利用對話回合間之內文關係來降低辨識錯誤率

Reducing Recognition Error Rate Based on Context Relation among Dialogue Turns

吳旭智(Hsu-Chih Wu) ； 李青憲(Ching-Hsien Lee)

《電腦與通訊》 120期 (2007/06) Pp. 65-70

https://doi.org/10.29917/CCLTJ.200706.0011

摘要

本論文提出一種利用對話回合間之內文關係，來降低對話系統辨識錯誤的方法。此方法利用分類元系統(Learning Classifier System)來分析現有的對話內容，找出多條描述對話內文關係的規則。其規則所描述的資訊是以對話回合(dialogue turns)為單位，且該規則可描述多個回合之間的關係。經過分類元系統訓練過後的規則組，可用來根據對話的歷史紀錄，決定目前對話中，每一種內文關係出現的機率。以此機率可語音辨識所產生的N-最佳清單(N-Best List)進行重算分數，藉此來降低辨識錯誤。

關鍵字

對話模型 ； 使用者模擬 ； 可靠度評估 ； 機器學習

並列摘要

Recently research on developing spoken dialogue systems to provide conversational practice for a learner of a foreign language has been conducted. One of the most critical aspects of such a system is speech recognition errors, since they often take the dialogue thread down a wrong turn that is very confusing to the student and may be irrecoverable. In this paper a Learning Classifier System technique is presented to assist the process of selection from a list of N-best candidates based on a high-level description of the semantics of the preceding dialogue. A significant reduction in sentence error rate can be achieved from 29.2% to 23.6%.

並列關鍵字

Dialogue Modeling ； User Simulation ； Confidence Measure ； Machine Learning