{
"title": "Lexical Normalization for Code-switched Data and its Effect on POS Tagging",
"authors": [
{
"name": "Rob van der Goot",
"email": "robv@itu.dk",
"affiliation": "IT University",
}, {
"name": "Özlem Çetinoğlu",
"email": "ozlem@ims.uni-stuttgart.de",
"affiliation": "IMS, University of Stuttgart"
}
],
"submission_date": "2021-04-01",
"github_link": "https://bitbucket.org/robvanderg/csmonoise",
"paper_link": "https://arxiv.org/pdf/2006.01175.pdf",
"allennlp_version": "1.3",
"datasets": [
{
"name": "Turkish-German Code-switching Normalisation Data",
"link": "https://github.com/ozlemcek/TrDeNormData"
},
{
"name": "Indonesian-English Code-switching Normalisation Data",
"link": "https://github.com/seelenbrecher/code-mixed-normalization"
}
],
"tags": ["Lexical normalization", "POS-tagging", "code-switching"]
}
Description:
Lexical normalization, the translation of noncanonical data to standard language, has shown to improve the performance of many natural language processing tasks on social media. Yet, using multiple languages in one utterance, also called code-switching (CS), is frequently overlooked by these normalization systems, despite its common use in social media. In this paper, we propose three normalization models specifically designed to handle codeswitched data which we evaluate for two language pairs: Indonesian-English (Id-En) and Turkish-German (Tr-De). For the latter, we introduce novel normalization layers and their
corresponding language ID and POS tags for the dataset, and evaluate the downstream effect of normalization on POS tagging. Results show that our CS-tailored normalization models outperform Id-En state of the art and Tr-De monolingual models, and lead to 5.4% relative performance increase for POS tagging as compared to unnormalized input.
Project metadata:
Description:
Lexical normalization, the translation of noncanonical data to standard language, has shown to improve the performance of many natural language processing tasks on social media. Yet, using multiple languages in one utterance, also called code-switching (CS), is frequently overlooked by these normalization systems, despite its common use in social media. In this paper, we propose three normalization models specifically designed to handle codeswitched data which we evaluate for two language pairs: Indonesian-English (Id-En) and Turkish-German (Tr-De). For the latter, we introduce novel normalization layers and their corresponding language ID and POS tags for the dataset, and evaluate the downstream effect of normalization on POS tagging. Results show that our CS-tailored normalization models outperform Id-En state of the art and Tr-De monolingual models, and lead to 5.4% relative performance increase for POS tagging as compared to unnormalized input.