Hierarchical Conversation Awareness Technology Based on Temporal Reasoning

View at publisher Summarise with VIRAT
Authors: Wancai Zhang, Yang Wenqing, Zhan Nan, Bu Xishuan
Cited by: 0
References: 1

Summary

No summary yet. VIRAT can read this paper and write a plain-language summary of its aim, methods, key findings and limitations.

Summarise with VIRAT Free to try with your Science Catalogs account.

Abstract

To address the problems of existing conversation awareness technologies that rely on static text feature extraction, struggle to capture temporal correlations in conversations, have weak multi-link reasoning capabilities, and are easily affected by useless information, this paper proposes a hierarchical conversation awareness technology based on temporal reasoning. This technology comprises three core modules: First, a dialog-task-specific temporal graph construction algorithm extracts role, sentence, and temporal features and constructs a structured temporal graph to enhance the multi-dimensional correlations between sentences; second, a gated temporal graph convolution algorithm constructs three reasoning links based on temporal, semantic, and role reasoning, and combines gating units to filter redundant information and integrate features; third, a temporal gating mechanism adaptively removes useless information based on contextual semantic similarity and dependencies. Experiments validated this approach on the DailyDialog and Ubuntu Dialogue Corpus datasets showed that the F1 scores for semantic understanding tasks reached 89.2% and 87.6%, respectively, representing improvements of 6.3% and 5.8% over the baseline BERT+BiLSTM model. The accuracy of useless information filtering reached 91.5% and 89.8%, with false positive rates as low as 4.2% and 5.1%. This technology effectively improves the accuracy and efficiency of conversational awareness, providing reliable technical support for scenarios such as intelligent customer service and conversational bots.

Fields of Study & Subjects

Speech and dialogue systems Topic Modeling AI in Service Interactions

Publication details

Journal / source
IEEE International Conference on Power, Electronics and Green Energy (ICPEGE)
Pages
959–964
Publication date
28 January 2026
Language
EN