We consider the problem of correcting errors made by English as a Second Language (ESL) writers and address two issues that are essential to making progress in ESL error correction- algorithm selection and model adaptation to the first language of the ESL learner. A variety of learning algorithms have been applied to correct ESL mistakes, but often comparisons were made between incomparable data sets. We conduct an extensive, fair comparison of four popular learning methods for the task, reversing conclusions from earlier evaluations. Our results hold for different training sets, genres, and feature sets. A second key issue in ESL error correction is the adaptation of a model to the first language of the writer. Errors made by non-native speakers exhibit certain regularities and, as we show, models perform much better when they use knowledge about error patterns of the nonnative writers. We propose a novel way to adapt a learned algorithm to the first language of the writer that is both cheaper to implement and performs better than other adaptation methods. 1

We introduce the social study of bullying to the NLP community. Bullying, in both physical and cyber worlds (the latter known as cyberbullying), has been recognized as a serious national health issue among adolescents. However, previous social studies of bullying are handicapped by data scarcity, while the few computational studies narrowly restrict themselves to cyberbullying which accounts for only a small fraction of all bullying episodes. Our main contribution is to present evidence that social media, with appropriate natural language processing techniques, can be a valuable and abundant data source for the study of bullying in both worlds. We identify several key problems in using such data sources and formulate them as NLP tasks, including text classification, role labeling, sentiment analysis, and topic modeling. Since this is an introductory paper, we present baseline results on these tasks using off-the-shelf NLP solutions, and encourage the NLP community to contribute better models in the future.

In this paper, we describe the University of Illinois system that participated in Helping Our Own (HOO), a shared task in text correction. We target several common errors, such as articles, prepositions, word choice, and punctuation errors, and we describe the approaches taken to address each error type. Our system is based on a combination of classifiers, combined with adaptation techniques for article and preposition detection. We ranked first in all three evaluation metrics (Detection, Recognition and Correction) among six participating teams. We also present type-based scores on preposition and article error correction and demonstrate that our approach achieves best performance in each task. 1

Open Information Extraction extracts relations from text without requiring a pre-specified domain or vocabulary. While existing techniques have used only shallow syntactic features, we investigate the use of semantic role labeling techniques for the task of Open IE. Semantic role labeling (SRL) and Open IE, although developed mostly in isolation, are quite related. We compare SRLbased open extractors, which perform computationally expensive, deep syntactic analysis, with TextRunner, an open extractor, which uses shallow syntactic analysis but is able to analyze many more sentences in a fixed amount of time and thus exploit corpus-level statistics. Our evaluation answers questions regarding these systems, including, can SRL extractors, which are trained on PropBank, cope with heterogeneous text found on the Web? Which extractor attains better precision, recall, f-measure, or running time? How does extractor performance vary for binary, n-ary and nested relations? How much do we gain by running multiple extractors? How do we select the optimal extractor given amount of data, available time, types of extractions desired?

This paper proposes a novel multiclass classification method and exhibits its advantage in the domain of text categorization with a large label space and, most importantly, when some of the labels were not observed in the training data. The key insight is the introduction of intermediate aspect variables that encode properties of the labels. Aspect variables serve as a joint representation for observed and unobserved labels. This way the classification problem can be viewed as a structure learning problem with natural constraints on assignments to the aspect variables. We solve the problem as a constrained optimization problem over multiple learners and show significant improvement in classifying short sentences into a large label space of categories, including previously unobserved categories. 1.

A psychologist and an AI researcher speculate on the future of social interaction between humans and androids (robots designed to look and act exactly like people). In reviewing the trajectory of currently developing robotics technologies, the level of android sophistication likely to be achieved in fifty years time is assessed. On the basis of psychological research, obstacles to creating an android indistinguishable from humans are considered. Implications of human-android social interaction from the standpoint of current psychological and AI research are discussed, with speculation on novel psychological issues likely to arise from such interaction. The science of psychology will face a remarkable new set of challenges in grappling with human-android interaction. Android-- 3 Speculations on Human-Android Interaction in the Near and Distant Future How would it feel to interact face-to-face with another person, without knowing whether that person was real? The technology needed to create androids – robots designed to look and just like human beings – is advancing rapidly. We may reach a point within the next hundred years at which androids are sufficiently sophisticated that they

In Semantic Role Labeling (SRL), it is reasonable to globally assign semantic roles due to strong dependencies among arguments. Some relations between arguments significantly characterize the structural information of argument structure. In this paper, we concentrate on thematic hierarchy that is a rank relation restricting syntactic realization of arguments. A loglinear model is proposed to accurately identify thematic rank between two arguments. To import structural information, we employ re-ranking technique to incorporate thematic rank relations into local semantic role classification results. Experimental results show that automatic prediction of thematic hierarchy can help semantic role classification. 1

A psychologist and an AI researcher speculate on the future of social interaction between humans and androids (robots designed to look and act exactly like people). In reviewing the trajectory of currently developing robotics technologies, the level of android sophistication likely to be achieved in fifty years time is assessed. On the basis of psychological research, obstacles to creating an android indistinguishable from humans are considered. Implications of human-android social interaction from the standpoint of current psychological and AI research are discussed, with speculation on novel psychological issues likely to arise from such interaction. The science of psychology will face a remarkable new set of challenges in grappling with human-android interaction.

This paper proposes a general learning framework for a class of problems that require learning over latent intermediate representations. Many natural language processing (NLP) decision problems are defined over an expressive intermediate representation that is not explicit in the input, leaving the algorithm with both the task of recovering a good intermediate representation and learning to classify correctly. Most current systems separate the learning problem into two stages by solving the first step of recovering the intermediate representation heuristically and using it to learn the final classifier. This paper develops a novel joint learning algorithm for both tasks, that uses the final prediction to guide the selection of the best intermediate representation. We evaluate our algorithm on three different NLP tasks – transliteration, paraphrase identification and textual entailment – and show that our joint method significantly improves performance. 1

Many prior studies have investigated the recovery of semantic arguments for nominal predicates. The models in many of these studies have assumed that arguments are independent of each other. This assumption simplifies the computational modeling of semantic arguments, but it ignores the joint nature of natural language. This paper presents a preliminary investigation into the joint modeling of implicit arguments for nominal predicates. The joint model uses propositional knowledge extracted from millions of Internet webpages to help guide prediction. 1