Nikesh Garera

Abhinav Goyal, Anupam Singh, Nikesh Garera

Automation of on-call customer support relies heavily on accurate and efficient speech-to-intent (S2I) systems. Building such systems using multi-component pipelines can pose various challenges because they require large annotated datasets, have higher latency, and have complex deployment. These pipelines are also prone to compounding errors. To overcome these challenges, we discuss an end-to-end (E2E) S2I model for customer support voicebot task in a bilingual setting. We show how we can solve E2E intent classification by leveraging a pre-trained automatic speech recognition (ASR) model with slight modification and fine-tuning on small annotated datasets. Experimental results show that our best E2E model outperforms a conventional pipeline by a relative ~27% on the F1 score.

Mandar Kulkarni, Soumya Chennabasavaraj, Nikesh Garera

With the broad reach of the internet and smartphones, e-commerce platforms have an increasingly diversified user base. Since native language users are not conversant in English, their preferred browsing mode is their regional language or a combination of their regional language and English. From our recent study on the query data, we noticed that many of the queries we receive are code-mix, specifically Hinglish i.e. queries with one or more Hindi words written in English (Latin) script. We propose a transformer-based approach for code-mix query translation to enable users to search with these queries. We demonstrate the effectiveness of pre-trained encoder-decoder models trained on a large corpus of the unlabeled English text for this task. Using generic domain translation models, we created a pseudo-labelled dataset for training the model on the search queries and verified the effectiveness of various data augmentation techniques. Further, to reduce the latency of the model, we use knowledge distillation and weight quantization. Effectiveness of the proposed method has been validated through experimental evaluations and A/B testing. The model is currently live on Flipkart app and website, serving millions of queries.

Tejpalsingh Siledar, Swaroop Nath, Sankara Sri Raghava Ravindra Muddu et al.

Evaluation of opinion summaries using conventional reference-based metrics rarely provides a holistic evaluation and has been shown to have a relatively low correlation with human judgments. Recent studies suggest using Large Language Models (LLMs) as reference-free metrics for NLG evaluation, however, they remain unexplored for opinion summary evaluation. Moreover, limited opinion summary evaluation datasets inhibit progress. To address this, we release the SUMMEVAL-OP dataset covering 7 dimensions related to the evaluation of opinion summaries: fluency, coherence, relevance, faithfulness, aspect coverage, sentiment consistency, and specificity. We investigate Op-I-Prompt a dimension-independent prompt, and Op-Prompts, a dimension-dependent set of prompts for opinion summary evaluation. Experiments indicate that Op-I-Prompt emerges as a good alternative for evaluating opinion summaries achieving an average Spearman correlation of 0.70 with humans, outperforming all previous approaches. To the best of our knowledge, we are the first to investigate LLMs as evaluators on both closed-source and open-source models in the opinion summarization domain.

Mandar Kulkarni, Nikesh Garera

With the democratization of e-commerce platforms, an increasingly diversified user base is opting to shop online. To provide a comfortable and reliable shopping experience, it's important to enable users to interact with the platform in the language of their choice. An accurate query translation is essential for Cross-Lingual Information Retrieval (CLIR) with vernacular queries. Due to internet-scale operations, e-commerce platforms get millions of search queries every day. However, creating a parallel training set to train an in-domain translation model is cumbersome. This paper proposes an unsupervised domain adaptation approach to translate search queries without using any parallel corpus. We use an open-domain translation model (trained on public corpus) and adapt it to the query data using only the monolingual queries from two languages. In addition, fine-tuning with a small labeled set further improves the result. For demonstration, we show results for Hindi to English query translation and use mBART-large-50 model as the baseline to improve upon. Experimental results show that, without using any parallel corpus, we obtain more than 20 BLEU points improvement over the baseline while fine-tuning with a small 50k labeled set provides more than 27 BLEU points improvement over the baseline.

Baban Gain, Ramakrishna Appicharla, Soumya Chennabasavaraj et al.

Community Question-Answering (CQA) portals serve as a valuable tool for helping users within an organization. However, making them accessible to non-English-speaking users continues to be a challenge. Translating questions can broaden the community's reach, benefiting individuals with similar inquiries in various languages. Translating questions using Neural Machine Translation (NMT) poses more challenges, especially in noisy environments, where the grammatical correctness of the questions is not monitored. These questions may be phrased as statements by non-native speakers, with incorrect subject-verb order and sometimes even missing question marks. Creating a synthetic parallel corpus from such data is also difficult due to its noisy nature. To address this issue, we propose a training methodology that fine-tunes the NMT system only using source-side data. Our approach balances adequacy and fluency by utilizing a loss function that combines BERTScore and Masked Language Model (MLM) Score. Our method surpasses the conventional Maximum Likelihood Estimation (MLE) based fine-tuning approach, which relies on synthetic target data, by achieving a 1.9 BLEU score improvement. Our model exhibits robustness while we add noise to our baseline, and still achieve 1.1 BLEU improvement and large improvements on TER and BLEURT metrics. Our proposed methodology is model-agnostic and is only necessary during the training phase. We make the codes and datasets publicly available at \url{https://www.iitp.ac.in/~ai-nlp-ml/resources.html#DomainAdapt} for facilitating further research.

Arnav Attri, Anuj Attri, Pushpak Bhattacharyya et al.

Product recommendations inherently involve comparisons, yet traditional opinion summarization often fails to provide holistic comparative insights. We propose the novel task of generating Query-Focused Comparative Explainable Summaries (QF-CES) using Multi-Source Opinion Summarization (M-OS). To address the lack of query-focused recommendation datasets, we introduce MS-Q2P, comprising 7,500 queries mapped to 22,500 recommended products with metadata. We leverage Large Language Models (LLMs) to generate tabular comparative summaries with query-specific explanations. Our approach is personalized, privacy-preserving, recommendation engine-agnostic, and category-agnostic. M-OS as an intermediate step reduces inference latency approximately by 40% compared to the direct input approach (DIA), which processes raw data directly. We evaluate open-source and proprietary LLMs for generating and assessing QF-CES. Extensive evaluations using QF-CES-PROMPT across 5 dimensions (clarity, faithfulness, informativeness, format adherence, and query relevance) showed an average Spearman correlation of 0.74 with human judgments, indicating its potential for QF-CES evaluation.

Tejpalsingh Siledar, Rupasai Rangaraju, Sankara Sri Raghava Ravindra Muddu et al.

In e-commerce, opinion summarization is the process of summarizing the consensus opinions found in product reviews. However, the potential of additional sources such as product description and question-answers (QA) has been considered less often. Moreover, the absence of any supervised training data makes this task challenging. To address this, we propose a novel synthetic dataset creation (SDC) strategy that leverages information from reviews as well as additional sources for selecting one of the reviews as a pseudo-summary to enable supervised training. Our Multi-Encoder Decoder framework for Opinion Summarization (MEDOS) employs a separate encoder for each source, enabling effective selection of information while generating the summary. For evaluation, due to the unavailability of test sets with additional sources, we extend the Amazon, Oposum+, and Flipkart test sets and leverage ChatGPT to annotate summaries. Experiments across nine test sets demonstrate that the combination of our SDC approach and MEDOS model achieves on average a 14.5% improvement in ROUGE-1 F1 over the SOTA. Moreover, comparative analysis underlines the significance of incorporating additional sources for generating more informative summaries. Human evaluations further indicate that MEDOS scores relatively higher in coherence and fluency with 0.41 and 0.5 (-1 to 1) respectively, compared to existing models. To the best of our knowledge, we are the first to generate opinion summaries leveraging additional sources in a self-supervised setting.

Swaroop Nath, Tejpalsingh Siledar, Sankara Sri Raghava Ravindra Muddu et al.

Reinforcement Learning from Human Feedback (RLHF) has become a dominating strategy in aligning Language Models (LMs) with human values/goals. The key to the strategy is learning a reward model ($\varphi$), which can reflect the latent reward model of humans. While this strategy has proven effective, the training methodology requires a lot of human preference annotation (usually in the order of tens of thousands) to train $\varphi$. Such a large-scale annotation is justifiable when it's a one-time effort, and the reward model is universally applicable. However, human goals are subjective and depend on the task, requiring task-specific preference annotations, which can be impractical to fulfill. To address this challenge, we propose a novel approach to infuse domain knowledge into $\varphi$, which reduces the amount of preference annotation required ($21\times$), omits Alignment Tax, and provides some interpretability. We validate our approach in E-Commerce Opinion Summarization, with a significant reduction in dataset size (to just $940$ samples) while advancing the SOTA ($\sim4$ point ROUGE-L improvement, $68\%$ of times preferred by humans over SOTA). Our contributions include a novel Reward Modeling technique and two new datasets: PromptOpinSumm (supervised data for Opinion Summarization) and OpinPref (a gold-standard human preference dataset). The proposed methodology opens up avenues for efficient RLHF, making it more adaptable to applications with varying human values. We release the artifacts (Code: github.com/efficient-rlhf. PromptOpinSumm: hf.co/prompt-opin-summ. OpinPref: hf.co/opin-pref) for usage under MIT License.

Anuj Attri, Arnav Attri, Pushpak Bhattacharyya et al.

Multi-source Opinion Summarization (M-OS) extends beyond traditional opinion summarization by incorporating additional sources of product metadata such as descriptions, key features, specifications, and ratings, alongside reviews. This integration results in comprehensive summaries that capture both subjective opinions and objective product attributes essential for informed decision-making. While Large Language Models (LLMs) have shown significant success in various Natural Language Processing (NLP) tasks, their potential in M-OS remains largely unexplored. Additionally, the lack of evaluation datasets for this task has impeded further advancements. To bridge this gap, we introduce M-OS-EVAL, a benchmark dataset for evaluating multi-source opinion summaries across 7 key dimensions: fluency, coherence, relevance, faithfulness, aspect coverage, sentiment consistency, specificity. Our results demonstrate that M-OS significantly enhances user engagement, as evidenced by a user study in which, on average, 87% of participants preferred M-OS over opinion summaries. Our experiments demonstrate that factually enriched summaries enhance user engagement. Notably, M-OS-PROMPTS exhibit stronger alignment with human judgment, achieving an average Spearman correlation of \r{ho} = 0.74, which surpasses the performance of previous methodologies.

Arnav Attri, Anuj Attri, Pushpak Bhattacharyya et al.

Customer reviews on e-commerce platforms capture critical affective signals that drive purchasing decisions. However, no existing research has explored the joint task of emotion detection and explanatory span identification in e-commerce reviews - a crucial gap in understanding what triggers customer emotional responses. To bridge this gap, we propose a novel joint task unifying Emotion detection and Opinion Trigger extraction (EOT), which explicitly models the relationship between causal text spans (opinion triggers) and affective dimensions (emotion categories) grounded in Plutchik's theory of 8 primary emotions. In the absence of labeled data, we introduce EOT-X, a human-annotated collection of 2,400 reviews with fine-grained emotions and opinion triggers. We evaluate 23 Large Language Models (LLMs) and present EOT-DETECT, a structured prompting framework with systematic reasoning and self-reflection. Our framework surpasses zero-shot and chain-of-thought techniques, across e-commerce domains.

Nikhil Kothari, Ravindra Nayak, Shreyas Shetty et al.

Large Language Models(LLMs) have shown exceptional abilities, yet training these models can be quite challenging. There is a strong dependence on the quality of data and finding the best instruction tuning set. Further, the inherent limitations in training methods create substantial difficulties to train relatively smaller models with 7B and 13B parameters. In our research, we suggest an improved training method for these models by utilising knowledge from larger models, such as a mixture of experts (8x7B) architectures. The scale of these larger models allows them to capture a wide range of variations from data alone, making them effective teachers for smaller models. Moreover, we implement a novel post-training domain alignment phase that employs domain-specific expert models to boost domain-specific knowledge during training while preserving the model's ability to generalise. Fine-tuning Mistral 7B and 2x7B with our method surpasses the performance of state-of-the-art language models with more than 7B and 13B parameters: achieving up to $7.9$ in MT-Bench and $93.04\%$ on AlpacaEval.

Sri Raghava Muddu, Rupasai Rangaraju, Tejpalsingh Siledar et al.

Opinion summarization in e-commerce encapsulates the collective views of numerous users about a product based on their reviews. Typically, a product on an e-commerce platform has thousands of reviews, each review comprising around 10-15 words. While Large Language Models (LLMs) have shown proficiency in summarization tasks, they struggle to handle such a large volume of reviews due to context limitations. To mitigate, we propose a scalable framework called Xl-OpSumm that generates summaries incrementally. However, the existing test set, AMASUM has only 560 reviews per product on average. Due to the lack of a test set with thousands of reviews, we created a new test set called Xl-Flipkart by gathering data from the Flipkart website and generating summaries using GPT-4. Through various automatic evaluations and extensive analysis, we evaluated the framework's efficiency on two datasets, AMASUM and Xl-Flipkart. Experimental results show that our framework, Xl-OpSumm powered by Llama-3-8B-8k, achieves an average ROUGE-1 F1 gain of 4.38% and a ROUGE-L F1 gain of 3.70% over the next best-performing model.

Abhinav Goyal, Nikesh Garera

Automatic Speech Recognition (ASR) plays a crucial role in voice-based applications. For applications requiring real-time feedback like Voice Search, streaming capability becomes vital. While LSTM/RNN and CTC based ASR systems are commonly employed for low-latency streaming applications, they often exhibit lower accuracy compared to state-of-the-art models due to a lack of future audio frames. In this work, we focus on developing accurate LSTM, attention, and CTC based streaming ASR models for large-scale Hinglish (a blend of Hindi and English) Voice Search. We investigate various modifications in vanilla LSTM training which enhance the system's accuracy while preserving its streaming capabilities. We also address the critical requirement of end-of-speech (EOS) detection in streaming applications. We present a simple training and inference strategy for end-to-end CTC models that enables joint ASR and EOS detection. The evaluation of our model on Flipkart's Voice Search, which handles substantial traffic of approximately 6 million queries per day, demonstrates significant performance gains over the vanilla LSTM-CTC model. Our model achieves a word error rate (WER) of 3.69% without EOS and 4.78% with EOS while also reducing the search latency by approximately ~1300 ms (equivalent to 46.64% reduction) when compared to an independent voice activity detection (VAD) model.

Anand A. Rajasekar, Nikesh Garera

Automatic question answering is an important yet challenging task in E-commerce given the millions of questions posted by users about the product that they are interested in purchasing. Hence, there is a great demand for automatic answer generation systems that provide quick responses using related information about the product. There are three sources of knowledge available for answering a user posted query, they are reviews, duplicate or similar questions, and specifications. Effectively utilizing these information sources will greatly aid us in answering complex questions. However, there are two main challenges present in exploiting these sources: (i) The presence of irrelevant information and (ii) the presence of ambiguity of sentiment present in reviews and similar questions. Through this work we propose a novel pipeline (MSQAP) that utilizes the rich information present in the aforementioned sources by separately performing relevancy and ambiguity prediction before generating a response. Experimental results show that our relevancy prediction model (BERT-QA) outperforms all other variants and has an improvement of 12.36% in F1 score compared to the BERT-base baseline. Our generation model (T5-QA) outperforms the baselines in all content preservation metrics such as BLEU, ROUGE and has an average improvement of 35.02% in ROUGE and 198.75% in BLEU compared to the highest performing baseline (HSSC-q). Human evaluation of our pipeline shows us that our method has an overall improvement in accuracy of 30.7% over the generation model (T5-QA), resulting in our full pipeline-based approach (MSQAP) providing more accurate answers. To the best of our knowledge, this is the first work in the e-commerce domain that automatically generates natural language answers combining the information present in diverse sources such as specifications, similar questions, and reviews data.