We present a system for the semantic role labeling task. The system combines a machine learning technique with an inference procedure based on integer linear programming that supports the incorporation of linguistic and structural constraints into the decision process. The system is tested on the data provided in the CoNLL-2004 shared task on semantic role labeling and achieves very competitive results. 1

We provide an experimental study of the role of syntactic parsing in semantic role labeling. Our conclusions demonstrate that syntactic parse information is clearly most relevant in the very first stage – the pruning stage. In addition, the quality of the pruning stage cannot be determined solely based on its recall and precision. Instead it depends on the characteristics of the output candidates that make downstream problems easier or harder. Motivated by this observation, we suggest an effective and simple approach of combining different semantic role labeling systems through joint inference, which significantly improves the performance. 1

Mappings to structured output spaces (strings, trees, partitions, etc.) are typically learned using extensions of classification algorithms to simple graphical structures (eg., linear chains) in which search and parameter estimation can be performed exactly. Unfortunately, in many complex problems, it is rare that exact search or parameter estimation is tractable. Instead of learning exact models and searching via heuristic means, we embrace this difficulty and treat the structured output problem in terms of approximate search. We present a framework for learning as search optimization, and two parameter updates with convergence theorems and bounds. Empirical evidence shows that our integrated approach to learning and decoding can outperform exact models at smaller computational cost. 1.

Reasoning with time 1 needs more than just a list of temporal expressions. TimeML—an emerging standard for temporal annotation as a language capturing properties and relationships among timedenoting expressions and events in text—is a good starting point for bridging the gap between temporal analysis of documents and reasoning with the information derived from them. Hard as TimeMLcompliant analysis is, the small size of the only currently available annotated corpus makes it even harder. We address this problem with a hybrid TimeML annotator, which uses cascaded finite-state grammars (for temporal expression analysis, shallow syntactic parsing, and feature generation) together with a machine learning component capable of effectively using large amounts of unannotated data. 1 Temporal Analysis of Documents Many information extraction tasks limit analysis of time to identifying a narrow class of time expressions, which literally specify a temporal point or an interval. For instance, a recent (2004) ACE task is that of temporal expression recognition and normalisation (TERN; see

We present a phrase recognition system based on perceptrons, and an online learning algorithm to train them together. The recognition strategy applies learning in two layers, first at word level, to filter words and form phrase candidates, second at phrase level, to rank phrases and select the optimal ones. We provide a global feedback rule which reflects the dependencies among perceptrons and allows to train them together online. Experimentation on Partial Parsing problems and Named Entity Extraction gives state-of-the-art results on the CoNLL public datasets. We also

Abstract. This work introduces a phrase recognition system based on perceptrons, and a global online learning algorithm to train them together. The method applies to complex domains in which some structure has to be recognized. This global problem is broken down into two layers of local subproblems: a filtering layer, which reduces the search space by identifying plausible phrase candidates, and a ranking layer, which discriminatively builds the optimal phrase structure. A recognitionbased feedback rule is presented which reflects to each local function its committed errors from a global point of view, and allows to train them together online as perceptrons. As a result, the learned functions automatically behave as filters and rankers, rather than binary classifiers, which we argue to be better for this type of problems. Extensive experimentation on partial parsing tasks gives state-of-the-art results and evinces the advantages of the global training method over optimizing each function locally, as in the traditional approach.

Abstract. Several phrase chunkers have been proposed over the past few years. Some state-of-the-art chunkers achieved better performance via integrating external resources, e.g., parsers and additional training data, or combining multiple learners. However, in many languages and domains, such external materials are not easily available and the combination of multiple learners will increase the cost of training and testing. In this paper, we propose a mask method to improve the chunking accuracy. The experimental results show that our chunker achieves better performance in comparison with other deep parsers and chunkers. For CoNLL-2000 data set, our system achieves 94.12 in F rate. For the base-chunking task, our system reaches 92.95 in F rate. When porting to Chinese, the performance of the base-chunking task is 92.36 in F rate. Also, our chunker is quite efficient. The complete chunking time of a 50K words document is about 50 seconds. 1

Abstract. This paper considers the hypothesis that voting between multiple data representations can be more accurate than voting between multiple learning models. This hypothesis has been considered before (cf. [San00]) but the focus was on voting methods rather than the data representations. In this paper, we focus on choosing specific data representations combined with simple majority voting. On the community standard CoNLL-2000 data set, using no additional knowledge sources apart from the training data, we achieved 94.01 Fβ=1 score for arbitrary phrase identification compared to the previous best Fβ=1 93.90. We also obtained 95.23 Fβ=1 score for Base NP identification. Significance tests show that our Base NP identification score is significantly better than the previous comparable best Fβ=1 score of 94.22. Our main contribution is that our model is a fast linear time approach and the previous best approach is significantly slower than our system. 1

abei,nanda,nicolas,roukos,sm1¢ We present a system for identifying and tracking named, nominal, and pronominal mentions of entities within a text document. Our maximum entropy model for mention detection combines two pre-existing named entity taggers (built to extract different entity categories), and other syntactic and morphological feature streams to achieve competitive performance. We developed a novel maximum entropy model for tracking all mentions of an entity within a document. We participated in the Automatic Content Extraction (ACE) evaluation and performed well. We describe our system and present results of the ACE evaluation. 1

Abstract. The design of TimeML as an expressive language for temporal information brings promises, and challenges; in particular, its representational properties raise the bar for traditional information extraction methods applied to the task of text-to-TimeML analysis. A reference corpus, such as TimeBank, is an invaluable asset in this situation; however, certain characteristics of TimeBank—size and consistency, primarily—present challenges of their own. We discuss the design, implementation, and performance of an automatic TimeML-compliant annotator, trained on TimeBank, and deploying a hybrid analytical strategy of mixing aggressive finitestate processing over linguistic annotations with a state-of-the-art machine learning technique capable of leveraging large amounts of unannotated data. The results we report are encouraging in the light of a close analysis of TimeBank; at the same time they are indicative of the need for more infrastructure work, especially in the direction of creating a larger and more robust reference corpus. 1