This paper addresses the pragmatic challenges that state-of-the-art question/answering systems face in trying to decompose complex information-seeking scenarios. We propose that question decomposition can be approached in one of two ways: either by approximating the domain-specific knowledge for a particular set of domains, or by identifying the decomposition strategies employed by human users. We also present preliminary results from experiments that confirm the viability of each of these approaches within an interactive Q/A context. 1

Information of interest to users is often distributed over a set of documents. Users can specify their request for information as a query/topic – a set of one or more sentences or questions. Producing a good summary of the relevant information relies on understanding the query and linking it with the associated set of documents. To “understand ” the query we expand it using encyclopedic knowledge in Wikipedia. The expanded query is linked with its associated documents through spreading activation in a graph that represents words and their grammatical connections in these documents. The topic expanded words and activated nodes in the graph are used to produce an extractive summary. The method proposed is tested on the DUC summarization data. The system implemented ranks high compared to the participating systems in the DUC competitions, confirming our hypothesis that encyclopedic knowledge is a useful addition to a summarization system. 1

We present a novel framework for answering complex questions that relies on question decomposition. Complex questions are decomposed by a procedure that operates on a Markov chain, by following a random walk on a bipartite graph of relations established between concepts related to the topic of a complex question and subquestions derived from topic-relevant passages that manifest these relations. Decomposed questions discovered during this random walk are then submitted to a state-of-the-art Question Answering (Q/A) system in order to retrieve a set of passages that can later be merged into a comprehensive answer by a Multi-Document Summarization (MDS) system. In our evaluations, we show that access to the decompositions generated using this method can significantly enhance the relevance and comprehensiveness of summarylength answers to complex questions.

In this paper, we describe how Language Computer Corporation’s GISTEX-TER question-directed summarization system combines multiple strategies for question decomposition and summary generation in order to produce summary-length answers to complex questions. In addition, we introduce a novel framework for question-directed summarization that uses a state-of-the-art textual entailment system (Hickl et al., 2006) in order to select a single responsive summary answer from amongst a number of candidate summaries. We show that by considering entailment relationships between sentences extracted for a summary, we can automatically create semantic “Pyramids ” that can be used to identify answer passages that are both relevant and responsive. 1

This paper describes FERRET, an interactive question-answering (Q/A) system designed to address the challenges of integrating automatic Q/A applications into real-world environments. FERRET utilizes a novel approach to Q/A – known as predictive questioning – which attempts to identify the questions (and answers) that users need by analyzing how a user interacts with a system while gathering information related to a particular scenario. 1

This paper describes how Language Computer Corporation’s Lite-GISTexter multi-document summarization system addresses the challenge of providing summary-length answers to the types of complex questions asked in DUC 2005. We show that techniques first used in the automatic answering of “relationship ” questions can be used in multi-document summarization in order to provide accurate and responsive summary answers to a wide range of complex questions. Although we have found that traditional multi-document summarization techniques do remain effective in producing summary answers to questions, we argue that the best results should be obtained by systems that focus summary answers by approximating the information need of complex questions. 1

This paper presents a topic-driven multidocument summarization approach that relies on linking documents to Wikipedia. Wikipedia provides structural support to retrieve relevant concepts from the documents to be summarized, and quantify the strength of the relations between them, thus expanding the topic. We identify concepts in the documents, and assign them scores that describe their relevance to the topic, their significance in general, and a machine-learned confidence that they should appear in the summary. Sentences are ranked according to the scores of the concepts within them and how much new information they provide. The best are extracted and compressed to form the summary. The system is trained and developed using the DUC 2005 and 2006 data. It was tested on the DUC 2007 data before deploying it on the update summarization task of TAC 2009. It performs 5th (compared to 30 peers) in DUC 2007, and 21st (of 52 peers) on the TAC 2009 update task. 1

The problem of using topic representations for multi-document summarization (MDS) has received considerable attention recently. In this paper, we describe five different topic representations and introduce a novel representation of topics based on topic themes. We present eight different methods of generating MDS and evaluate each of these methods on a large set of topics used in past DUC workshops. Our evaluation results show a significant improvement in the quality of summaries based on topic themes over MDS methods that use other alternative topic representations.