Abstract

Social Media Platforms (SMPs) like Facebook, Twitter, Instagram etc. have large user base all around the world that generates huge amount of data every second. This includes a lot of posts by fake and spam users, typically used by many organisations around the globe to have competitive edge over others. In this work, we aim at detecting such user accounts in Twitter using a novel approach. We show how to distinguish between Genuine and Spam accounts in Twitter using a combination of Graph Representation Learning and Natural Language Processing techniques.

Abstract

Code-Switching is the embedding of linguistic units or phrases from two or more languages in a single sentence. This phenomenon is practiced in all multilingual communities and is prominent in social media. Consequently, there is a growing need to understand code-switched translations by translating the code-switched text into one of the standard languages or vice versa. Neural Machine translation is a well-studied research problem in the monolingual text. In this paper, we have used the gated convolutional sequences to sequence networks for English-Hinglish translation. The convolutions in the model help to identify the compositional structure in the sequences more easily. The model relies on gating and performs multiple attention steps at encoder and decoder layers.

Abstract

Multimodal Machine Translation (MMT) enriches the source text with visual information for translation. It has gained popularity in recent years, and several pipelines have been proposed in the same direction. Yet, the task lacks quality datasets to illustrate the contribution of visual modality in the translation systems. In this paper, we propose our system for the Multimodal Translation Task of WAT 2021 from English to Hindi. We propose to use mBART, a pretrained multilingual sequence-to-sequence model, for the textual-only translations. Further, we bring the visual information to a textual domain by extracting object tags from the image and enhance the input for the multimodal task. We also explore the robustness of our system by systematically degrading the source text. Finally, we achieve a BLEU score of 44.6 and 51.6 on the test set and challenge set of the task.

Abstract

Memes are one of the most popular types of content used to spread information online. They can influence a large number of people through rhetorical and psychological techniques. The task, Detection of Persuasion Techniques in Texts and Images, is to detect these persuasive techniques in memes. It consists of three subtasks: (A) Multi-label classification using textual content, (B) Multi-label classification and span identification using textual content, and (C) Multi-label classification using visual and textual content. In this paper, we propose a transfer learning approach to fine-tune BERT-based models in different modalities. We also explore the effectiveness of ensembles of models trained in different modalities. We achieve an F1-score of 57.0, 48.2, and 52.1 in the corresponding subtasks.

Abstract

Sentiment-to-sentiment transfer involves changing the sentiment of the given text while preserving the underlying information. In this work, we present a model SentiInc for sentiment-to-sentiment transfer using unpaired mono-sentiment data. Existing sentiment-tosentiment transfer models ignore the valuable sentiment-specific details already present in the text. We address this issue by providing a simple framework for encoding sentiment-specific information in the target sentence while preserving the content information. This is done by incorporating sentiment based loss in the back-translation based style transfer. Extensive experiments over the Yelp dataset show that the SentiInc outperforms state-of-the-art methods by a margin of as large as ∼11% in G-score. The results also demonstrate that our model produces sentiment-accurate and information-preserved sentences.

Abstract

Recent technological advancements in the Internet and Social media usage have resulted in the evolution of faster and efficient platforms of communication. These platforms include visual, textual and speech mediums and have brought a unique social phenomenon called Internet memes. Internet memes are in the form of images with witty, catchy, or sarcastic text descriptions. In this paper, we present a multi-modal sentiment analysis system using deep neural networks combining Computer Vision and Natural Language Processing. Our aim is different than the normal sentiment analysis goal of predicting whether a text expresses positive or negative sentiment; instead, we aim to classify the Internet meme as a positive, negative, or neutral, identify the type of humor expressed and quantify the extent to which a particular effect is being expressed. Our system has been developed using CNN and LSTM and outperformed the baseline score.

Abstract

Off-colour humour is a category of humour which is considered by many to be in poor taste or overly vulgar. Most commonly, off-colour humour contains remarks on particular ethnic group or gender, violence, domestic abuse, acts concerned with sex, excessive swearing or profanity. Blue humour, dark humour and insult humour are types of off-colour humour. Blue and dark humour unlike insult humour are not outrightly insulting in nature but are often misclassified because of the presence of insults and harmful speech. As the primary contributions of this paper we provide an original data-set consisting of nearly 15,000 instances and a novel approach towards resolving the problem of separating dark and blue humour from offensive humour which is essential so that free-speech on the internet is not curtailed. Our experiments show that deep learning methods outperforms other n-grams based approaches like SVM’s, Naive Bayes and Logistic Regression by a large margin.

Abstract

Question Answer pair generation is a task that has been worked upon by multiple researchers in many languages. It has been a topic of interest due to its extensive uses in different fields like self assessment, academics, busi ness website FAQs etc. Many experiments were conducted on Question Answering pair generation in English, concentrating on basic Whquestions with a rulebased approach. We have built the first hybrid machine learning and rulebased solution in Telugu which is ef ficient for short stories or short passages in children’s books. Our work covers the funda mental question forms with the question types: adjective, yes/no, adverb, verb, when, where, whose, quotative, and quantitative(how many/ how much). We constructed rules for question generation using POS tags and UD tags along with linguistic information of the surrounding context of the word.

Abstract

Shayari is a form of poetry mainly popular in the Indian subcontinent, in which the poet expresses his emotions and feelings in a very poetic manner. It is one of the best ways to express our thoughts and opinions. Therefore, it is of prime importance to have an annotated corpus of Hindi shayaris for the task of sentiment analysis. In this paper, we introduce SUKHAN, a dataset consisting of Hindi shayaris along with sentiment polarity labels. To the best of our knowledge, this is the first corpus of Hindi shayaris annotated with sentiment polarity information. This corpus contains a total of 733 Hindi shayaris of various genres. Also, this dataset is of utmost value as all the annotation is done manually by five annotators and this makes it a very rich dataset for training purposes. This annotated corpus is also used to build baseline sentiment classification models using machine learning techniques.

Abstract

The phenomenon of mixing the vocabulary and syntax of multiple languages within the same utterance is called Code-Mixing. This is more evident in multilingual societies. In this paper, we have developed a system for SemEval 2020: Task 9 on Sentiment Analysis for Code-Mixed Social Media Text. Our system first generates two types of embeddings for the social media text. In those, the first one is character level embeddings to encode the character level information and to handle the out-of-vocabulary entries and the second one is FastText word embeddings for capturing morphology and semantics. These two embeddings were passed to the LSTM network and the system outperformed the baseline model.