publications

2021

1. Mention Memory: incorporating textual knowledge into Transformers through entity mention attention

   Michiel Jong*, Yury Zemlyanskiy*, Nicholas FitzGerald, Fei Sha, and William Cohen

   2021

   Abs HTML PDF Code

   Natural language understanding tasks such as open-domain question answering often require retrieving and assimilating factual information from multiple sources. We propose to address this problem by integrating a semi-parametric representation of a large text corpus into a Transformer model as a source of factual knowledge. Specifically, our method represents knowledge with ‘mention memory’, a table of dense vector representations of every entity mention in a corpus. The proposed model - TOME - is a Transformer that accesses the information through internal memory layers in which each entity mention in the input passage attends to the mention memory. This approach enables synthesis of and reasoning over many disparate sources of information within a single Transformer model. In experiments using a memory of 150 million Wikipedia mentions, TOME achieves strong performance on several open-domain knowledge-intensive tasks, including the claim verification benchmarks HoVer and FEVER and several entity-based QA benchmarks. We also show that the model learns to attend to informative mentions without any direct supervision. Finally we demonstrate that the model can generalize to new unseen entities by updating the memory without retraining.

2. ReadTwice: Reading Very Large Documents with Memories

   Yury Zemlyanskiy, Joshua Ainslie, Michiel Jong, Philip Pham, Ilya Eckstein, and Fei Sha

   NAACL-HLT 2021

   Abs HTML PDF Code

   Knowledge-intensive tasks such as question answering often require assimilating information from different sections of large inputs such as books or article collections. We propose ReadTwice, a simple and effective technique that combines several strengths of prior approaches to model long-range dependencies with Transformers. The main idea is to read text in small segments, in parallel, summarizing each segment into a memory table to be used in a second read of the text. We show that the method outperforms models of comparable size on several question answering (QA) datasets and sets a new state of the art on the challenging NarrativeQA task, with questions about entire books. Source code and pre-trained checkpoints for ReadTwice can be found at https://goo.gle/research-readtwice.

3. DOCENT: Learning Self-Supervised Entity Representations from Large Document Collections

   Yury Zemlyanskiy, Sudeep Gandhe, Ruining He, Bhargav Kanagal, Anirudh Ravula, Juraj Gottweis, Fei Sha, and Ilya Eckstein

   EACL 2021

   Abs HTML PDF Code

   This paper explores learning rich self-supervised entity representations from large amounts of the associated text. Once pre-trained, these models become applicable to multiple entity-centric tasks such as ranked retrieval, knowledge base completion, question answering, and more. Unlike other methods that harvest self-supervision signals based merely on a local context within a sentence, we radically expand the notion of context to include any available text related to an entity. This enables a new class of powerful, high-capacity representations that can ultimately distill much of the useful information about an entity from multiple text sources, without any human supervision. We present several training strategies that, unlike prior approaches, learn to jointly predict words and entities – strategies we compare experimentally on downstream tasks in the TV-Movies domain, such as MovieLens tag prediction from user reviews and natural language movie search. As evidenced by results, our models match or outperform competitive baselines, sometimes with little or no fine-tuning, and can scale to very large corpora. Finally, we make our datasets and pre-trained models publicly available. This includes Reviews2Movielens (see https://goo.gle/research-docent ), mapping the up to 1B word corpus of Amazon movie reviews (He and McAuley, 2016) to MovieLens tags (Harper and Konstan, 2016), as well as Reddit Movie Suggestions (see this https URL ) with natural language queries and corresponding community recommendations.

2019

1. Self-Attentive, Multi-Context One-Class Classification for Unsupervised Anomaly Detection on Text

   Lukas Ruff, Yury Zemlyanskiy, Robert A. Vandermeulen, Thomas Schnake, and Marius Kloft

   ACL 2019

   Abs HTML PDF Code

   There exist few text-specific methods for unsupervised anomaly detection, and for those that do exist, none utilize pre-trained models for distributed vector representations of words. In this paper we introduce a new anomaly detection method—Context Vector Data Description (CVDD)—which builds upon word embedding models to learn multiple sentence representations that capture multiple semantic contexts via the self-attention mechanism. Modeling multiple contexts enables us to perform contextual anomaly detection of sentences and phrases with respect to the multiple themes and concepts present in an unlabeled text corpus. These contexts in combination with the self-attention weights make our method highly interpretable. We demonstrate the effectiveness of CVDD quantitatively as well as qualitatively on the well-known Reuters, 20 Newsgroups, and IMDB Movie Reviews datasets.

2018

1. Aiming to Know You Better Perhaps Makes Me a More Engaging Dialogue Partner

   Yury Zemlyanskiy, and Fei Sha

   CoNLL 2018

   Abs HTML PDF Code

   There have been several attempts to define a plausible motivation for a chit-chat dialogue agent that can lead to engaging conversations. In this work, we explore a new direction where the agent specifically focuses on discovering information about its interlocutor. We formalize this approach by defining a quantitative metric. We propose an algorithm for the agent to maximize it. We validate the idea with human evaluation where our system outperforms various baselines. We demonstrate that the metric indeed correlates with the human judgments of engagingness.

2014

1. Extracting translation pairs from social network content

   Matthias Eck, Yury Zemlyanskiy, Joy Zhang, and Alex Waibel

   IWSLT 2014

   Abs HTML PDF

   We introduce two methods to collect additional training data for statistical machine translation systems from public social network content. The first method identifies multilingual content where the author self-translated their own post to reach additional friends, fans or customers. Once identified, we can split the post in the language segments and extract translation pairs from this content. The second methods considers web links (URLs) that users add as part of their post to point the reader to a video, article or website. If the same URL is shared from different language users, there is a chance they might give the same comment in their respective language. We use a support vector machine (SVM) as a classifier to identify true translations from all candidate pairs. We collected additional translation pairs using both methods for the language pairs Spanish-English and Portuguese-English. Testing the collected data as additional training data for statistical machine translations on in-domain test sets resulted in very significant improvements of up to 5 BLEU.