We discuss several improvements to the Question Generation Shared Task Evaluation Challenge (QGSTEC) system developed at the University of Pennsylvania in 2010. In addition to enhancing the question generation rules, we have implemented two new components to improve the ranking process. We use topic scoring, a technique developed for summarization, to identify important information for questioning, and language model probabilities to measure grammaticality. Preliminary experiments show that our approach is feasible.

This paper shows that automatically generated questions can help classify children’s spoken responses to a reading tutor teaching them to generate their own questions. We use automatic question generation to model and classify children’s prompted spoken questions about stories. On distinguishing complete and incomplete questions from irrelevant speech and silence, a language model built from automatically generated questions out-performs a trigram language model that does not exploit the structure of questions.

Automated question generation has been explored for a broad range of tasks. However, an important task for which limited work on question generation has been undertaken is writing support. Writing support systems, particularly for novice writers who are acquiring the fundamentals of writing, can scaffold the complex processes that bear on writing. Novice writers face significant challenges in creative writing. Their stories often lack the expressive prose that characterizes texts produced by their expert writer counterparts. A story that is composed by a novice writer may also lack a compelling plot, may not effectively utilize a story’s setting, characters, and props, and may describe events that play out in an unpredictable or confusing order. We propose an automatic question generation framework that is designed to stimulate the cognitive processes associated with creative writing. The framework utilizes semantic role labeling and discourse parsing applied to the initial drafts of the writer’s passage to generate questions to promote creativity.

Text annotations are of great use to researchers in the language sciences, and much effort has been invested in creating annotated corpora for an wide variety of purposes. Unfortunately, software support for these corpora tends to be quite limited: it is usually ad-hoc, poorly designed and documented, or not released for public use. I describe an annotation tool, the Story Workbench, which provides a generic platform for text annotation. It is free, open-source, cross-platform, and user friendly. It provides a number of common text annotation operations, including representations (e.g., tokens, sentences, parts of speech), functions (e.g., generation of initial annotations by algorithm, checking annotation validity by rule, fully manual manipulation of annotations) and tools (e.g., distributing texts to annotators via version control, merging doubly-annotated texts into a single file). The tool is extensible at many different levels, admitting new representations, algorithm, and tools. I enumerate ten important features and illustrate how they support the annotation process at three levels: (1) annotation of individual texts by a single annotator, (2) double-annotation of texts by two annotators and an adjudicator, and (3) annotation scheme development. The Story Workbench is scheduled for public release in March 2012.

In this paper we propose a data intensive approach for inferring sentence-internal temporal relations. Temporal inference is relevant for practical NLP applications which either extract or synthesize temporal information (e.g., summarisation, question answering). Our method bypasses the need for manual coding by exploiting the presence of markers like after", which overtly signal a temporal relation. We first show that models trained on main and subordinate clauses connected with a temporal marker achieve good performance on a pseudo-disambiguation task simulating temporal inference (during testing the temporal marker is treated as unseen and the models must select the right marker from a set of possible candidates). Secondly, we assess whether the proposed approach holds promise for the semi-automatic creation of temporal annotations. Specifically, we use a model trained on noisy and approximate data (i.e., main and subordinate clauses) to predict intra-sentential relations present in TimeBank, a corpus annotated rich temporal information. Our experiments compare and contrast several probabilistic models differing in their feature space, linguistic assumptions and data requirements. We evaluate performance against gold standard corpora and also against human subjects.

This paper applies machine learning and the mathematics of chaos to the task of designing indoor rock-climbing routes. Chaotic variation has been used to great advantage on music and dance, but the challenges here are quite different, beginning with the representation. We present a formalized system for transcribing rock climbing problems, then describe a variation generator that is designed to support human route-setters in designing new and interesting climbing problems. This variation generator, termed Strange Beta, combines chaos and machine learning, using the former to introduce novelty and the latter to smooth transitions in a manner that is consistent with the style of the climbs This entails parsing the domain-specific natural language that rock climbers use to describe routes and movement and then learning the patterns in the results. We validated this approach with a pilot study in a small university rock climbing gym, followed by a large blinded study in a commercial climbing gym, in cooperation with experienced climbers and expert route setters. The results show that {\sc Strange Beta} can help a human setter produce routes that are at least as good as, and in some cases better than, those produced in the traditional manner.

Combinatory categorial grammar (CCG) is a grammar formalism used for natural language parsing. CCG assigns structured lexical categories to words and uses a small set of combinatory rules to combine these categories to parse a sentence. In this work we propose and implement a new approach to CCG parsing that relies on a prominent knowledge representation formalism, answer set programming (ASP) - a declarative programming paradigm. We formulate the task of CCG parsing as a planning problem and use an ASP computational tool to compute solutions that correspond to valid parses. Compared to other approaches, there is no need to implement a specific parsing algorithm using such a declarative method. Our approach aims at producing all semantically distinct parse trees for a given sentence. From this goal, normalization and efficiency issues arise, and we deal with them by combining and extending existing strategies. We have implemented a CCG parsing tool kit - AspCcgTk - that uses ASP as its main computational means. The C&C supertagger can be used as a preprocessor within AspCcgTk, which allows us to achieve wide-coverage natural language parsing.

At a computational level, language processing tasks are traditionally processed in a language-only space/context, isolated from perception and action. However, at a cognitive level, language processing has been shown experimentally to be embodied, i.e. to inform and be informed by perception and action. In this paper, we argue that embodied cognition dictates the development of a new generation of language processing tools that bridge the gap between the symbolic and the sensorimotor representation spaces. We describe that tasks and challenges such tools need to address and provide an overview of the first such suite of processing tools developed in the framework of the POETICON project.

Activity recognition is a widely researched area with applications in health care, security and other domains. With each recognition system considering its own set of activities and sensors, it is difficult to compare the performance of these different systems and more importantly it makes the task of selecting an appropriate set of technologies and tools for recognizing an activity challenging. In this work-in-progress paper we attempt to characterize activities in terms of a complexity measure. We define activity complexity along three dimensions – sensing, computation and performance and illustrate different parameters that parameterize these dimensions. We look at grammars for representing activities and use grammar complexity as a measurement for activity complexity. Then we describe how these measurements can help evaluate the complexity of activities of daily living that are commonly considered by various researchers.

In a separate study, we were interested in understanding people's Q&A habits on Twitter. Finding questions within Twitter turned out to be a difficult challenge, so we considered applying some traditional NLP approaches to the problem. On the one hand, Twitter is full of idiosyncrasies, which make processing it difficult. On the other it is very restricted in length and tends to employ simple syntactic constructions, which could help the performance of NLP processing. In order to find out the viability of NLP and Twitter, we built a pipeline of tools to work specifically with Twitter input for the task of finding questions in tweets. This work is still preliminary, but in this paper we discuss the techniques we used and the lessons we learned.