Results 1 -
6 of
6
USING DIALOG-ACTIVITY SIMILARITY FOR SPOKEN INFORMATION RETRIEVAL
"... We want to enable users to locate desired information in spoken audio documents using not only the words, but also dialog activities. Following previous research, we infer this information from prosodic features, however, instead of retrieval by matching to a predefined finite set of activities, we ..."
Abstract
-
Cited by 4 (4 self)
- Add to MetaCart
(Show Context)
We want to enable users to locate desired information in spoken audio documents using not only the words, but also dialog activities. Following previous research, we infer this information from prosodic features, however, instead of retrieval by matching to a predefined finite set of activities, we estimate similarity using a vector space representation. Utterances close in this vector space are frequently similar not only pragmatically, but also topically. Using this we implemented a dialog-based query-by-example function and built it into an interface for use in combination with normal lexical search. Evaluating its utility by an experiment with four searchers doing twenty tasks each, we found that searchers used the new feature and considered it helpful, but only for some search tasks. 1. Two Views of Audio Search
Where in Dialog Space does Uh-huh Occur?
"... In what dialog situations and contexts do backchannels commonly occur? This paper examines this question using a newly developed notion of dialog space, defined by orthogonal, prosody-derived dimensions. Taking 3363 instances of uh-huh, found in the Switchboard corpus, we examine where in this space ..."
Abstract
-
Cited by 1 (1 self)
- Add to MetaCart
(Show Context)
In what dialog situations and contexts do backchannels commonly occur? This paper examines this question using a newly developed notion of dialog space, defined by orthogonal, prosody-derived dimensions. Taking 3363 instances of uh-huh, found in the Switchboard corpus, we examine where in this space they tend to occur. While the results largely agree with previous descriptions and observations, we find several novel aspects, relating to rhythm, polarity, and the details of the low-pitch cue. Index Terms: backchannels, feedback, prosody, context, principal component analysis, dimensions, dialog activities
Detecting Differences in Communication During Two Types of Patient Handovers: A Linguistic Construct Categorization Approach
"... Patient handovers are a critical point in the patient care process. Software to identify differences in communication content and strategies across different types of patient handovers could be helpful in customizing physician training programs. To determine whether there were differences, Linguisti ..."
Abstract
- Add to MetaCart
Patient handovers are a critical point in the patient care process. Software to identify differences in communication content and strategies across different types of patient handovers could be helpful in customizing physician training programs. To determine whether there were differences, Linguistic Inquiry and Word Count (LIWC) software was used. The primary measure was the LIWC output score, which is the frequency of mention of words in a construct category divided by the total number of words in the handover transcript. Two types of constructs were investigated: 1) content, which included name/age, care plan, prognosis, and family, and 2) strategy, which included questioning and collaborative cross-checks. We hypothesized that the Emergency Department (ED) to hospital transfer compared to Intensive Care Unit (ICU) sign-outs would have more discussion of family and less of the patient’s prognosis, as well as more collaborative cross-checks. A two-tailed t-test was used to detect differences. One hypothesis was confirmed, that there was less discussion of prognosis in the ED as compared to the ICU handover. Unexpected findings were less discussion of the care plan and more questioning in the ED as compared to the ICU handover. Findings confirm that both communication content and strategies are different for the two types of patient handovers and that an automated analysis approach can detect differences across a set of handover transcripts.
Sub-lexical Dialogue Act Classification in a Spoken Dialogue System Support for the Elderly with Cognitive Disabilities
"... This paper presents a dialogue act classification for a spoken dialogue system that delivers necessary information to elderly subjects with mild dementia. Lexical features have been shown to be effective for classification, but the automatic transcription of spontaneous speech demands expensive lang ..."
Abstract
- Add to MetaCart
(Show Context)
This paper presents a dialogue act classification for a spoken dialogue system that delivers necessary information to elderly subjects with mild dementia. Lexical features have been shown to be effective for classification, but the automatic transcription of spontaneous speech demands expensive language modeling. Therefore, this paper proposes a classifier that does not require language modeling and that uses sub-lexical features instead of lexical features. This classifier operates on sequences of phonemes obtained by a phoneme recognizer and exhaustively analyzes the saliency of all possible sub-sequences using a support vector machine with a string kernel. An empirical study of a dialogue corpus containing elderly speech showed that the sub-lexical classifier was robust against the poor modeling of language and it performed better than a lexical classifier that used hidden Markov models of words. Index Terms: dialogue acts, support vector machines, string kernels, spontaneous speech, elderly speech, dementia
APPROVED:
"... I would like to give thanks to all my family and friends. Without their support all these years, I would have never gotten to where I am right now. I want to thank my family for their sacrifices and patience. I want to thank my friends for their support and company during the good and tough times. I ..."
Abstract
- Add to MetaCart
(Show Context)
I would like to give thanks to all my family and friends. Without their support all these years, I would have never gotten to where I am right now. I want to thank my family for their sacrifices and patience. I want to thank my friends for their support and company during the good and tough times. I want to give my deepest thanks to Nigel Ward for all his help as an advisor and mentor. I thank David Novick for being a mentor and teacher and for all the helpful feedback all these years. I thank Jon Amastae for all the helpful comments and being my committee member. I would like to give deep thanks to Shreyas Karkhedkar for his help with the Respond features, for his help being someone I could always talk to when I had trouble, and his help as one of my best friends during this process. This work was supported in part by NSF Award IIS-0914868. Previous studies show that immediate and long range prosodic context provide beneficial information when applied to a language model.
