The robustness of pretrained language models(PLMs) is generally measured using performance drops on two or more domains. However, we do not yet understand the inherent robustness achieved by contributions from different layers of a PLM. We systematically analyze the robustness of these representations layer by layer from two perspectives. First, we measure the robustness of representations by using domain divergence between two domains. We find that i) Domain variance increases from the lower to the upper layers for vanilla PLMs; ii) Models continuously pretrained on domain-specific data (DAPT)(Gururangan et al., 2020) exhibit more variance than their pretrained PLM counterparts; and that iii) Distilled models (e.g., DistilBERT) also show greater domain variance. Second, we investigate the robustness of representations by analyzing the encoded syntactic and semantic information using diagnostic probes. We find that similar layers have similar amounts of linguistic information for data from an unseen domain.

Code-switching is the communication phenomenon where the speakers switch between different languages during a conversation. With the widespread adoption of conversational agents and chat platforms, code-switching has become an integral part of written conversations in many multi-lingual communities worldwide. Therefore, it is essential to develop techniques for understanding and summarizing these conversations. Towards this objective, we introduce the task of abstractive summarization of Hindi-English (Hi-En) code-switched conversations. We also develop the first code-switched conversation summarization dataset - GupShup, which contains over 6,800 Hi-En conversations and their corresponding human-annotated summaries in English (En) and Hi-En. We present a detailed account of the entire data collection and annotation process. We analyze the dataset using various code-switching statistics. We train state-of-the-art abstractive summarization models and report their performances using both automated metrics and human evaluation. Our results show that multi-lingual mBART and multi-view seq2seq models obtain the best performances on this new dataset. We also conduct an extensive qualitative analysis to provide insight into the models and some of their shortcomings.

This paper describes our submission to the 5th edition of the Social Media Mining for Health Applications (SMM4H) shared task 1. Task 1 aims at the automatic classification of tweets that mention a medication or a dietary supplement. This task is specifically challenging due to its highly imbalanced dataset, with only 0.2% of the tweets mentioning a drug. For our submission, we particularly focused on several pretrained encoders for text classification. We achieve an F1 score of 0.75 for the positive class on the test set.

In this paper we present our approach and the system description for Sub Task A and Sub Task B of SemEval 2019 Task 6: Identifying and Categorizing Offensive Language in Social Media. Sub Task A involves identifying if a given tweet is offensive and Sub Task B involves detecting if an offensive tweet is targeted towards someone (group or an individual). Our models for Sub Task A is based on an ensemble of Convolutional Neural Network and Bidirectional LSTM, whereas for Sub Task B, we rely on a set of heuristics derived from the training data. We provide detailed analysis of the results obtained using the trained models. Our team ranked 5th out of 103 participants in Sub Task A, achieving a macro F1 score of 0.807, and ranked 8th out of 75 participants achieving a macro F1 of 0.695.

In this paper we present our approach to tackle the Suggestion Mining from Online Reviews and Forums Sub-Task A. Given a review, we are asked to predict whether the review consists of a suggestion or not. Our model is based on Universal Language Model Fine-tuning for Text Classification. We apply various pre-processing techniques before training the language and the classification model. We further provide analysis of the model. Our team ranked 10th out of 34 participants, achieving an F1 score of 0.7011.

In this paper, we present our approach and the system description for the Social Media Mining for Health Applications (SMM4H) Shared Task 1,2 and 4 (2019). Our main contribution is to show the effectiveness of Transfer Learning approaches like BERT and ULMFiT, and how they generalize for the classification tasks like identification of adverse drug reaction mentions and reporting of personal health problems in tweets. We show the use of stacked embeddings combined with BLSTM+CRF tagger for identifying spans mentioning adverse drug reactions in tweets. We also show that these approaches perform well even with imbalanced dataset in comparison to undersampling and oversampling.

Social media-based text mining in healthcare has received special attention in recent times due to the enhanced accessibility of social media sites like Twitter. The increasing trend of spreading important information in distress can help patients reach out to prospective blood donors in a time bound manner. However such manual efforts are mostly inefficient due to the limited network of a user. In a novel step to solve this problem, we present an annotated Emergency Blood Donation Request (EBDR) dataset to classify tweets referring to the necessity of urgent blood donation requirement. Additionally, we also present an automated feature-based SVM classification technique that can help selective EBDR tweets reach relevant personals as well as medical authorities. Our experiments also present a quantitative evidence that linguistic along with handcrafted heuristics can act as the most representative set of signals this task with an accuracy of 97.89%.