Multiple topic identification in human/human conversations

Abstract

This paper deals with the automatic analysis of conversations between a customer and an agent in a call centre of a customer care service. The purpose of the analysis is to hypothesize themes about problems and complaints discussed in the conversation. Themes are defined by the application documentation topics. A conversation may contain mentions that are irrelevant for the application purpose and multiple themes whose mentions may be interleaved portions of a conversation that cannot be well defined. Two methods are proposed for multiple theme hypothesization. One of them is based on a cosine similarity measure using a bag of features extracted from the entire conversation. The other method introduces the concept of thematic density distributed around specific word positions in a conversation. In addition to automatically selected words, word bigrams with possible gaps between successive words are also considered and selected. Experimental results show that the results obtained with the proposed methods outperform the results obtained with support vector machines on the same data. Furthermore, using the theme skeleton of a conversation from which thematic densities are derived, it will be possible to extract components of an automatic conversation report to be used for improving the service performance.

A Critical Review for Trustworthy and Explainable Structural Health Monitoring and Risk Prognosis of Bridges with Human-In-The-Loop

“…However, the authors found that relatively fewer studies comprehend humans' sensing mechanisms and relevant reliability issues [86]. Some studies designed automated audio processing methods for augmenting human perception so that workers could better hear task-relevant information or commands in complex and changing environments with noisy backgrounds [87,88]. Few studies were on human individuals' sensing reliability of smelling and tasting [89,90].…”

Semi-Supervised Training of Transformer and Causal Dilated Convolution Network with Applications to Speech Topic Classification

“…In the field of the sound event classification, most of the initial researchers used the energy on the Mel band or the MFCC as the input value [18], and then combined with the Gaussian mixture model (GMM) or a support vector machine (SVM). Further, more speech feature filter banks or time-frequency descriptors have been developed to classify speech events in combination with the CNN, RNN, and convolutional recurrent neural networks (CRNNs) and have demonstrated advanced performance in DCASE challenges [19]. However, these models discard the temporal order of frame-level features in their construction, leading to unsatisfactory final results.…”

Topic Spotting using Hierarchical Networks with Self Attention

“…Topic spotting has been shown to be important in commercial dialog systems (Bost et al, 2013;Jokinen et al, 2002) directly dealing with the customers. Topical information is useful for speech recognition systems (Iyer and Ostendorf, 1999) as well as in audio document retrieval systems (Hazen et al, 2007;Hazen, 2011).…”