As part of this planning project, the PIs test the feasibility of collecting a corpus of conversational speech including both broadcast and telephone conversations. The corpus is annotated to support research in extractive and abstractive summarization of opinion and attitude in speech. The goal of the pilot annotation effort is the adaptation and refinement of current opinion and attitude annotation schemes for conversational data. The PIs are also organizing a workshop to be held at the 2011 meeting of the Association for Computational Linguistics during which they plan to solicit annotation desiderata and feedback from researchers who are the future users of the resource. The pilot annotations include abstractive and extractive summaries, rich mark-up with existing automatic tools for prosodic event detection, discourse relations, topic words and extractive summaries from current baselines.
It is increasingly important to track opinions and information on a wide spectrum of issues, and increasingly difficult to do so in the face of enormous amounts of information in textual and audio form. However, speech data is notoriously hard to search using current technologies, so developing new tools for this type of speech search is particularly important. The corpus the PIs are developing will make possible the development of automatic techniques to deal with this problem. Users will benefit from better tools to identify and summarize opinions that concern their daily choices related to health, diet, purchases, and the environment.
All successful applications for language analysis rely on a large corpus of annotated data in which people have performed the task which we wish to automate. For example for speech recognition the annotated data may consist of an audio recording of a person speaking and the human transcription of the speech; for machine translation the annotated data is a translation from a text from one language to another. Statistical methods applied to these annotations allow for rather accurate approximations of the task to be done by a machine. Our proposal was to create a novel large scale resource to support two tasks that have high potential impact for human-computer interaction and for information access but for which there is practically no existing annotated corpus: automatic summarization of speech and affect recognition in speech. The specific goal of the project was to solicit the opinion of a large number of experts and researchers in related fields, in order to clarify the exact annotations that would be most beneficial for future research. We also planned to more widely discuss open issues in these two fields, reaching out to researchers working on text summarization, sentiment and opinion analysis, and emotion recognition. To achieve our goals, we organized two workshops and gave four tutorials, two oriented toward students specifically and two aimed at a diverse audience at premier conferences on speech and natural language processing. We organized a workshop, "Automatic Summarization for Different Genres, Media, and Languages", collocated with the Annual Meeting of the Association for Computational Linguistics (ACL) in 2011. The workshop included a mix of peer reviewed poster presentations, two invited talks and a discussion panel on the need for corpus development for summarization in general and speech summarization specifically. In October 2012 we organized an invitation-only workshop attended by 25 prominent scholars working on emotion, affect and opinion analysis. They presented their work and discussed common issues and desiderata for large-scale annotations. In our efforts to bring more interdisciplinary awareness we presented tutorials on text and speech summarization at the 2011 Meeting of ACL, the 2011 INTERSPEECH conference; In addition we gave student-oriented tutorials on summarization at the Johns Hopkins University summer school on language technology and on affect processing in text and speech at the European Summer School on Logic, Language and Computation. Based on our detailed discussions with researchers across three mostly independent fields---summarization, affect detection from speech and face and opinion analysis---we have prepared a detailed annotation framework, and submitted an NSF CRI proposal for data creation. The annotation includes summarization of spoken interaction and affect. Numerous researchers from the three fields we reached out enthusiastically supported the proposal.
