Tregex and Tsurgeon: tools for querying and manipulating tree data structures

By Roger Levy, Galen Andrew. International Conference on Language Resources and Evaluation, 2006

421 citations recorded by the source corpus.

With syntactically annotated corpora becoming increasingly available for a variety of languages and grammatical frameworks, tree query tools have proven invaluable to linguists and computer scientists for both data exploration and corpus-based research. We provide a combined engine for tree query (Tregex) and manipulation (Tsurgeon) that can operate on arbitrary tree data structures with no need for preprocessing. Tregex remedies several expressive and implementational limitations of existing query tools, while Tsurgeon is to our knowledge the most expressive tree manipulation utility available.

Subfields: Artificial Intelligence, Computer Networks and Communications, Information Systems

View Paper: https://www.semanticscholar.org/paper/4e751a9017dbd825418d35afba6442df958a973c

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 Computer Networks and Communications

In Information Systems

Citations in the corpus

Referential translation machines for predicting semantic similarity (2015, methodology): “Figure 2 depicts the parsing output obtained by CCL for the following sentence from WSJ23 3 : Many fund managers argue that now ’ s the time to buy : We use Tregex (Levy and Andrew 2006) for visualizing the output parse trees presented on the left of Fig.”
Automatic analysis of syntactic complexity in second language writing (2010, methodology): “Given that it is only possible to search parse trees one by one using Tregex, we specify that a T-unit can only occur within a sentence punctuated by the writer (Homburg 1984, Ishikawa 1995).”
Yandex School of Data Analysis approach to English-Turkish translation at WMT16 News Translation Task (2016, methodology): “Rule-based reorderer modiﬁes parse trees using rules similar to Tregex (Levy and Andrew, 2006), adapted to dependency trees 1 .”
Generating Natural Language Questions to Support Learning On-Line (2013, methodology): “As seen in Wyse and Piwek (2009), who use Stanford Tregex (Levy and Andrew, 2006) for pattern matching, the specificity of syntactic patterns can make it difficult to specify a syntactic pattern of the desired scope.”
Part-of-Speech Tagging from 97% to 100%: Is It Time for Some Linguistics? (2011, methodology): “Here, we use Tsurgeon scripts [21], which work by matching a tree pattern using Tregex (a tgrep-like language; cf. [22]) and then performing operations on matched parts of the tree.”
Referential Translation Machines for Predicting Translation Quality (2014, methodology): “We use Tregex (Levy and Andrew, 2006) for visualizing the output parse trees presented on the left.”
Modeling Semantic Containment and Exclusion in Natural Language Inference (2008, background): “We deﬁne a list of operator types affecting projectivity (e.g., implicatives like refuse to , prepositions like without ), and for each type we specify its arity and a Tregex tree pattern (Levy and Andrew, 2006) which permits us to identify its occurrences in our Treebank parses.”
On the negativity of negation (2010, methodology): “…precision was called for, so I parsed the data using the freely available Stanford parser (Klein & Manning 2003a,b) and then extracted the requisite patterns using the Stanford Tregex program (Levy & Andrew 2006), which provides intuitive functionality for extracting patterns from parsetrees.”
Self-Training PCFG Grammars with Latent Annotations Across Languages (2009, methodology): “We used slightly modified versions of the treebanks; empty nodes and nonterminal-yield unary rules5, e.g., NP→VP, are deleted using tsurgeon (Levy and Andrew, 2006).”
The Stanford Typed Dependencies Representation (2008, methodology): “Then for each grammatical relation, patterns are deﬁned over the phrase structure parse tree us-ing the tree-expression syntax deﬁned by tregex (Levy and Andrew, 2006).”
Learning to recognize features of valid textual entailments (2006, methodology): “…rules, heads of constituents are ﬁrst identiﬁed using a mod-iﬁed version of the Collins head rules that favor semantic heads (such as lexical verbs rather than aux-iliaries), and dependents of heads are typed using tregex patterns (Levy and Andrew, 2006), an extension of the tgrep pattern language.”
A Gold Standard Dependency Corpus for English (2014, methodology): “The primary source of the improvement is improvements in the underlying Tregex library, with more compact matching expressions making up a smaller but still substantial fraction of the improvement.”
Natural Logic for Textual Inference (2007, background): “We define a list of downward-monotone and non-monotone expressions, and for each item we specify its arity and a Tregex pattern (Levy and Andrew, 2006) which permits us to identify its occurrences.”
The subject-relative advantage in Chinese: Evidence for expectation-based processing (2015, methodology): “We carried out four corpus searches on the Chinese Treebank 7.0 (Xue, Xia, Chiou, & Palmer, 2005), which consists of 51,447 fully parsed sentences (1,196,329 words) using the pattern matching tool Tregex (Levy & Andrew, 2006).”
Generating Typed Dependency Parses from Phrase Structure Parses (2006, methodology): “For each grammatical relation, we deﬁne one or more patterns over the phrase structure parse tree (using the tree-expression syntax deﬁned by tregex (Levy and Andrew, 2006)).”