Dragomir R. Radev

The goal of sentence and document modeling is to accurately represent the meaning of sentences and documents for various Natural Language Processing tasks. In this work, we present Dependency Sensitive Convolutional Neural Networks (DSCNN) as a general-purpose classification system for both sentences and documents. DSCNN hierarchically builds textual representations by processing pretrained word embeddings via Long Short-Term Memory networks and subsequently extracting features with convolution operators. Compared with existing recursive neural models with tree structures, DSCNN does not rely on parsers and expensive phrase labeling, and thus is not restricted to sentence-level tasks. Moreover, unlike other CNN-based models that analyze sentences locally by sliding windows, our system captures both the dependency information within each sentence and relationships across sentences in the same document. Experiment results demonstrate that our approach is achieving state-of-the-art performance on several tasks, including sentiment analysis, question type classification, and subjectivity classification.

This paper presents a comparison of classification methods for linguistic typology for the purpose of expanding an extensive, but sparse language resource: the World Atlas of Language Structures (WALS) (Dryer and Haspelmath, 2013). We experimented with a variety of regression and nearest-neighbor methods for use in classification over a set of 325 languages and six syntactic rules drawn from WALS. To classify each rule, we consider the typological features of the other five rules; linguistic features extracted from a word-aligned Bible in each language; and genealogical features (genus and family) of each language. In general, we find that propagating the majority label among all languages of the same genus achieves the best accuracy in label pre- diction. Following this, a logistic regression model that combines typological and linguistic features offers the next best performance. Interestingly, this model actually outperforms the majority labels among all languages of the same family.

This paper summarizes ongoing research in Natural-Language-Processing-driven citation analysis and describes experiments and motivating examples of how this work can be used to enhance traditional scientometrics analysis that is based on simply treating citations as a ‘vote’ from the citing paper to cited paper. In particular, we describe our dataset for citation polarity and citation purpose, present experimental results on the automatic detection of these indicators, and demonstrate the use of such annotations for studying research dynamics and scientific summarization. We also look at two complementary problems that show up in Natural-Language-Processing-driven citation analysis for a specific target paper. The first problem is extracting citation context, the implicit citation sentences that do not contain explicit anchors to the target paper. The second problem is extracting reference scope, the target relevant segment of a complicated citing sentence that cites multiple papers. We show how these tasks can be helpful in improving sentiment analysis and citation-based summarization.

New scientific concepts, interpreted broadly, are continuously introduced in the literature, but relatively few concepts have a long‐term impact on society. The identification of such concepts is a challenging prediction task that would help multiple parties—including researchers and the general public—focus their attention within the vast scientific literature. In this paper we present a system that predicts the future impact of a scientific concept, represented as a technical term, based on the information available from recently published research articles. We analyze the usefulness of rich features derived from the full text of the articles through a variety of approaches, including rhetorical sentence analysis, information extraction, and time‐series analysis. The results from two large‐scale experiments with 3.8 million full‐text articles and 48 million metadata records support the conclusion that full‐text features are significantly more useful for prediction than metadata‐only features and that the most accurate predictions result from combining the metadata and full‐text features. Surprisingly, these results hold even when the metadata features are available for a much larger number of documents than are available for the full‐text features.

Sophisticated documents like legal cases and biomedical articles can contain unusually long sentences. Extractive summarizers can select such sentences—potentially adding hundreds of unnecessary words to the summary—or exclude them and lose important content. Sentence simplification or compression seems on the surface to be a promising solution. However, compression removes words before the selection algorithm can use them, and simplification generates sentences that may be ambiguous in an extractive summary. We therefore compare the performance of an extractive summarizer selecting from the sentences of the original document with that of the summarizer selecting from sentences shortened in three ways: simplification, compression, and disaggregation, which splits one sentence into several according to rules designed to keep all meaning. We find that on legal cases and biomedical articles, these shortening methods generate ungrammatical output. Human evaluators performed an extrinsic evaluation consisting of comprehension questions about the summaries. Evaluators given compressed, simplified, or disaggregated versions of the summaries answered fewer questions correctly than did those given summaries with unaltered sentences. Error analysis suggests 2 causes: Altered sentences sometimes interact with the sentence selection algorithm, and alterations to sentences sometimes obscure information in the summary. We discuss future work to alleviate these problems.

The New Yorker publishes a weekly captionless cartoon. More than 5,000 readers submit captions for it. The editors select three of them and ask the readers to pick the funniest one. We describe an experiment that compares a dozen automatic methods for selecting the funniest caption. We show that negative sentiment, human-centeredness, and lexical centrality most strongly match the funniest captions, followed by positive sentiment. These results are useful for understanding humor and also in the design of more engaging conversational agents in text and multimodal (vision+text) systems. As part of this work, a large set of cartoons and captions is being made available to the community.

Graphs are a powerful representation formalism that can be applied to a variety of aspects related to language processing. We provide an overview of how Natural Language Processing problems have been projected into the graph framework, focusing in particular on graph construction – a crucial step in modeling the data to emphasize the phenomena targeted.

The ACL Anthology is a large collection of research papers in computational linguistics. Citation data were obtained using text extraction from a collection of PDF files with significant manual postprocessing performed to clean up the results. Manual annotation of the references was then performed to complete the citation network. We analyzed the networks of paper citations, author citations, and author collaborations in an attempt to identify the most central papers and authors. The analysis includes general network statistics, PageRank, metrics across publication years and venues, the impact factor and h‐index, as well as other measures.

Automatically identifying the sentiment polarity of words is a very important task that has been used as the essential building block of many natural language processing systems such as text classification, text filtering, product review analysis, survey response analysis, and on-line discussion mining. We propose a method for identifying the sentiment polarity of words that applies a Markov random walk model to a large word relatedness graph, and produces a polarity estimate for any given word. The model can accurately and quickly assign a polarity sign and magnitude to any word. It can be used both in a semi-supervised setting where a training set of labeled words is used, and in a weakly supervised setting where only a handful of seed words is used to define the two polarity classes. The method is experimentally tested using a gold standard set of positive and negative words from the General Inquirer lexicon. We also show how our method can be used for three-way classification which identifies neutral words in addition to positive and negative words. Our experiments show that the proposed method outperforms the state-of-the-art methods in the semi-supervised setting and is comparable to the best reported values in the weakly supervised setting. In addition, the proposed method is faster and does not need a large corpus. We also present extensions of our methods for identifying the polarity of foreign words and out-of-vocabulary words.