Jiwei Li

A Hierarchical Neural Autoencoder for Paragraphs and Documents

Natural language generation of coherent long texts like paragraphs or longer documents is a challenging problem for recurrent networks models. In this paper, we explore an important step toward this generation task: training an LSTM (Longshort term memory) auto-encoder to preserve and reconstruct multi-sentence paragraphs. We introduce an LSTM model that hierarchically builds an embedding for a paragraph from embeddings for sentences and words, then decodes this embedding to reconstruct the original paragraph. We evaluate the reconstructed paragraph using standard metrics like ROUGE and Entity Grid, showing that neural models are able to encode texts in a way that preserve syntactic, semantic, and discourse coherence. While only a first step toward generating coherent text units from neural models, our work has the potential to significantly impact natural language generation and summarization1.

Deep Reinforcement Learning for Dialogue Generation

Recent neural models of dialogue generation offer great promise for generating responses for conversational agents, but tend to be shortsighted, predicting utterances one at a time while ignoring their influence on future outcomes. Modeling the future direction of a dialogue is crucial to generating coherent, interesting dialogues, a need which led traditional NLP models of dialogue to draw on reinforcement learning. In this paper, we show how to integrate these goals, applying deep reinforcement learning to model future reward in chatbot dialogue. The model simulates dialogues between two virtual agents, using policy gradient methods to reward sequences that display three useful conversational properties: informativity (non-repetitive turns), coherence, and ease of answering (related to forward-looking function). We evaluate our model on diversity, length as well as with human judges, showing that the proposed algorithm generates more interactive responses and manages to foster a more sustained conversation in dialogue simulation. This work marks a first step towards learning a neural conversational model based on the long-term success of dialogues.

A Persona-Based Neural Conversation Model

We present persona-based models for handling the issue of speaker consistency in neural response generation. A speaker model encodes personas in distributed embeddings that capture individual characteristics such as background information and speaking style. A dyadic speaker-addressee model captures properties of interactions between two interlocutors. Our models yield qualitative performance improvements in both perplexity and BLEU scores over baseline sequence-to-sequence models, with similar gains in speaker consistency as measured by human judges.

Visualizing and Understanding Neural Models in NLP.

While neural networks have been successfully applied to many NLP tasks the resulting vector-based models are very difficult to interpret. For example its not clear how they achieve {\em compositionality}, building sentence meaning from the meanings of words and phrases. In this paper we describe four strategies for visualizing compositionality in neural models for NLP, inspired by similar work in computer vision. We first plot unit values to visualize compositionality of negation, intensification, and concessive clauses, allow us to see well-known markedness asymmetries in negation. We then introduce three simple and straightforward methods for visualizing a units {\em salience}, the amount it contributes to the final composed meaning: (1) gradient back-propagation, (2) the variance of a token from the average word node, (3) LSTM-style gates that measure information flow. We test our methods on sentiment using simple recurrent nets and LSTMs. Our general-purpose methods may have wide applications for understanding compositionality and other semantic properties of deep networks , and also shed light on why LSTMs outperform simple recurrent nets,

Adversarial Learning for Neural Dialogue Generation

We apply adversarial training to open-domain dialogue generation, training a system to produce sequences that are indistinguishable from human-generated dialogue utterances. We cast the task as a reinforcement learning problem where we jointly train two systems: a generative model to produce response sequences, and a discriminator—analagous to the human evaluator in the Turing test— to distinguish between the human-generated dialogues and the machine-generated ones. In this generative adversarial network approach, the outputs from the discriminator are used to encourage the system towards more human-like dialogue. Further, we investigate models for adversarial evaluation that uses success in fooling an adversary as a dialogue evaluation metric, while avoiding a number of potential pitfalls. Experimental results on several metrics, including adversarial evaluation, demonstrate that the adversarially-trained system generates higher-quality responses than previous baselines

Towards a General Rule for Identifying Deceptive Opinion Spam

Consumers’ purchase decisions are increasingly influenced by user-generated online reviews. Accordingly, there has been growing concern about the potential for posting deceptive opinion spam— fictitious reviews that have been deliberately written to sound authentic, to deceive the reader. In this paper, we explore generalized approaches for identifying online deceptive opinion spam based on a new gold standard dataset, which is comprised of data from three different domains (i.e. Hotel, Restaurant, Doctor), each of which contains three types of reviews, i.e. customer generated truthful reviews, Turker generated deceptive reviews and employee (domain-expert) generated deceptive reviews. Our approach tries to capture the general difference of language usage between deceptive and truthful reviews, which we hope will help customers when making purchase decisions and review portal operators, such as TripAdvisor or Yelp, investigate possible fraudulent activity on their sites. 1

When Are Tree Structures Necessary for Deep Learning of Representations

Recursive neural models, which use syntactic parse trees to recursively generate representations bottom-up, are a popular architecture. However there have not been rigorous evaluations showing for exactly which tasks this syntax-based method is appropriate. In this paper, we benchmark recursive neural models against sequential recurrent neural models, enforcing applesto-apples comparison as much as possible. We investigate 4 tasks: (1) sentiment classification at the sentence level and phrase level; (2) matching questions to answerphrases; (3) discourse parsing; (4) semantic relation extraction. Our goal is to understand better when, and why, recursive models can outperform simpler models. We find that recursive models help mainly on tasks (like semantic relation extraction) that require longdistance connection modeling, particularly on very long sequences. We then introduce a method for allowing recurrent models to achieve similar performance: breaking long sentences into clause-like units at punctuation and processing them separately before combining. Our results thus help understand the limitations of both classes of models, and suggest directions for improving recurrent models.

Do Multi-Sense Embeddings Improve Natural Language Understanding?

Learning a distinct representation for each sense of an ambiguous word could lead to more powerful and fine-grained models of vector-space representations. Yet while ‘multi-sense’ methods have been proposed and tested on artificial wordsimilarity tasks, we don’t know if they improve real natural language understanding tasks. In this paper we introduce a multisense embedding model based on Chinese Restaurant Processes that achieves state of the art performance on matching human word similarity judgments, and propose a pipelined architecture for incorporating multi-sense embeddings into language understanding. We then test the performance of our model on part-of-speech tagging, named entity recognition, sentiment analysis, semantic relation identification and semantic relatedness, controlling for embedding dimensionality. We find that multi-sense embeddings do improve performance on some tasks (part-of-speech tagging, semantic relation identification, semantic relatedness) but not on others (named entity recognition, various forms of sentiment analysis). We discuss how these differences may be caused by the different role of word sense information in each of the tasks. The results highlight the importance of testing embedding models in real applications.

Recursive Deep Models for Discourse Parsing

Text-level discourse parsing remains a challenge: most approaches employ features that fail to capture the intentional, semantic, and syntactic aspects that govern discourse coherence. In this paper, we propose a recursive model for discourse parsing that jointly models distributed representations for clauses, sentences, and entire discourses. The learned representations can to some extent learn the semantic and intentional import of words and larger discourse units automatically,. The proposed framework obtains comparable performance regarding standard discoursing parsing evaluations when compared against current state-of-art systems.

Weakly Supervised User Profile Extraction from Twitter

While user attribute extraction on social media has received considerable attention, existing approaches, mostly supervised, encounter great difficulty in obtaining gold standard data and are therefore limited to predicting unary predicates (e.g., gender). In this paper, we present a weaklysupervised approach to user profile extraction from Twitter. Users’ profiles from social media websites such as Facebook or Google Plus are used as a distant source of supervision for extraction of their attributes from user-generated text. In addition to traditional linguistic features used in distant supervision for information extraction, our approach also takes into account network information, a unique opportunity offered by social media. We test our algorithm on three attribute domains: spouse, education and job; experimental results demonstrate our approach is able to make accurate predictions for users’ attributes based on their tweets. 1