Sentiment analysis has been an active area of research in the past two decades and recently, with the advent of social media, there has been an increasing demand for sentiment analysis on social media texts. Since the social media texts are not in one language and are largely code-mixed in nature, the traditional sentiment classification models fail to produce acceptable results. This paper tries to solve this very research problem and uses bi-directional LSTMs along with language tagging, to facilitate sentiment tagging of code-mixed Tamil texts that have been extracted from social media. The presented algorithm, when evaluated on the test data, garnered precision, recall, and F1 scores of 0.59, 0.66, and 0.58 respectively.

JUNLP@Dravidian-CodeMix-FIRE2020: Sentiment Classification of Code-Mixed Tweets using Bi-Directional RNN and Language Tags

Code-mixing is a phenomenon which arises mainly in multilingual societies. Multilingual people, who are well versed in their native languages and also English speakers, tend to code-mix using English-based phonetic typing and the insertion of anglicisms in their main language. This linguistic phenomenon poses a great challenge to conventional NLP domains such as Sentiment Analysis, Machine Translation, and Text Summarization, to name a few. In this work, we focus on working out a plausible solution to the domain of Code-Mixed Sentiment Analysis. This work was done as participation in the SemEval-2020 Sentimix Task, where we focused on the sentiment analysis of English-Hindi code-mixed sentences. our username for the submission was “sainik.mahata” and team name was “JUNLP”. We used feature extraction algorithms in conjunction with traditional machine learning algorithms such as SVR and Grid Search in an attempt to solve the task. Our approach garnered an f1-score of 66.2% when tested using metrics prepared by the organizers of the task.

/pdf/junlp-at-semeval-2020-task-9-sentiment-analysis-of-hindi-1pf7db8naw.pdf

JUNLP at SemEval-2020 Task 9: Sentiment Analysis of Hindi-English Code Mixed Data Using Grid Search Cross Validation

Code-mixing is a phenomenon which arises mainly in multilingual societies. Multilingual people, who are well versed in their native languages and also English speakers, tend to code-mix using English-based phonetic typing and the insertion of anglicisms in their main language. This linguistic phenomenon poses a great challenge to conventional NLP domains such as Sentiment Analysis, Machine Translation, and Text Summarization, to name a few. In this work, we focus on working out a plausible solution to the domain of Code-Mixed Sentiment Analysis. This work was done as participation in the SemEval-2020 Sentimix Task, where we focused on the sentiment analysis of English-Hindi code-mixed sentences. our username for the submission was "sainik.mahata" and team name was "JUNLP". We used feature extraction algorithms in conjunction with traditional machine learning algorithms such as SVR and Grid Search in an attempt to solve the task. Our approach garnered an f1-score of 66.2\% when tested using metrics prepared by the organizers of the task.

JUNLP@SemEval-2020 Task 9:Sentiment Analysis of Hindi-English code mixed data using Grid Search Cross Validation

Code-switching, the combination of more than one language within a single utterance is popular in social media sites for informal communication, generating a huge amount of data, though not fully a...

Deep Insights of Erroneous Bengali–English Code-Mixed Bilingual Language

Due to their widespread use around the globe, especially in a multilingual country like India, digital assistants are programmed to be capable of understanding speech and conversing for an array of languages. The objective is to understand user intent and respond with human expression, achieving a great degree of accuracy and efficiency. However, a multilingual society such as India has grown quite comfortable using a combination of two or more languages while communicating, commonly referenced as code-switching. Code-switching is not inherently taken care of by today’s intelligent assistants. Due to the lack of standardized dataset, the presented work includes corpus generation of code-switching frequently asked queries was generated. It includes a study of intent classification of code-switching queries with various vectorizers and classifier combinations. This work also comprises a pipeline to identify the user intent and provide a desired response using keyword extraction and named entity recognition models on code-switching queries. Standardized code-switching metrics applied on the generated corpus to estimate the strength of code-switching yielded high intensity. The results of the intent classification study on code-switching queries obtained by evaluating the models on standardized classification metrics have yielded high accuracy. Named entity recognition in code-switching queries also have yielded high precision and recall scores.

Regional Language Code-Switching for Natural Language Understanding and Intelligent Digital Assistants

In this work, we have proposed an efficient and less resource intensive strategy for parsing and analyzing switching points in code-mixed data. Specifically, we have explored the rules of code-switching in Hindi–English code-mixed data. The work involves code-mixed text extraction, translation of the extracted texts to its pure form, forming word pairs, annotation of these using of Parts-of-Speech tags and recognition of the rules that govern switching in code-mixed text. We have created three models, viz. baseline model, lexicon-based model, and machine learning-based model, and found out the individual accuracies of these models.

Sushnat Makhija

Papers

Analyzing Code-Switching Rules for English–Hindi Code-Mixed Text