Inducing multilingual text analysis tools via robust projection across aligned corpora

Abstract

This paper describes a system and set of algorithms for automatically inducing stand-alone monolingual part-of-speech taggers, base noun-phrase bracketers, named-entity taggers and morphological analyzers for an arbitrary foreign language. Case studies include French, Chinese, Czech and Spanish.Existing text analysis tools for English are applied to bilingual text corpora and their output projected onto the second language via statistically derived word alignments. Simple direct annotation projection is quite noisy, however, even with optimal alignments. Thus this paper presents noise-robust tagger, bracketer and lemmatizer training procedures capable of accurate system bootstrapping from noisy and incomplete initial projections.Performance of the induced stand-alone part-of-speech tagger applied to French achieves 96% core part-of-speech (POS) tag accuracy, and the corresponding induced noun-phrase bracketer exceeds 91% F-measure. The induced morphological analyzer achieves over 99% lemmatization accuracy on the complete French verbal system. This achievement is particularly noteworthy in that it required absolutely no hand-annotated training data in the given language, and virtually no language-specific knowledge or resources beyond raw text. Performance also significantly exceeds that obtained by direct annotation projection.

Automatically Inducing a Part-of-Speech Tagger by Projecting from Multiple Source Languages Across Aligned Corpora

“…We hypothesize that a large component of the error rate in the automatically induced text analysis tools generated by [22] is due to morphosyntactic differences between the source and target languages that are specific to each source-target language pair. Therefore, training POS taggers on additional source languages should result in multiple classifiers which produce independently distributed errors on the target language.…”

“…Labelling data by hand is time-consuming; a natural goal is therefore to generate text analysis tools automatically, using minimal resources. Yarowsky et al [22] present methods for automatically inducing various monolingual text analysis tools for an arbitrary target language, using only the corresponding text analysis tool for a source language and a parallel corpus between the source and target languages. Hwa et al [15] induce a parser for Chinese text via projection from English using a similar method to that of [22].…”

“…Yarowsky et al [22] present methods for automatically inducing various monolingual text analysis tools for an arbitrary target language, using only the corresponding text analysis tool for a source language and a parallel corpus between the source and target languages. Hwa et al [15] induce a parser for Chinese text via projection from English using a similar method to that of [22]. Cucerzan and Yarowsky [8] present a method for bootstrapping a POS tagger for an arbitrary target language using only a bilingual dictionary between the source and target languages, a "basic library reference grammar" for the target language, and an existing corpus in the target language.…”

Cross-lingual alignment transfer: a chicken-and-egg story?

“…A widely-used methodology consists in generating automatic annotations for the resource-poor language by projecting linguistic information through word alignment links (see eg. (Yarowsky et al, 2001; for PoS tagging, (Hwa et al, 2005;Lacroix et al, 2016a) for dependency parsing, (Ehrmann et al, 2011) for Named Entity Recognition, (Kozhevnikov and Titov, 2013) for Semantic Role Labeling, etc.). Implementing this methodology requires the existence of (a) parallel corpora aligned at the word level, and (b) annotation and/or tools on the resource-rich side.…”

Combination of Statistical Word Alignments Based on Multiple Preprocessing Schemes

“…1 Although phrasebased approaches to SMT tend to be robust to wordalignment errors (Lopez and Resnik, 2006), improving word-alignment is still useful for other NLP research that is more sensitive to alignment quality, e.g., projection of information across parallel corpora (Yarowsky et al, 2001).…”