Information distillation aims to extract the most useful pieces of information related to a given query from massive, possibly multilingual, audio and textual document sources. One critical componentin a distillation engine is detecting sentences to be extracted from each relevant document. In this paper, we present a statistical sentence extraction approach for distillation. Basically, we frame this task as a classi�cation problem, where each candidate sentence in documents is classi�ed as relevant to the query or not. These documents may be in textual or audio format and in a number of languages. For audio documents, we use both manual and automatic transcriptions, for non-English documents, we use automatic translations. In this work, we use AdaBoost, a discriminative classi�cation method with both lexical and semantic features. The results indicate 11%-13 % relative improvement over a baseline keyword-spotting-based approach. We also show the robustness of our method on the audio subset of the document sources using manual and automatic transcriptions. Index Terms — information distillation, information extraction, language understanding, speech processing, natural language processing

We propose a novel method of extractive summarization of lecture speech based on unsupervised learning of its rhetorical structure. We present empirical evidence showing that rhetorical structure is the underlying semantics which is then rendered in linguistic and acoustic/prosodic forms in lecture speech. We present a first thorough investigation of the relative contribution of linguistic versus acoustic features and show that, at least for lecture speech, what is said is more important than how it is said. We base our experiments on conference speeches and corresponding presentation slides as the latter is a faithful description of the rhetorical structure of the former. We find that discourse features from broadcast news are not applicable to lecture speech. By using rhetorical structure information in our summarizer, its performance reaches 67.87 % ROUGE-L F-measure at 30 % compression, surpassing all previously reported results. The performance is also superior to the 66.47 % ROUGE-L F-measure of baseline summarization performance without rhetorical information. We also show that, despite a 29.7 % character error rate in speech recognition, extractive summarization performs relatively well, underlining the fact that spontaneity in lecture speech does not affect the central meaning of lecture speech.

We present the first known empirical study on speech summarization without lexical features for Mandarin broadcast news. We evaluate acoustic, lexical and structural features as predictors of summary sentences. We find that the summarizer yields good performance at the average F-measure of 0.5646 even by using the combination of acoustic and structural features alone, which are independent of lexical features. In addition, we show that structural features are superior to lexical features and our summarizer performs surprisingly well at the average F-measure of 0.3914 by using only acoustic features. These findings enable us to summarize speech without placing a stringent demand on speech recognition accuracy. 1

Feature-based approaches are widely used in the task of extractive meeting summarization. In this paper, we analyze and evaluate the effectiveness of different types of features using Forward Feature Selection in an SVM classifier. In addition to features used in prior studies, we introduce topic related features and demonstrate that these features are helpful for meeting summarization. We also propose a new way to resample the sentences based on their salience scores for model training and testing. The experimental results on both the human transcripts and recognition output, evaluated by the ROUGE summarization metrics, show that feature selection and data resampling help improve the system performance. Index Terms — meeting summarization, forward feature selection, resampling, TFIDF 1.

The CALO Meeting Assistant provides for distributed meeting capture, annotation, automatic transcription and semantic analysis of multi-party meetings, and is part of the larger CALO personal assistant system. This paper summarizes the CALO-MA architecture and its speech recognition and understanding components, which include realtime and offline speech transcription, dialog act segmentation and tagging, question-answer pair identification, action item recognition, and summarization. 1.

Abstract—Speech contains additional information than text that can be valuable for automatic speech summarization. In this paper, we evaluate how to effectively use acoustic/prosodic features for extractive meeting summarization, and how to integrate prosodic features with lexical and structural information for further improvement. To properly represent prosodic features, we propose different normalization methods based on speaker, topic, or local context information. Our experimental results show that using only the prosodic features we achieve better performance than using the non-prosodic information on both the human transcripts and recognition output. In addition, a decision-level combination of the prosodic and non-prosodic features yields further gain, outperforming the individual models. I.

Prosody, or intonation, is a critically important component of spoken communication. The automatic extraction of prosodic information is necessary for machines to process speech with human levels of proficiency. In this thesis we describe work on the automatic detection and classification of prosodic events – specifically, pitch accents and prosodic phrase boundaries. We present novel techniques, feature representations and state of the art performance in each of these tasks. We also present three proof-of-concept applications – speech summarization, story segmentation and non-native speech assessment – showing that access to hypothesized prosodic event information can be used to improve the performance of downstream spoken language processing tasks. We believe the contributions of this thesis advance the understanding of prosodic events and the use of prosody in spoken language processing towards the goal of human-like processing of speech by machines.

Advances in multimedia technologies have enabled the creation of huge archives of audio-video recordings of meetings, and there is burgeoning interest in developing meeting browsers to help users better leverage these archives. A recent study has shown that extractive summaries provide a more efficient way of navigating meeting content than simply reading through the transcript and using the audio-video record, or navigating via keyword search [15]. The extractive summary technique identifies informative dialogue acts to generate general purpose summaries. These summaries can still be lengthy. Recently, we have developed a decisionfocused summarization system that presents only 1-2 % of the recordings related to decision making. In this paper, we describe a task-based evaluation in which we compare the decision-focused summaries to the general purpose summaries. Our results indicate that the more focused summaries help users perform the decision debriefing task more effectively and improve perceived efficiency. In addition, this study also investigates the effect of automatic summaries and transcription on task effectiveness, report quality, and users ’ perceptions of task success. Author Keywords Meeting browser, automatic summarization, multimedia information

This paper presents a pilot study of opinion summarization on conversations. We create a corpus containing extractive and abstractive summaries of speaker’s opinion towards a given topic using 88 telephone conversations. We adopt two methods to perform extractive summarization. The first one is a sentence-ranking method that linearly combines scores measured from different aspects including topic relevance, subjectivity, and sentence importance. The second one is a graph-based method, which incorporates topic and sentiment information, as well as additional information about sentence-to-sentence relations extracted based on dialogue structure. Our evaluation results show that both methods significantly outperform the baseline approach that extracts the longest utterances. In particular, we find that incorporating dialogue structure in the graph-based method contributes to the improved system performance. 1

User-contributed content is creating a surge on the Internet. A list of “buzzing topics ” can effectively monitor the surge and lead people to their topics of interest. Yet a topic phrase alone, such as “SXSW”, can rarely present the information clearly. In this paper, we propose to explore a variety of text sources for summarizing the Twitter topics, including the tweets, normalized tweets via a dedicated tweet normalization system, web contents linked from the tweets, as well as integration of different text sources. We employ the concept-based optimization framework for topic summarization, and conduct both automatic and human evaluation regarding the summary quality. Performance differences are observed for different input sources and types of topics. We also provide a comprehensive analysis regarding the task challenges. 1