Luchen Tan

Wei Yang, Yuqing Xie, Aileen Lin et al.

We demonstrate an end-to-end question answering system that integrates BERT with the open-source Anserini information retrieval toolkit. In contrast to most question answering and reading comprehension models today, which operate over small amounts of input text, our system integrates best practices from IR with a BERT-based reader to identify answers from a large corpus of Wikipedia articles in an end-to-end fashion. We report large improvements over previous results on a standard benchmark test collection, showing that fine-tuning pretrained BERT with SQuAD is sufficient to achieve high accuracy in identifying answer spans.

Mohan Zhang, Luchen Tan, Zhengkai Tu et al.

In the paraphrase generation task, source sentences often contain phrases that should not be altered. Which phrases, however, can be context dependent and can vary by application. Our solution to this challenge is to provide the user with explicit tags that can be placed around any arbitrary segment of text to mean "don't change me!" when generating a paraphrase; the model learns to explicitly copy these phrases to the output. The contribution of this work is a novel data generation technique using distant supervision that allows us to start with a pretrained sequence-to-sequence model and fine-tune a paraphrase generator that exhibits this behavior, allowing user-controllable paraphrase generation. Additionally, we modify the loss during fine-tuning to explicitly encourage diversity in model output. Our technique is language agnostic, and we report experiments in English and Chinese.

He Bai, Peng Shi, Jimmy Lin et al.

Transformers are powerful for sequence modeling. Nearly all state-of-the-art language models and pre-trained language models are based on the Transformer architecture. However, it distinguishes sequential tokens only with the token position index. We hypothesize that better contextual representations can be generated from the Transformer with richer positional information. To verify this, we propose a segment-aware Transformer (Segatron), by replacing the original token position encoding with a combined position encoding of paragraph, sentence, and token. We first introduce the segment-aware mechanism to Transformer-XL, which is a popular Transformer-based language model with memory extension and relative position encoding. We find that our method can further improve the Transformer-XL base model and large model, achieving 17.1 perplexity on the WikiText-103 dataset. We further investigate the pre-training masked language modeling task with Segatron. Experimental results show that BERT pre-trained with Segatron (SegaBERT) can outperform BERT with vanilla Transformer on various NLP tasks, and outperforms RoBERTa on zero-shot sentence representation learning.

He Bai, Peng Shi, Jimmy Lin et al.

The semantics of a text is manifested not only by what is read, but also by what is not read. In this article, we will study how the implicit "not read" information such as end-of-paragraph (\eop) and end-of-sequence (\eos) affect the quality of text generation. Specifically, we find that the pre-trained language model GPT2 can generate better continuations by learning to generate the \eop in the fine-tuning stage. Experimental results on English story generation show that \eop can lead to higher BLEU score and lower \eos perplexity. We also conduct experiments on a self-collected Chinese essay dataset with Chinese-GPT2, a character level LM without \eop or \eos during pre-training. Experimental results show that the Chinese GPT2 can generate better essay endings with \eop.

Ruixue Zhang, Wei Yang, Luyun Lin et al.

Techniques for automatically extracting important content elements from business documents such as contracts, statements, and filings have the potential to make business operations more efficient. This problem can be formulated as a sequence labeling task, and we demonstrate the adaption of BERT to two types of business documents: regulatory filings and property lease agreements. There are aspects of this problem that make it easier than "standard" information extraction tasks and other aspects that make it more difficult, but on balance we find that modest amounts of annotated data (less than 100 documents) are sufficient to achieve reasonable accuracy. We integrate our models into an end-to-end cloud platform that provides both an easy-to-use annotation interface as well as an inference interface that allows users to upload documents and inspect model outputs.

Wei Yang, Yuqing Xie, Luchen Tan et al.

Recently, a simple combination of passage retrieval using off-the-shelf IR techniques and a BERT reader was found to be very effective for question answering directly on Wikipedia, yielding a large improvement over the previous state of the art on a standard benchmark dataset. In this paper, we present a data augmentation technique using distant supervision that exploits positive as well as negative examples. We apply a stage-wise approach to fine tuning BERT on multiple datasets, starting with data that is "furthest" from the test data and ending with the "closest". Experimental results show large gains in effectiveness over previous approaches on English QA datasets, and we establish new baselines on two recent Chinese QA datasets.

Luchen Tan, Jimmy Lin, Adam Roegiest et al.

We examine the effects of different latency penalties in the evaluation of push notification systems, as operationalized in the TREC 2015 Microblog track evaluation. The purpose of this study is to inform the design of metrics for the TREC 2016 Real-Time Summarization track, which is largely modeled after the TREC 2015 evaluation design.

Luchen Tan, Clarke L. A. Clarke

Rank similarity measures provide a method for quantifying differences between search engine results without the need for relevance judgments. For example, the providers of a search service might use such measures to estimate the impact of a proposed algorithmic change across a large number of queries - perhaps millions - identifying those queries where the impact is greatest. In this paper, we propose and validate a family of rank similarity measures, each derived from an associated effectiveness measure. Each member of the family is based on the maximization of effectiveness difference under this associated measure. Computing this maximized effectiveness difference (MED) requires the solution of an optimization problem that varies in difficulty, depending on the associated measure. We present solutions for several standard effectiveness measures, including nDCG, MAP, and ERR. Through an experimental validation, we show that MED reveals meaningful differences between retrieval runs. Mathematically, MED is a metric, regardless of the associated measure. Prior work has established a number of other desiderata for rank similarity in the context of search, and we demonstrate that MED satisfies these requirements. Unlike previous proposals, MED allows us to directly translate assumptions about user behavior from any established effectiveness measure to create a corresponding rank similarity measure. In addition, MED cleanly accommodates partial relevance judgments, and if complete relevance information is available, it reduces to a simple difference between effectiveness values.