TnT – A Statistical Part-of-Speech Tagger

By Thorsten Brants. Applied Natural Language Processing Conference, 2000

2,005 citations recorded by the source corpus.

Trigrams'n'Tags (TnT) is an efficient statistical part-of-speech tagger. Contrary to claims found elsewhere in the literature, we argue that a tagger based on Markov models performs at least as well as other current approaches, including the Maximum Entropy framework. A recent comparison has even shown that TnT performs significantly better for the tested corpora. We describe the basic model of TnT, the techniques used for smoothing and for handling unknown words. Furthermore, we present evaluations on two corpora.

Subfields: Artificial Intelligence

View Paper: https://www.semanticscholar.org/paper/9bd13ac82c011a91542df1a705cf46bb787764cb

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

Citations in the corpus

Polyglot: Distributed Word Representations for Multilingual NLP (2013, result): “We compare our results to a similar experiment conducted in their work, where they trained a TnT tagger (Brants, 2000) on several treebanks.”
Enriching the Knowledge Sources Used in a Maximum Entropy Part-of-Speech Tagger (2000, background): “Among recent top performing methods are Hidden Markov Models (Brants 2000), maximum entropy approaches (Ratnaparkhi 1996), and transformation-based learning (Brill 1994).”
Multilingual Part-of-Speech Tagging with Bidirectional Long Short-Term Memory Models and Auxiliary Loss (2016, methodology): “We hence use three different types of taggers: our implementation of a bi-LSTM; TNT (Brants, 2000)—a second order HMM with suffix trie handling for OOVs; and a freely available CRF-based tagger.”
From distributional to semantic similarity (2004, methodology): “Brill and Wu (1998) call this complementary disagreement complementarity.”
Vector Space Models of Lexical Meaning (2015, methodology): “The part-of-speech tags are from the Penn Treebank tagset (Marcus et al., 1993) and could be automatically applied using any number of freely available part-of-speech taggers (Brants, 2000; Toutanova et al., 2003; Curran & Clark, 2003).”
A Reconfigurable Stochastic Tagger for Languages with Complex Tag Structure (2003, methodology): “for a multitude of binary features (Brants, 2000; Megyesi, 2001), even for large structured tagsets and Slavonic free word-order (Hajič et al.”
Deriving a Large-Scale Taxonomy from Wikipedia (2007, methodology): “In order to ensure precision in applying the patterns, both the Wikipedia and Tipster corpora were preprocessed by a pipeline consisting of a trigram-based statistical POS tagger (Brants, 2000) and a SVM-based chunker (Kudoh & Matsumoto, 2000), to identify noun phrases (NPs).”
A Protégé Plug-In for Ontology Extraction from Text Based on Linguistic Analysis (2004, methodology): “Preconditions are implemented as XPATH expressions over the XML-based linguistic annotation.”
Recovering Semantics of Tables on the Web (2011, methodology): “Boundaries of potential class labels, C , in the text are approximated from the part-of-speech tags (obtained using the TnT tagger [4]) of the sentence words.”
TectoMT: Highly Modular MT System with Tectogrammatics Used as Transfer Layer (2008, methodology): “B3 : Tag the tokens with PTB-style POS tags using a tagger (Brants, 2000).”
Weakly-Supervised Acquisition of Open-Domain Classes and Class Attributes from Web Documents and Query Logs (2008, methodology): “The textual portion of the documents is cleaned of HTML, tokenized, split into sentences and part-ofspeech tagged using the TnT tagger (Brants, 2000).”
FreeLing: An Open-Source Suite of Language Analyzers (2004, methodology): “The PoS tagger is a classical trigram HMM tagger in the style of (Cutting et al., 1992; Brants, 2000), trained on WSJ for English, and on 100,000 words of hand-disambiguated corpus for Spanish and Catalan.”
Unsupervised Part-of-Speech Tagging with Bilingual Graph-Based Projections (2011, methodology): “• Feature-HMM: The vanilla feature-HMM of Berg-Kirkpatrick et al. (2010) (i.e. no additional constraint feature) served as a second baseline.”
Extracting implicit knowledge from text (2009, methodology): “This list was further refined by examining trigram frequencies as reported in the web-derived n-gram collection of Brants and Franz (2006). For each title of the form (Adj N) .”
SVMTool: A general POS Tagger Generator Based on Support Vector Machines (2004, result): “In both cases results clearly outperform the HMM–based T n T part– of–speech tagger (Brants, 2000), compared exactly under the same conditions.”
Challenging Language-Dependent Segmentation for Arabic: An Application to Machine Translation and Part-of-Speech Tagging (2017, methodology): “Probabilistic taggers like HMM-based (Brants, 2000) and sequence learning models like CRF (Lafferty et al., 2001) consider previous words and/or tags to predict the tag of the current word.”
Piggyback: Using Search Engines for Robust Cross-Domain Named Entity Recognition (2011, methodology): “We also reannotated the lowercased datasets with PoS categories using the retrained TnT PoS tag-ger (Brants, 2000) to avoid using non-plausible PoS information.”