This paper investigates a generative history-based parsing model that synchronises the derivation of non-planar graphs representing semantic dependencies with the derivation of dependency trees representing syntactic structures. To process non-planarity online, the semantic transition-based parser uses a new technique to dynamically reorder nodes during the derivation. While the synchronised derivations allow different structures to be built for the semantic non-planar graphs and syntactic dependency trees, useful statistical dependencies between these structures are modeled using latent variables. The resulting synchronous parser achieves competitive performance on the CoNLL-2008 shared task, achieving relative error reduction of 12 % in semantic F score over previously proposed synchronous models that cannot process non-planarity online. 1

Many classification problems with structured outputs can be regarded as a set of interrelated sub-problems where constraints dictate valid variable assignments. The standard approaches to these problems include either independent learning of individual classifiers for each of the sub-problems or joint learning of the entire set of classifiers with the constraints enforced during learning. We propose an intermediate approach where we learn these classifiers in a sequence using previously learned classifiers to guide learning of the next classifier by enforcing constraints between their outputs. We provide a theoretical motivation to explain why this learning protocol is expected to outperform both alternatives when individual problems have different ‘complexity’. This analysis motivates an algorithm for choosing a preferred order of classifier learning. We evaluate our technique on artificial experiments and on the entity and relation identification problem where the proposed method outperforms both joint and independent learning. 1

In this paper we present a machine learning system that finds the scope of negation in biomedical texts. The system consists of two memory-based engines, one that decides if the tokens in a sentence are negation signals, and another that finds the full scope of these negation signals. Our approach to negation detection differs in two main aspects from existing research on negation. First, we focus on finding the scope of negation signals, instead of determining whether a term is negated or not. Second, we apply supervised machine learning techniques, whereas most existing systems apply rule-based algorithms. As far as we know, this way of approaching the negation scope finding task is novel. 1

Our submission to the CoNLL-2008 shared task (Surdeanu et al., 2008) focused on applying a novel method for semantic role labeling to the shared task. Our system first simplifies each sentence to be labeled using a set of hand-constructed rules; the weights of the system are trained on semantic role labeling data to generate simplifications which are as useful as possible for semantic role labeling. Our system is only a semantic role labeling system, and thus did not receive a score for Syntactic Dependencies (or, by extension, a score for the complete problem). Unlike most systems in the shared task, our system took constituency parses as input. On the subtask of semantic dependencies, our system obtained an F1 score of 76.17, the highest in the open task. In this paper we give a high-level overview of the sentence simplification system, and discuss and analyze the modifications to this system required for the CoNLL-2008 shared task. 1 Sentence Simplification The main technical interest of our method is a sentence simplification system. This system is described in depth in (Vickrey and Koller, 2008); for lack of space, we omit many details, including a discussion of related work, from this paper. Current semantic role labeling systems rely primarily on syntactic features in order to identify and classify roles. Features derived from a syntactic parse of the sentence have proven particularly useful (Gildea and Jurafsky, 2002). For example, the syntactic subject of “eat ” is nearly always the

In this paper we present a method for unsupervised semantic role induction which we formalize as a graph partitioning problem. Argument instances of a verb are represented as vertices in a graph whose edge weights quantify their role-semantic similarity. Graph partitioning is realized with an algorithm that iteratively assigns vertices to clusters based on the cluster assignments of neighboring vertices. Our method is algorithmically and conceptually simple, especially with respect to how problem-specific knowledge is incorporated into the model. Experimental results on the CoNLL 2008 benchmark dataset demonstrate that our model is competitive with other unsupervised approaches in terms of F1 whilst attaining significantly higher cluster purity.

We present a normalisation framework for linguistic representations and illustrate its use by normalising the Stanford Dependency graphs (SDs) produced by the Stanford parser into Labelled Stanford Dependency graphs (LSDs). The normalised representations are evaluated both on a testsuite of constructed examples and on free text. The resulting representations improve on standard Predicate/Argument structures produced by SRL by combining role labelling with the semantically oriented features of SDs. Furthermore, the proposed normalisation framework opens the way to stronger normalisation processes which should be useful in reducing the burden on inference. 1

M. F. and P. A. were supported by the FET programme (EU

We propose a generative model based on Temporal Restricted Boltzmann Machines for transition based dependency parsing. The parse tree is built incrementally using a shiftreduce parse and an RBM is used to model each decision step. The RBM at the current time step induces latent features with the help of temporal connections to the relevant previous steps which provide context information. Our parser achieves labeled and unlabeled attachment scores of 88.72 % and 91.65 % respectively, which compare well with similar previous models and the state-of-the-art. 1

Abstract: We present a cost-effective strategy for the creation of a mid-size fine-grained Spanish dependency tree bank of surface-, deep-syntactic and semantic structures as defined in the Meaning-Text Theory. The strategy starts from a small seed dependency corpus, the AnCora corpus, whose annotation is considerably more coarse-grained than our target annotation. We show that this discrepancy can be bridged largely by automatic means. This allows us to develop the resources with limited human effort within a limited period of time. We also propose a preliminary evaluation of the actual amount of work that the annotation process requires.

We formulate the problem of nonprojective dependency parsing as a polynomial-sized integer linear program. Our formulation is able to handle non-local output features in an efficient manner; not only is it compatible with prior knowledge encoded as hard constraints, it can also learn soft constraints from data. In particular, our model is able to learn correlations among neighboring arcs (siblings and grandparents), word valency, and tendencies toward nearlyprojective parses. The model parameters are learned in a max-margin framework by employing a linear programming relaxation. We evaluate the performance of our parser on data in several natural languages, achieving improvements over existing state-of-the-art methods. 1