practical applications.

Web-based discussion fora proliferate on the Internet. These fora consist of threads about specific matters. Existing forum search facilities provide an easy way for finding threads of interest. However, understanding the content of threads is not always trivial. This problem becomes more pressing as threads become longer. It frustrates users that are looking for specific information and also makes it more difficult to make valuable contributions to a discussion. We postulate that having a concise summary of a thread would greatly help forum users. But, how would we best create such summaries? In this paper, we present an automated method of summarising threads in discussion fora. Compared with summarisation of unstructured texts and spoken dialogues, the structural characteristics of threads give important advantages. We studied how to best exploit these characteristics. Messages in threads contain both explicit and implicit references to each other and are structured. Therefore, we term the threads hierarchical dialogues. Our proposed summarisation algorithm produces one summary of an hierarchical dialogue by ‘cherry-picking’ sentences out of the original messages that make up a thread. We try to select sentences usable for obtaining an overview of the discussion. Our method is built around a set of heuristics based on observations of real fora discussions. The data used for this research was in Dutch, but the developed method equally applies to other languages. We evaluated our approach using a prototype. Users judged our summariser as very useful, half of them indicating they would use it regularly or always when visiting fora.

Lapata and Brew (Computational Linguistics, vol. 30, 2004, pp. 295–313) (hereafter LB04) obtain from untagged texts a statistical prior model that is able to generate class preferences for ambiguous Lewin (English Verb Classes and Alternations: A Preliminary Investigation, 1993, University of Chicago Press) verbs (hereafter Levin). They also show that their informative priors, incorporated into a Naive Bayes classifier deduced from hand-tagged data (HTD), can aid in verb class disambiguation. We re-analyse LB04's prior model and show that a single factor (the joint probability of class and frame) determines the predominant class for a particular verb in a particular frame. This means that the prior model cannot be sensitive to fine-grained lexical distinctions between different individual verbs falling in the same class.

We replicate LB04's supervised disambiguation experiments on large-scale data, using deep parsers rather than the shallow parser of LB04. In addition, we introduce a method for training our classifier without using HTD. This relies on knowledge of Levin class memberships to move information from unambiguous to ambiguous instances of each class. We regard this system as unsupervised because it does not rely on human annotation of individual verb instances. Although our unsupervised verb class disambiguator does not match the performance of the ones that make use of HTD, it consistently outperforms the random baseline model. Our experiments also demonstrate that the informative priors derived from untagged texts help improve the performance of the classifier trained on untagged data.

To work effectively in information-rich environments, knowledge workers must be able to distil the most appropriate information from the deluge of information available to them. This is difficult to do manually. Natural language engineers can support these workers by developing information delivery tools, but because of the wide variety of contexts in which information is acquired and delivered, these tools have tended to be domain-specific, ad hoc solutions that are hard to generalise. This paper discusses Myriad, a platform that generalises the integration of sets of resources to a variety of information delivery contexts. Myriad provides resources from natural language generation for discourse planning as well as a service-based architecture for data access. The nature of Myriad's resources is driven by engineering concerns. It focuses on resources that reason about and generate from coarse-grained units of information, likely to be provided by existing information sources, and it supports the integration of pipe-lined planning and template mechanisms. The platform is illustrated in the context of three information delivery applications and is evaluated with respect to its utility.

The distributional similarity methods have proven to be a valuable tool for the induction of semantic similarity. Until now, most algorithms use two-way co-occurrence data to compute the meaning of words. Co-occurrence frequencies, however, need not be pairwise. One can easily imagine situations where it is desirable to investigate co-occurrence frequencies of three modes and beyond. This paper will investigate tensor factorization methods to build a model of three-way co-occurrences. The approach is applied to the problem of selectional preference induction, and automatically evaluated in a pseudo-disambiguation task. The results show that tensor factorization, and non-negative tensor factorization in particular, is a promising tool for Natural Language Processing (nlp).

We investigate the use of instance-based ranking methods for surface realization in natural language generation. Our approach to instance-based natural language generation (IBNLG) employs two components: a rule system that ‘overgenerates’ a number of realization candidates from a meaning representation and an instance-based ranker that scores the candidates according to their similarity to examples taken from a training corpus. We develop an efficient search technique for identifying the optimal candidate based on a novel extension of the A* algorithm. The rule system is produced automatically from a semantically annotated fragment of the Penn Treebank II containing management succession texts. We detail the annotation scheme and grammar induction algorithm and evaluate the efficiency and output of the generator. We also discuss issues such as input coverage (completeness) and fluency that are relevant to surface generation in general.

Thesauri, which list the most salient semantic relations between words, have mostly been compiled manually. Therefore, the inclusion of an entry depends on the subjective decision of the lexicographer. As a consequence, those resources are usually incomplete. In this paper, we propose an unsupervised methodology to automatically discover pairs of semantically related words by highlighting their local environment and evaluating their semantic similarity in local and global semantic spaces. This proposal differs from all other research presented so far as it tries to take the best of two different methodologies, i.e. semantic space models and information extraction models. In particular, it can be applied to extract close semantic relations, it limits the search space to few, highly probable options and it is unsupervised.