How may I help you?
This paper focuses on the task of automatically routing telephone calls based on a user's fluently spoken response to the open-ended prompt of ``How may I help you?. Section 2 of the paper describes an experimental spoken dialog system for call-routing plus subsequent automatic processing of information retrieval and form-filling functions. The dialog is based upon a feedback control model. Section 3 presents a database of \(10K\) fluently spoken transactions between customers and human agents. Section 4 exposes the Spoken Language Understanding (SLU) algorithms that are used for call classification. A central, confirmed supposition is the following: it is not necessary to recognize and understand every nuance of the speech, but only those fragments which are salient to the task. This strategy leads to a methodology where understanding is based upon recognition of such salient fragments and combinations thereof. The proposed SLU methodology comprises three main components: (1) To automatically acquire the salient grammar fragments from the data, modelling those parts of the language which are meaningful for the task plus their statistical associations to the machine actions. (2) To recognize these fragments in fluent speech, searching the output of a large vocabulary speech recognizer. (3) To use these multiple recognized fragments to classify the call-type of an utterance. These methods have been embedded within a spoken dialog system, with subsequent processing for information retrieval and form-filling. Section 5 reports on experimental results for call-classification from the \(1OK\) speech database, trained on \(8K\) utterances and tested on \(1K\). These experiments give evidence that understanding rate is significantly greater than recognition rate, and confirm the intuition that for a good SLU there is necessary to be recognized only those fragments that are salient to the considered task. For the entire collection see [Zbl 0927.00044].
- scientific article; zbMATH DE number 1753976 (Why is no real title available?)
- A hierarchical tag-graph search scheme with layered grammar rules for spontaneous speech understanding
- scientific article; zbMATH DE number 1928631 (Why is no real title available?)
- Speech act modeling in a spoken dialog system using a fuzzy fragment-class Markov model
- Robust numeric recognition in spoken language dialogue
- Integration of utterance verification with statistical language modeling and spoken language understanding
- Grouping utterances in spoken dialogues
This page was built for publication: How may I help you?
Report a bug (only for logged in users!)Click here to report a bug for this page (MaRDI item Q4942755)
