Kaushal Kumar Maurya

In this work, we focus on the task of machine translation (MT) from extremely low-resource language (ELRLs) to English. The unavailability of parallel data, lack of representation from larger multilingual pre-trained models, and limited monolingual data hinder the development of MT systems for ELRLs. However, many ELRLs often share lexical similarities with high-resource languages (HRLs) due to factors such as dialectical variations, geographical proximity, and language structure. We utilize this property to transfer cross-lingual signals from closely related HRL to enable MT for ELRLs. Specifically, we propose a novel unsupervised approach based on selective candidate extraction and noise injection to generate noisy HRLs training data. The noise injection acts as a regularizer and the trained model learns to handle lexical variations such as spelling, grammar, and vocabulary changes, leading to improved cross-lingual transfer to ELRLs. The selective candidates are extracted using BPE merge operations and edit operations, and noise injection is performed using greedy, top-p, and top-k sampling strategies. We evaluate our proposed model on 12 ELRLs from the FLORES-200 benchmark in a zero-shot setting, encompassing two closely related HRLs and LRL families. The proposed model significantly outperformed all the strong baselines, demonstrating its efficacy. The proposed model has comparable performance with the supervised noise injection model.

Query auto-completion (QAC) aims at suggesting plausible completions for a given query prefix. Traditionally, QAC systems have leveraged tries curated from historical query logs to suggest most popular completions. In this context, there are two specific scenarios that are difficult to handle for any QAC system: short prefixes (which are inherently ambiguous) and unseen prefixes. Recently, personalized Natural Language Generation (NLG) models have been proposed to leverage previous session queries as context for addressing these two challenges. However, such NLG models suffer from two drawbacks: (1) some of the previous session queries could be noisy and irrelevant to the user intent for the current prefix, and (2) NLG models cannot directly incorporate historical query popularity. This motivates us to propose a novel NLG model for QAC, Trie-NLG, which jointly leverages popularity signals from trie and personalization signals from previous session queries. We train the Trie-NLG model by augmenting the prefix with rich context comprising of recent session queries and top trie completions. This simple modeling approach overcomes the limitations of trie-based and NLG-based approaches, and leads to state-of-the-art performance. We evaluate the Trie-NLG model using two large QAC datasets. On average, our model achieves huge 57% and 14% boost in MRR over the popular trie-based lookup and the strong BART-based baseline methods, respectively.

Diverse headline generation is an NLP task where given a news article, the goal is to generate multiple headlines that are true to the content of the article, but are different among themselves. This task aims to exhibit and exploit semantically similar one-to-many relationships between a source news article and multiple target headlines. Towards this, we propose a novel model called DivHSK. It has two components: KeySelect for selecting the important keywords, and SeqGen, for finally generating the multiple diverse headlines. In KeySelect, we cluster the self-attention heads of the last layer of the pre-trained encoder and select the most-attentive theme and general keywords from the source article. Then, cluster-specific keyword sets guide the SeqGen, a pre-trained encoder-decoder model, to generate diverse yet semantically similar headlines. The proposed model consistently outperformed existing literature and our strong baselines and emerged as a state-of-the-art model. Additionally, We have also created a high-quality multi-reference headline dataset from news articles.

Recently, the NLP community has witnessed a rapid advancement in multilingual and cross-lingual transfer research where the supervision is transferred from high-resource languages (HRLs) to low-resource languages (LRLs). However, the cross-lingual transfer is not uniform across languages, particularly in the zero-shot setting. Towards this goal, one promising research direction is to learn shareable structures across multiple tasks with limited annotated data. The downstream multilingual applications may benefit from such a learning setup as most of the languages across the globe are low-resource and share some structures with other languages. In this paper, we propose a novel meta-learning framework (called Meta-X_NLG) to learn shareable structures from typologically diverse languages based on meta-learning and language clustering. This is a step towards uniform cross-lingual transfer for unseen languages. We first cluster the languages based on language representations and identify the centroid language of each cluster. Then, a meta-learning algorithm is trained with all centroid languages and evaluated on the other languages in the zero-shot setting. We demonstrate the effectiveness of this modeling on two NLG tasks (Abstractive Text Summarization and Question Generation), 5 popular datasets and 30 typologically diverse languages. Consistent improvements over strong baselines demonstrate the efficacy of the proposed framework. The careful design of the model makes this end-to-end NLG setup less vulnerable to the accidental translation problem, which is a prominent concern in zero-shot cross-lingual NLG tasks.

Despite the recent advancement in NLP research, cross-lingual transfer for natural language generation is relatively understudied. In this work, we transfer supervision from high resource language (HRL) to multiple lowresource languages (LRLs) for natural language generation (NLG). We consider four NLG tasks (text summarization, question generation, news headline generation, and distractor generation) and three syntactically diverse languages, i.e., English, Hindi, and Japanese. We propose an unsupervised crosslingual language generation framework (called ZmBART) that does not use any parallel or pseudo-parallel/back-translated data. In this framework, we further pre-train mBART sequence-to-sequence denoising auto-encoder model with an auxiliary task using monolingual data of three languages. The objective function of the auxiliary task is close to the target tasks which enriches the multi-lingual latent representation of mBART and provides good initialization for target tasks. Then, this model is fine-tuned with task-specific supervised English data and directly evaluated with low-resource languages in the Zero-shot setting. To overcome catastrophic forgetting and spurious correlation issues, we applied freezing model component and data arguclosely related languages. We also show that the proposed character-span noise injection performs better than the unigram-character noise injection.mentation approaches respectively. This simple modeling approach gave us promising results. We experimented with few-shot training (with 1000 supervised data-points) which boosted the model performance further. We performed several ablations and cross-lingual transferability analysis to demonstrate the robustness of ZmBART.

The task of generating incorrect options for multiple-choice questions is termed as distractor generation problem. The task requires high cognitive skills and is extremely challenging to automate. Existing neural approaches for the task leverage encoder-decoder architecture to generate long distractors. However, in this process two critical points are ignored - firstly, many methods use Jaccard similarity over a pool of candidate distractors to sample the distractors. This often makes the generated distractors too obvious or not relevant to the question context. Secondly, some approaches did not consider the answer in the model, which caused the generated distractors to be either answer-revealing or semantically equivalent to the answer.In this paper, we propose a novel Hierarchical Multi-Decoder Network (HMD-Net) consisting of one encoder and three decoders, where each decoder generates a single distractor. To overcome the first problem mentioned above, we include multiple decoders with a dis-similarity loss in the loss function. To address the second problem, we exploit richer interaction between the article, question, and answer with a SoftSel operation and a Gated Mechanism. This enables the generation of distractors that are in context with questions but semantically not equivalent to the answers. The proposed model outperformed all the previous approaches significantly in both automatic and manual evaluations. In addition, we also consider linguistic features and BERT contextual embedding with our base model which further push the model performance.