Conventional customer service chatbots are usually based on human dialogue, yet significant issues in terms of data scale and privacy. In this paper, we present SuperAgent, a customer service chatbot that leverages large-scale and publicly available ecommerce data. Distinct from existing counterparts, SuperAgent takes advantage of data from in-page product descriptions as well as user-generated content from ecommerce websites, which is more practical and cost-effective when answering repetitive questions, freeing up human support staff to answer much higher value questions. We demonstrate SuperAgent as an add-on extension to mainstream web browsers and show its usefulness to user's online shopping experience.

Natural language inference aims to predict whether a premise sentence can infer another hypothesis sentence. Recent progress on this task only relies on a shallow interaction between sentence pairs, which is insufficient for modeling complex relations. In this paper, we present an attention-fused deep matching network (AF-DMN) for natural language inference. Unlike existing models, AF-DMN takes two sentences as input and iteratively learns the attention-aware representations for each side by multi-level interactions. Moreover, we add a self-attention mechanism to fully exploit local context information within each sentence. Experiment results show that AF-DMN achieves state-of-the-art performance and outperforms strong baselines on Stanford natural language inference (SNLI), multi-genre natural language inference (MultiNLI), and Quora duplicate questions datasets.

No abstract

Annotating linguistic data is often a complex, time consuming and expensive endeavor. Even with strict annotation guidelines, human subjects often deviate in their analyses, each bring different biases, interpretations of the task and levels of consistency. The aim of this paper is to explore a way to find out the inconsistencies in the corpus TreeBank which is used for syntactic analysis through the procedure we study the inconsistencies of verb phrase tagging in the corpus TreeBank. At the same time, we can analyze the inconsistencies of verb phrase tagging which are found in the corpus TreeBank in order that we can find a way to improve the consistency of verb phrase tagging automatically which is effective to improve the quality of corpus.

Existing neural machine translation (NMT) systems utilize sequence-to-sequence neural networks to generate target translation word by word, and then make the generated word at each time-step and the counterpart in the references as consistent as possible. However, the trained translation model tends to focus on ensuring the accuracy of the generated target word at the current time-step and does not consider its future cost which means the expected cost of generating the subsequent target translation (i.e., the next target word). To respond to this issue, we propose a simple and effective method to model the future cost of each target word for NMT systems. In detail, a timedependent future cost is estimated based on the current generated target word and its contextual information to boost the training of the NMT model. Furthermore, the learned future context representation at the current time-step is used to help the generation of the next target word in the decoding. Experimental results on three widely-used translation datasets, including the WMT14 German-to-English, WMT14 English-to-French, and WMT17 Chinese-to-English, show that the proposed approach achieves significant improvements over strong Transformer-based NMT baseline.