Generalized Inference with Multiple Semantic Role Labeling Systems

By P. Koomen, Vasin Punyakanok, Dan Roth, Wen-tau Yih. 2005

156 citations recorded by the source corpus.

We present an approach to semantic role labeling (SRL) that takes the output of multiple argument classifiers and combines them into a coherent predicate-argument output by solving an optimization problem. The optimization stage, which is solved via integer linear programming, takes into account both the recommendation of the classifiers and a set of problem specific constraints, and is thus used both to clean the classification results and to ensure structural integrity of the final role labeling. We illustrate a significant improvement in overall SRL performance through this inference.

Subfields: Artificial Intelligence, Natural Language Processing, Information Extraction

How researchers see this paper

The researchers whose trusted networks rank it highest, strongest advocate first. Every researcher gets a different order.

In Artificial Intelligence

In Natural Language Processing

In Information Extraction

Citations in the corpus

Natural Language Processing (Almost) from Scratch (2011, methodology): “…(2001) 93.91% (b) CHUNK System F1 Ando and Zhang (2005) 89.31% Florian et al. (2003) 88.76% Kudo and Matsumoto (2001) 88.31% (c) NER System F1 Koomen et al. (2005) 77.92% Pradhan et al. (2005) 77.30% Haghighi et al. (2005) 77.04% (d) SRL Table 2: State-of-the-art systems on four NLP tasks.”
End-to-end learning of semantic role labeling using recurrent neural networks (2015, methodology): “Similarly, Koomen et al. (Koomen et al., 2005) combined the system in another way.”
Deep Semantic Role Labeling: What Works and What’s Next (2017, neutral): “SRL Constraints Punyakanok et al. (2008); Täckström et al.”
Introduction to the CoNLL-2005 Shared Task: Semantic Role Labeling (2005, methodology): “Another team used also a related learning approach, SNoW, which is a Winnowbased network of linear separators (Punyakanok et al., 2005).”
Special Issue Introduction: Semantic Role Labeling: An Introduction to the Special Issue (2008, unknown): “2005b), or (2) several outputs from the same SRL system obtained by changing input annotations or other internal parameters (Koomen et al. 2005; Toutanova, Haghighi, and Manning 2005).”
Inferring Robot Task Plans from Human Team Meetings: A Generative Modeling Approach with Logic-Based Prior (2013, background): “Processing human dialogue into more machine understandable forms is an important research area (Kruijff, Janıcek, and Lison 2010; Tellex et al. 2011; Koomen et al. 2005; Palmer, Gildea, and Xue 2010; Pradhan et al. 2004), but we view this as a separate problem and do not focus on it in this paper.”
The Importance of Syntactic Parsing and Inference in Semantic Role Labeling (2008, methodology): “As with our full syntactic parse–based SRL system (Koomen et al. 2005), our shallow parsing–based SRL system is based on the system that achieves very competitive results and was one of the top systems in the CoNLL-2004 shared task competition (Carreras and Màrquez 2004).”
Deep Semantic Role Labeling with Self-Attention (2017, methodology): “ syntactic features for capturing the overall sentence structure. Combination of different syntactic parsers was also proposed to avoid prediction risk which was introduced by Surdeanu et al. (2003); Koomen et al. (2005); Pradhan et al. (2013). Beyond these traditional methods above, Collobert et al”
Tree Kernels for Semantic Role Labeling (2008, background): “Furthermore, recent work (Haghighi, Toutanova, and Manning 2005; Punyakanok et al. 2005) has shown that, to achieve high labeling accuracy, joint inference should be applied on the whole predicate–argument structure.”
Interactive and interpretable machine learning models for human machine collaboration (2015, background): “Processing human dialogue into more machine-understandable forms is an important research area [86, 124, 84, 104, 110], but I view this as a separate problem and do not focus on it in this thesis.”
Robust Parsing of the Proposition Bank (2006, methodology): “…use (Charniak, 2000)’s parse trees both for training and testing as well as various other information sources including sets of n -best parse trees (Punyakanok et al., 2005; Haghighi et al., 2005) or chunks (Marquez et al., 2005; Pradhan et al., 2005) and named entities (Surdeanu and Turmo, 2005).”
Emotions from Text: Machine Learning for Text-based Emotion Prediction (2005, methodology): “…us-ing a variation of the Winnow update rule implemented in the SNoW learning architecture (Carl-son et al., 1999), 1 which learns a linear classiﬁer in feature space, and has been successful in several NLP applications, e.g. semantic role labeling (Koomen, Punyakanok, Roth and Yih, 2005).”
Inferring Team Task Plans from Human Meetings: A Generative Modeling Approach with Logic-Based Prior (2015, unknown): “…forms is an important research area (Kruijﬀ, Janıcek, & Lison, 2010; Tellex, Kollar, Dick-erson, Walter, Banerjee, Teller, & Roy, 2011; Koomen, Punyakanok, Roth, & Yih, 2005; Palmer, Gildea, & Xue, 2010; Pradhan, Ward, Hacioglu, Martin, & Jurafsky, 2004), but we view this as a…”