基于指针网络的多标签多类意图联合提取和检测方法

A Pointer Network-based Approach for Joint Extraction and Detection of Multi-Label Multi-Class Intents

October 29, 2024

作者: Ankan Mullick, Sombit Bose, Abhilash Nandy, Gajula Sai Chaitanya, Pawan Goyal

cs.AI

摘要

在面向任务的对话系统中，意图检测对于解释用户查询并提供适当回应至关重要。现有研究主要解决了单一意图的简单查询，缺乏处理具有多个意图和提取不同意图范围的复杂查询的有效系统。此外，缺乏多语言、多意图数据集也是一个显著问题。本研究解决了三个关键任务：从查询中提取多个意图范围、检测多个意图以及开发多语言多标签意图数据集。我们引入了一个新颖的多标签多类别意图检测数据集（MLMCID-dataset），从现有基准数据集中精心筛选而来。我们还提出了一个基于指针网络的架构（MLMCID），以六元组的形式提取意图范围并检测多个意图，标签包括粗粒度和细粒度。全面的分析表明，相对于各种数据集上的基准方法，我们基于指针网络的系统在准确性和F1分数方面表现出优越性。

English

In task-oriented dialogue systems, intent detection is crucial for interpreting user queries and providing appropriate responses. Existing research primarily addresses simple queries with a single intent, lacking effective systems for handling complex queries with multiple intents and extracting different intent spans. Additionally, there is a notable absence of multilingual, multi-intent datasets. This study addresses three critical tasks: extracting multiple intent spans from queries, detecting multiple intents, and developing a multi-lingual multi-label intent dataset. We introduce a novel multi-label multi-class intent detection dataset (MLMCID-dataset) curated from existing benchmark datasets. We also propose a pointer network-based architecture (MLMCID) to extract intent spans and detect multiple intents with coarse and fine-grained labels in the form of sextuplets. Comprehensive analysis demonstrates the superiority of our pointer network-based system over baseline approaches in terms of accuracy and F1-score across various datasets.