Unsupervised models for morpheme segmentation and morphology learning

By Mathias Creutz, Krista Lagus. TSLP, 2007

429 citations recorded by the source corpus.

View Paper: https://www.semanticscholar.org/paper/02711b7e8c73779db802cb00b6cbf407d0872b66

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

Hybrid Morphological Segmentation for Phrase-Based Machine Translation (2016, methodology): “In order to force the Morfessor method to follow the linguistic morphs produced by Omorfi, we added some new features to the Morfessor Baseline implementation by Virpioja et al. (2013).”
Between words and characters: A Brief History of Open-Vocabulary Modeling and Tokenization in NLP (2021, background): “Bostrom and Durrett (2020) claim that Uni-gramLM obtains better segmentation than BPE, both qualitatively (they tend to better correspond to morphemes and Morfessor (Creutz and Lagus, 2007) morphs) and quantitatively (they improve BERT-style models’ performance on modern NLP tasks a little in Englis”
Morphology-aware statistical machine translation based on morphs induced in an unsupervised manner (2007, methodology): “Morfessor (Creutz and Lagus, 2007) is a method for finding morpheme-like units (morphs) of a language in an unsupervised manner.”
Improving sub-word language modeling for Turkish speech recognition (2012, background): “[4] T. Aksungurlu, S. Parlak, H. Sak, and M. Sarac;:lar, "Turkc;:e haber bultenleri ic;:in dil modelleme yak laslmlanmn karsllastlrmasl comparison of language mod eling approaches for turkish broadcast news," .”
A Primer on Neural Network Models for Natural Language Processing (2015, methodology): “…to model the embedding vector of a word as a sum of the word-specific vector if such vector is available, with vectors for the different morphological components that comprise it (the components are derived using Morfessor (Creutz & Lagus, 2007), an unsupervised morphological segmentation method).”
Morfessor 2.0: Python Implementation and Extensions for Morfessor Baseline (2013, methodology): “Prior In the general formulation of Morfessor (Creutz and Lagus, 2007), the model parameters θ are divided into a lexicon L and grammar G.”
Co-learning of Word Representations and Morpheme Representations (2014, methodology): “The ﬁrst method is called Morfessor , which is a public tool implemented based on the minimum descriptions length algo-rithm (Creutz and Lagus, 2007).”
Unsupervised Morphology Rivals Supervised Morphology for Arabic MT (2012, methodology): “The work of Mermer and Akın (2010) and Mer-mer and Saraclar (2011) attempts to integrate morphology and MT more closely than we do, by incorporating bilingual alignment probabilities into a Gibbs-sampled version of Morfessor for Turkish-to-English MT.”
Unsupervised Lexicon Discovery from Acoustic Input (2015, background): “…distribution of phoneme sequences in the input (Cartwright and Brent, 1994; Brent, 1999a; Goldsmith, 2001; Argamon et al., 2004; Goldsmith, 2006; Creutz and Lagus, 2007; Goldwater et al., 2009; Mochihashi et al., 2009; El-sner et al., 2013; Neubig et al., 2012; Heymann et al., 2013; De Marcken,…”
WordICA—emergence of linguistic representations for words by independent component analysis (2010, methodology): “The applied method (Creutz and Lagus 2007) segments words into statistical morphemes based on a corpus of text only, i.e. it does not depend on any given set of morphemes, and is able to segment previously unseen words.”
Morfessor in the Morpho Challenge (2006, background): “Among others, de Marcken (1996), Brent (1999), Goldsmith (2001), and Creutz and Lagus (2002; 2003; 2004; 2005a; 2006) have shown that the above type of model produces segmentations that resemble linguistic morpheme segmentations, when formulated mathematically in a probabilistic framework or…”
Unsupervised Multilingual Learning for Morphological Segmentation (2008, methodology): “However, to demonstrate the competitiveness of this baseline model, we also provide results using MORFESSOR (Creutz and Lagus, 2007), a state-of-the-art unsupervised system for morphological segmentation.”
Unsupervised Morphology Induction Using Word Embeddings (2015, methodology): “In contrast to previous approaches that combine morphology with vector-based word representations (Luong et al., 2013; Botha and Blunsom, 2014), we do not rely on an external morphological analyzer, such as Morfessor (Creutz and La- * Work done at Google, now at Human Longevity Inc. gus, 2007).”