Making Tree Kernels Practical for Natural Language Learning

By Alessandro Moschitti. Conference of the European Chapter of the Association for Computational Linguistics, 2006

343 citations recorded by the source corpus.

View Paper: https://www.semanticscholar.org/paper/e1f9ad97af9f09b206e3681f080fcf996e56a5e3

How researchers see this paper

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

1. Aliaksei Severyn
2. Danilo Croce
3. Alessandro Moschitti
4. Grzegorz Chrupała
5. Bonaventura Coppola
6. Roberto Basili
7. Afra Alishahi
8. Yonatan Belinkov

Citations in the corpus

• Syntactic/Semantic Structures for Textual Entailment Recognition (2010, methodology): “To assess the benefit of our approach, we carried out comparative experiments with previous work: especially with the method described in (Zanzotto and Moschitti, 2006; Zanzotto et al., 2009).”
• Modelling input texts: from Tree Kernels to Deep Learning (2015, methodology): “This reduces the time for selecting the matching-node pairs carried out in PTK [93, 94].”
• KERMIT: Complementing Transformer Architectures with Encoders of Explicit Syntactic Interpretations (2020, unknown): “Structural kernels have been defined for both constituency-based (Collins and Duffy, 2002; Moschitti, 2006) and dependency-based parse trees (Culotta and Sorensen, 2004).”
• A General Purpose FrameNet-based Shallow Semantic Parser (2010, methodology): “Such analysis asked for a customized rework of the SVM-Light ”Tree Kernel” extension (Moschitti, 2006), which now includes a specific machine learning model caching capability.”
• Tree Kernels for Semantic Role Labeling (2008, methodology): “9 We employed the SVM-light-TK software , which encodes fast tree kernel evaluation (Moschitti 2006b), and combinations between multiple feature vectors and trees in the SVM-light software (Joachims 1999).”
• Syntactic and Semantic Kernels for Short Text Pair Categorization (2009, methodology): “A closer relationship between questions and answers can be exploited with models presented in (Moschitti and Zanzotto, 2007; Zanzotto and Moschitti, 2006).”
• A Shortest Dependency Path Based Convolutional Neural Network for Protein-Protein Relation Extraction (2016, methodology): “The kernels include bag-of-words kernel [10], all-path kernel [11], subset-tree kernel [12], edit-distance kernel [8], and graph kernel [13], and they have shown effectiveness in PPI task.”
• Correlating Neural and Symbolic Representations of Language (2019, background): “…parse trees as a metric which quantifies similarity between trees as the number of overlapping tree fragments between them, and introduce a polynomial time algorithm to compute these kernels; Moschitti (2006) pro-pose an efficient algorithm for computing tree kernels in linear average running time.”
• Structured Lexical Similarity via Convolution Kernels on Dependency Trees (2011, methodology): “This is very valuable as previous work showed that tree kernels (TK) alone perform lower than models based on manually engineered features for SRL tasks, e.g., (Mos-chitti, 2004; Giuglea and Moschitti, 2004; Giu-glea and Moschitti, 2006; Moschitti, 2006b; Che et al., 2006; Moschitti et al., 2008).”