Improved Sentiment Detection via Label Transfer from Monolingual to Synthetic Code-Switched Text

Proceedings of the 57th Annual Meeting of the Association for Computational Linguistics, pages 3528–3537 Florence, Italy, July 28 - August 2, 2019. c⃝2019 Association for Computational Linguistics 3528 Improved Sentiment Detection via Label Transfer from Monolingual to Synthetic Code-Switched Text Bidisha Samanta IIT Kharagpur bidisha@ iitkgp.ac.in Niloy Ganguly IIT Kharagpur niloy@ cse.iitkgp.ac.in Soumen Chakrabarti IIT Bombay soumen@ cse.iitb.ac.in Abstract Multilingual writers and speakers often alter- nate between two languages in a single dis- course, a practice called “code-switching”. Existing sentiment detection methods are usu- ally trained on sentiment-labeled monolingual text. Manually labeled code-switched text, es- pecially involving minority languages, is ex- tremely rare. Consequently, the best mono- lingual methods perform relatively poorly on code-switched text. We present an ef- fective technique for synthesizing labeled code-switched text from labeled monolingual text, which is more readily available. The idea is to replace carefully selected subtrees of constituency parses of sentences in the resource-rich language with suitable token spans selected from automatic translations to the resource-poor language. By augment- ing scarce human-labeled code-switched text with plentiful synthetic code-switched text, we achieve significant improvements in sentiment labeling accuracy (1.5%, 5.11%, 7.20%) for three different language pairs (English-Hindi, English-Spanish and English-Bengali). We also get significant gains for hate speech de- tection: 4% improvement using only synthetic text and 6% if augmented with real text. 1 Introduction Sentiment analysis on social media is critical for commerce and governance. Multilingual social media users often use code-switching, particularly to express emotion (Rudra et al., 2016). However, a basic requirement to train any sentiment analysis (SA) system is the availability of large sentiment- labeled corpora. These are extremely challenging to obtain (Chittaranjan et al., 2014; Vyas et al., 2014; Barman et al., 2014), requiring volunteers fluent in multiple languages. We present CSGen, a system which provides supervised SA algorithms with synthesized unlim- ited sentiment-tagged code-switched text, without involving human labelers of code-switched text, or any linguistic theory or grammar for code- switching. These texts can then train state-of- the-art SA algorithms which, until now, primarily worked with monolingual text. A common scenario in code-switching is that a resource-rich source language is mixed with a resource-poor target language. Given a sentiment- labeled source corpus, we first create a parallel corpus by translating to the target language, us- ing a standard translator. Although existing neural machine translators (NMTs) can translate a com- plete source sentence to a target sentence with good quality, it is difficult to translate only des- ignated source segments in isolation because of missing context and lack of coherent semantics. Among our key contributions is a suite of approaches to automatic segment conversion. Broadly, given a source segment selected for code- switching, we propose intuitive ways to select a corresponding segment from the target sentence, based on maximum similarity or minimum dis- similarity with the source segment, so that the segment blends naturally in the outer source con- text. Finally, the generated synthetic sentence is tagged with the same sentiment label as the source sentence. The source segment …

Content truncated to 500 words