In this paper, we focus on the problem of using sentence compression techniques to improve multi-document summarization. We propose an innovative sentence compression method by considering every node in the constituent parse tree and deciding its status -remove or retain. Integer liner programming with discriminative training is used to solve the problem. Under this model, we incorporate various constraints to improve the linguistic quality of the compressed sentences. Then we utilize a pipeline summarization framework where sentences are first compressed by our proposed compression model to obtain top-n candidates and then a sentence selection module is used to generate the final summary. Compared with state-ofthe-art algorithms, our model has similar ROUGE-2 scores but better linguistic quality on TAC data.

We propose to use user simulation for testing during the development of a sophisticated dialog system. While the limited behaviors of the state-of-the-art user simulation may not cover important aspects in the dialog system testing, our proposed approach extends the functionality of the simulation so that it can be used at least for the early stage testing before the system reaches stable performance for evaluation involving human users. The proposed approach includes a set of evaluation measures that can be computed automatically from the interaction logs between the user simulator and the dialog system. We first validate these measures on human user dialogs using user satisfaction scores. We also build a regression model to estimate the user satisfaction scores using these evaluation measures. Then, we apply the evaluation measures on a simulated dialog corpus trained from the real user corpus. We show that the user satisfaction scores estimated from the simulated corpus are not statistically different from the real users' satisfaction scores.

This paper describes a fast algorithm that selects features for conditional maximum entropy modeling. Berger et al. (1996) presents an incremental feature selection (IFS) algorithm, which computes the approximate gains for all candidate features at each selection stage, and is very time-consuming for any problems with large feature spaces. In this new algorithm, instead, we only compute the approximate gains for the top-ranked features based on the models obtained from previous stages. Experiments on WSJ data in Penn Treebank are conducted to show that the new algorithm greatly speeds up the feature selection process while maintaining the same quality of selected features. One variant of this new algorithm with look-ahead functionality is also tested to further confirm the good quality of the selected features. The new algorithm is easy to implement, and given a feature space of size F, it only uses O(F) more space than the original IFS algorithm.

This paper describes our effort on the task of edited region identification for parsing disfluent sentences in the Switchboard corpus. We focus our attention on exploring feature spaces and selecting good features and start with analyzing the distributions of the edited regions and their components in the targeted corpus. We explore new feature spaces of a partof-speech (POS) hierarchy and relaxed for rough copy in the experiments. These steps result in an improvement of 43.98% percent relative error reduction in F-score over an earlier best result in edited detection when punctuation is included in both training and testing data [Charniak and Johnson 2001], and 20.44% percent relative error reduction in F-score over the latest best result where punctuation is excluded from the training and testing data [Johnson and Charniak 2004].

No abstract