A phone call is still one of the primary preferred channels for seniors to express their needs, ask questions, and inform potential problems to their health insurance plans. Alignment Healthis a next-generation, consumer-centric organization that is providing a variety of Medicare Advantage Products for seniors. We combine our proprietary technology platform, AVA, and our high-touch clinical model to provide seniors with care as it should be: high quality, low cost, and accompanied by a vastly improved consumer experience. Our members have the ability to connect with our member services and concierge teams 24/7 for a wide variety of ever-changing reasons through different channels, such as phone, email, and messages. We strive to provide an excellent member experience and ensure our members are getting the help and information they need at every touch —ideally, even before they reach us. This requires ongoing monitoring of reasons for contacting us, ensuring agents are equipped with the right tools and information to serve members, and coming up with proactive strategies to eliminate the need for the call when possible.We developed an NLP-based dynamic call reason tagging and reporting pipeline with an optimized human-in-the-loop approach to enable accurate call reason reporting and monitoring with the ability to see high-level trends as well as drill down into more granular sub-reasons. Our system produces 96.4% precision and 30%-50% better recall in tagging calls with proper reasons. We have also consistently achieved a 60+ Net Promoter Score (NPS) score, which illustrates high consumer satisfaction.

Modelling prosody variation is critical for synthesizing natural and expressive speech in end-to-end text-to-speech (TTS) systems. In this paper, a cross-utterance conditional VAE (CUC-VAE) is proposed to estimate a posterior probability distribution of the latent prosody features for each phoneme by conditioning on acoustic features, speaker information, and text features obtained from both past and future sentences. At inference time, instead of the standard Gaussian distribution used by VAE, CUC-VAE allows sampling from an utterance-specific prior distribution conditioned on cross-utterance information, which allows the prosody features generated by the TTS system to be related to the context and is more similar to how humans naturally produce prosody. The performance of CUC-VAE is evaluated via a qualitative listening test for naturalness, intelligibility and quantitative measurements, including word error rates and the standard deviation of prosody attributes. Experimental results on LJ-Speech and LibriTTS data show that the proposed CUC-VAE TTS system improves naturalness and prosody diversity with clear margins.

Recently, retrieval models based on dense representations are dominant in passage retrieval tasks, due to their outstanding ability in terms of capturing semantics of input text compared to the traditional sparse vector space models. A common practice of dense retrieval models is to exploit a dual-encoder architecture to represent a query and a passage independently. Though efficient, such a structure loses interaction between the query-passage pair, resulting in inferior accuracy. To enhance the performance of dense retrieval models without loss of efficiency, we propose a GNN-encoder model in which query (passage) information is fused into passage (query) representations via graph neural networks that are constructed by queries and their top retrieved passages. By this means, we maintain a dual-encoder structure, and retain some interaction information between query-passage pairs in their representations, which enables us to achieve both efficiency and efficacy in passage retrieval. Evaluation results indicate that our method significantly outperforms the existing models on MSMARCO, Natural Questions and TriviaQA datasets, and achieves the new state-of-the-art on these datasets.

Text matching is a fundamental research problem in natural language understanding. Interaction-based approaches treat the text pair as a single sequence and encode it through cross encoders, while representation-based models encode the text pair independently with siamese or dual encoders. Interaction-based models require dense computations and thus are impractical in real-world applications. Representation-based models have become the mainstream paradigm for efficient text matching. However, these models suffer from severe performance degradation due to the lack of interactions between the pair of texts. To remedy this, we propose a Virtual InteRacTion mechanism (VIRT) for improving representation-based text matching while maintaining its efficiency. In particular, we introduce an interactive knowledge distillation module that is only applied during training. It enables deep interaction between texts by effectively transferring knowledge from the interaction-based model. A light interaction strategy is designed to fully leverage the learned interactive knowledge. Experimental results on six text matching benchmarks demonstrate the superior performance of our method over several state-of-the-art representation-based models. We further show that VIRT can be integrated into existing methods as plugins to lift their performances.

Synonym and antonym practice are the most common practices in our early childhood. It correlated our known words to a better place deep in our intuition. At the beginning of life for a machine, we would like to treat the machine as a baby and built a similar training for it as well to present a qualified performance. In this paper, we present an ensemble model for sentence logistics classification, which outperforms the state-of-art methods. Our approach essentially builds on two models including ERNIE-M and DeBERTaV3. With cross validation and random seeds tuning, we select the top performance models for the last soft ensemble and make them vote for the final answer, achieving the top 6 performance.

Detecting sarcasm and verbal irony from people’s subjective statements is crucial to understanding their intended meanings and real sentiments and positions in social scenarios. This paper describes the X-PuDu system that participated in SemEval-2022 Task 6, iSarcasmEval - Intended Sarcasm Detection in English and Arabic, which aims at detecting intended sarcasm in various settings of natural language understanding. Our solution finetunes pre-trained language models, such as ERNIE-M and DeBERTa, under the multilingual settings to recognize the irony from Arabic and English texts. Our system ranked second out of 43, and ninth out of 32 in Task A: one-sentence detection in English and Arabic; fifth out of 22 in Task B: binary multi-label classification in English; first out of 16, and fifth out of 13 in Task C: sentence-pair detection in English and Arabic.

Sentiment analysis is a classical problem of natural language processing. SemEval 2022 sets a problem on the structured sentiment analysis in task 10, which is also a study-worthy topic in research area. In this paper, we propose a method which can predict structured sentiment information on multiple languages with limited data. The ERNIE-M pretrained language model is employed as a lingual feature extractor which works well on multiple language processing, followed by a graph parser as a opinion extractor. The method can predict structured sentiment information with high interpretability. We apply data augmentation as the given datasets are so small. Furthermore, we use K-fold cross-validation and DeBERTaV3 pretrained model as extra English embedding generator to train multiple models as our ensemble strategies. Experimental results show that the proposed model has considerable performance on both monolingual and cross-lingual tasks.

Existing works on rumor resolution have shown great potential in recognizing word appearance and user participation. However, they ignore the intrinsic propagation mechanisms of rumors and present poor adaptive ability when unprecedented news emerges. To exploit the fine-grained rumor diffusion patterns and generalize rumor resolution methods, we formulate a predecessor task to identify triggering posts, and then exploit their characteristics to facilitate rumor verification. We design a tree-structured annotation interface and extend PHEME dataset with labels on the message level. Data analysis shows that triggers play a critical role in verifying rumors and present similar lingual patterns across irrelevant events. We propose a graph-based model considering the direction and interaction of information flow to implement role-aware rumor resolution. Experimental results demonstrate the effectiveness of our proposed model and progressive scheme.

Sentiment analysis has attracted increasing attention in e-commerce. The sentiment polarities underlying user reviews are of great value for business intelligence. Aspect category sentiment analysis (ACSA) and review rating prediction (RP) are two essential tasks to detect the fine-to-coarse sentiment polarities. ACSA and RP are highly correlated and usually employed jointly in real-world e-commerce scenarios. While most public datasets are constructed for ACSA and RP separately, which may limit the further exploitation of both tasks. To address the problem and advance related researches, we present a large-scale Chinese restaurant review dataset ASAP including 46, 730 genuine reviews from a leading online-to-offline (O2O) e-commerce platform in China. Besides a 5-star scale rating, each review is manually annotated according to its sentiment polarities towards 18 pre-defined aspect categories. We hope the release of the dataset could shed some light on the field of sentiment analysis. Moreover, we propose an intuitive yet effective joint model for ACSA and RP. Experimental results demonstrate that the joint model outperforms state-of-the-art baselines on both tasks.

We consider the problem of scaling automated suggested replies for a commercial email application to multiple languages. Faced with increased compute requirements and low language resources for language expansion, we build a single universal model for improving the quality and reducing run-time costs of our production system. However, restricted data movement across regional centers prevents joint training across languages. To this end, we propose a multi-lingual multi-task continual learning framework, with auxiliary tasks and language adapters to train universal language representation across regions. The experimental results show positive cross-lingual transfer across languages while reducing catastrophic forgetting across regions. Our online results on real user traffic show significant CTR and Char-saved gain as well as 65% training cost reduction compared with per-language models. As a consequence, we have scaled the feature in multiple languages including low-resource markets.

Clinical trials often require that patients meet eligibility criteria (e.g., have specific conditions) to ensure the safety and the effectiveness of studies. However, retrieving eligible patients for a trial from the electronic health record (EHR) database remains a challenging task for clinicians since it requires not only medical knowledge about eligibility criteria, but also an adequate understanding of structured query language (SQL). In this paper, we introduce a new dataset that includes the first-of-its-kind eligibility-criteria corpus and the corresponding queries for criteria-to-sql (Criteria2SQL), a task translating the eligibility criteria to executable SQL queries. Compared to existing datasets, the queries in the dataset here are derived from the eligibility criteria of clinical trials and include Order-sensitive, Counting-based, and Boolean-type cases which are not seen before. In addition to the dataset, we propose a novel neural semantic parser as a strong baseline model. Extensive experiments show that the proposed parser outperforms existing state-of-the-art general-purpose text-to-sql models while highlighting the challenges presented by the new dataset. The uniqueness and the diversity of the dataset leave a lot of research opportunities for future improvement.

Cross-domain sentiment classification aims to predict sentiment polarity on a target domain utilizing a classifier learned from a source domain. Most existing adversarial learning methods focus on aligning the global marginal distribution by fooling a domain discriminator, without taking category-specific decision boundaries into consideration, which can lead to the mismatch of category-level features. In this work, we propose an adversarial category alignment network (ACAN), which attempts to enhance category consistency between the source domain and the target domain. Specifically, we increase the discrepancy of two polarity classifiers to provide diverse views, locating ambiguous features near the decision boundaries. Then the generator learns to create better features away from the category boundaries by minimizing this discrepancy. Experimental results on benchmark datasets show that the proposed method can achieve state-of-the-art performance and produce more discriminative features.

Multiple emotions with different intensities are often evoked by events described in documents. Oftentimes, such event information is hidden and needs to be discovered from texts. Unveiling the hidden event information can help to understand how the emotions are evoked and provide explainable results. However, existing studies often ignore the latent event information. In this paper, we proposed a novel interpretable relevant emotion ranking model with the event information incorporated into a deep learning architecture using the event-driven attentions. Moreover, corpus-level event embeddings and document-level event distributions are introduced respectively to consider the global events in corpus and the document-specific events simultaneously. Experimental results on three real-world corpora show that the proposed approach performs remarkably better than the state-of-the-art emotion detection approaches and multi-label approaches. Moreover, interpretable results can be obtained to shed light on the events which trigger certain emotions.

Existing neural models usually predict the tag of the current token independent of the neighboring tags. The popular LSTM-CRF model considers the tag dependencies between every two consecutive tags. However, it is hard for existing neural models to take longer distance dependencies between tags into consideration. The scalability is mainly limited by the complex model structures and the cost of dynamic programming during training. In our work, we first design a new model called “high order LSTM” to predict multiple tags for the current token which contains not only the current tag but also the previous several tags. We call the number of tags in one prediction as “order”. Then we propose a new method called Multi-Order BiLSTM (MO-BiLSTM) which combines low order and high order LSTMs together. MO-BiLSTM keeps the scalability to high order models with a pruning technique. We evaluate MO-BiLSTM on all-phrase chunking and NER datasets. Experiment results show that MO-BiLSTM achieves the state-of-the-art result in chunking and highly competitive results in two NER datasets.

Text might express or evoke multiple emotions with varying intensities. As such, it is crucial to predict and rank multiple relevant emotions by their intensities. Moreover, as emotions might be evoked by hidden topics, it is important to unveil and incorporate such topical information to understand how the emotions are evoked. We proposed a novel interpretable neural network approach for relevant emotion ranking. Specifically, motivated by transfer learning, the neural network is initialized to make the hidden layer approximate the behavior of topic models. Moreover, a novel error function is defined to optimize the whole neural network for relevant emotion ranking. Experimental results on three real-world corpora show that the proposed approach performs remarkably better than the state-of-the-art emotion detection approaches and multi-label learning methods. Moreover, the extracted emotion-associated topic words indeed represent emotion-evoking events and are in line with our common-sense knowledge.

Text might contain or invoke multiple emotions with varying intensities. As such, emotion detection, to predict multiple emotions associated with a given text, can be cast into a multi-label classification problem. We would like to go one step further so that a ranked list of relevant emotions are generated where top ranked emotions are more intensely associated with text compared to lower ranked emotions, whereas the rankings of irrelevant emotions are not important. A novel framework of relevant emotion ranking is proposed to tackle the problem. In the framework, the objective loss function is designed elaborately so that both emotion prediction and rankings of only relevant emotions can be achieved. Moreover, we observe that some emotions co-occur more often while other emotions rarely co-exist. Such information is incorporated into the framework as constraints to improve the accuracy of emotion detection. Experimental results on two real-world corpora show that the proposed framework can effectively deal with emotion detection and performs remarkably better than the state-of-the-art emotion detection approaches and multi-label learning methods.

Knowledge embedding, which projects triples in a given knowledge base to d-dimensional vectors, has attracted considerable research efforts recently. Most existing approaches treat the given knowledge base as a set of triplets, each of whose representation is then learned separately. However, as a fact, triples are connected and depend on each other. In this paper, we propose a graph aware knowledge embedding method (GAKE), which formulates knowledge base as a directed graph, and learns representations for any vertices or edges by leveraging the graph’s structural information. We introduce three types of graph context for embedding: neighbor context, path context, and edge context, each reflects properties of knowledge from different perspectives. We also design an attention mechanism to learn representative power of different vertices or edges. To validate our method, we conduct several experiments on two tasks. Experimental results suggest that our method outperforms several state-of-art knowledge embedding models.

Zara, or ‘Zara the Supergirl’ is a virtual robot, that can exhibit empathy while interacting with an user, with the aid of its built in facial and emotion recognition, sentiment analysis, and speech module. At the end of the 5-10 minute conversation, Zara can give a personality analysis of the user based on all the user utterances. We have also implemented a real-time emotion recognition, using a CNN model that detects emotion from raw audio without feature extraction, and have achieved an average of 65.7% accuracy on six different emotion classes, which is an impressive 4.5% improvement from the conventional feature based SVM classification. Also, we have described a CNN based sentiment analysis module trained using out-of-domain data, that recognizes sentiment from the speech recognition transcript, which has a 74.8 F-measure when tested on human-machine dialogues.