Sumit Chopra

A Neural Attention Model for Abstractive Sentence Summarization

Summarization based on text extraction is inherently limited, but generation-style abstractive methods have proven challenging to build. In this work, we propose a fully data-driven approach to abstractive sentence summarization. Our method utilizes a local attention-based model that generates each word of the summary conditioned on the input sentence. While the model is structurally simple, it can easily be trained end-to-end and scales to a large amount of training data. The model shows significant performance gains on the DUC-2004 shared task compared with several strong baselines.

Question Answering with Subgraph Embeddings

This paper presents a system which learns to answer questions on a broad range of topics from a knowledge base using few hand-crafted features. Our model learns low-dimensional embeddings of words and knowledge base constituents; these representations are used to score natural language questions against candidate answers. Training our system using pairs of questions and structured representations of their answers, and pairs of question paraphrases, yields competitive results on a recent benchmark of the literature.

Abstractive Sentence Summarization with Attentive Recurrent Neural Networks

Abstractive Sentence Summarization generates a shorter version of a given sentence while attempting to preserve its meaning. We introduce a conditional recurrent neural network (RNN) which generates a summary of an input sentence. The conditioning is provided by a novel convolutional attention-based encoder which ensures that the decoder focuses on the appropriate input words at each step of generation. Our model relies only on learned features and is easy to train in an end-to-end fashion on large data sets. Our experiments show that the model significantly outperforms the recently proposed state-of-the-art method on the Gigaword corpus while performing competitively on the DUC-2004 shared task.ive Sentence Summarization generates a shorter version of a given sentence while attempting to preserve its meaning. We introduce a conditional recurrent neural network (RNN) which generates a summary of an input sentence. The conditioning is provided by a novel convolutional attention-based encoder which ensures that the decoder focuses on the appropriate input words at each step of generation. Our model relies only on learned features and is easy to train in an end-to-end fashion on large data sets. Our experiments show that the model significantly outperforms the recently proposed state-of-the-art method on the Gigaword corpus while performing competitively on the DUC-2004 shared task.

#TagSpace: Semantic Embeddings from Hashtags

We describe a convolutional neural network that learns feature representations for short textual posts using hashtags as a supervised signal. The proposed approach is trained on up to 5.5 billion words predicting 100,000 possible hashtags. As well as strong performance on the hashtag prediction task itself, we show that its learned representation of text (ignoring the hashtag labels) is useful for other tasks as well. To that end, we present results on a document recommendation task, where it also outperforms a number of baselines.

Machine Learning and the Spatial Structure of House Prices and Housing Returns

Economists do not have reliable measures of current house values, let alone housing returns. This ignorance underlies the illiquidity of mortgage-backed securities, which in turn feeds back to deepen the sub-prime crisis. Using a massive new data tape of housing transactions in L.A., we demonstrate systematic patterns in the error associated with using the ubiquitous repeat sales methodology to understand house values. In all periods, the resulting indices under-predict sales prices of less expensive homes, and over-predict prices of more expensive homes. The recent period has produced errors that are not only unprecedentedly large in absolute value, but highly systematic: after a few years in which the indices under-predicted prices, they now significantly over-predict them. We introduce new machine learning techniques from computer science to correct for prediction errors that have geographic origins. The results are striking. Accounting for geography significantly reduces the extent of the prediction error, removes many of the systematic patterns, and results in far less deterioration in model performance in the recent period.