IVR Customer Service Intent Recognition

IVR (Interactive Voice Response) customer service intent recognition refers to a technology application that combines AI artificial intelligence voice processing technology, allowing users to interact with customer service systems through voice commands, and automatically recognizes customer semantics to provide corresponding services. In Taiwan, such systems have been widely applied in banking, telecommunications, healthcare and other fields, enhancing service efficiency and customer experience.

Core Functions and Workflow

1. Receiving Voice Input After a customer calls, the system plays an automated voice menu (for example: "Please briefly describe your needs, and we will assist you."). Customers do not need to press keys to select, but directly express their needs through voice.

2. Automatic Speech Recognition (ASR) The system uses Automatic Speech Recognition technology to convert voice into text. For example, when a customer says "I want to check my bill", the system converts this content into text input.

3. Natural Language Understanding (NLU) The system uses Natural Language Understanding technology to analyze semantics and determine the user's actual needs. For example:

   • Lexical analysis: Keywords like "check" and "bill" indicate the need is related to billing.

   • Intent identification: Determines that the customer's purpose is "check bill".

4. Response and Routing The system provides corresponding services based on semantic analysis results, with the following possible options:

   • Direct response: If the need can be handled automatically, for example "Your bill amount is 1200 NTD, payment deadline is December 15".

   • Transfer to agent: If the need is more complex, the system will automatically transfer to customer service personnel in the corresponding department and simultaneously provide a semantic summary to reduce repetitive communication.

Technical Challenges and Limitations

1. Difficulty Understanding Diverse Expressions

   • User language expressions may not be standard, for example: "What's going on with my bill?" or "How to handle payment issues?" Such semantics are somewhat ambiguous, and models like BERT sometimes cannot accurately determine specific needs.

   • In Taiwan, there is also the challenge of multilingual and multi-dialect (Mandarin, Taiwanese, Hakka) mixing, and language models still lack sufficient support for these language data.

2. Limitations of Intent Subdivision

   • Although modern NLP models can process large amounts of text data, they cannot fully grasp professional knowledge or special intents in certain industries. For example: "I want to know the specific date of my last payment" may require linking different systems to answer correctly.

   • Even though BERT performs well in handling shorter conversation fragments, long sentences or complex semantic expressions can cause model confusion.

3. Data Bias and Incomplete Corpus

   • Training language models requires massive localized corpus. If data is insufficient or biased toward a single expression form, it will lead to insufficient model adaptability to special contexts. For example, Taiwan-specific language habits such as "top up" and "number portability" may lack sufficient contextual corpus in the model.

4. Context and Memory Limitations

   • Customer conversations usually have contextual relevance, for example "Regarding the payment I just mentioned, I have other questions" - this kind of multi-turn conversation requires the system to remember previous intents. Existing NLP models have limited performance in this application.

   • If intent determination is incorrect, users may have to restate their needs, causing frustration.

5. Low Error Tolerance

   • Customers have limited patience with customer service systems. If the voice system makes incorrect judgments, customers may feel frustrated and ultimately request to speak with a real person directly.

Industry Status

Currently, many enterprises' voice customer service systems still use traditional key-selection processes. Taking SinoPac Bank's voice customer service system as an example, its design fully considers business diversity, providing multi-level menu options to guide users. However, there is still much room for optimization and improvement in the user interaction experience.

The emergence of LLM (Large Language Models) and RAG (Retrieval-Augmented Generation) has brought revolutionary changes to semantic recognition and overall IVR systems, making voice customer service systems more intelligent, accurate, and adaptive, overcoming many limitations of traditional NLP technologies.

永豐銀行 Bank SinoPac - 客服中心 – 連絡客服 – 語音服務bank.sinopac.com

Solution

The following will guide you step by step through MaiAgent's powerful and accurate LLM and RAG features to create a superior semantic recognition assistant.

Operation Steps

1. Data Preparation

In the past, BERT required large amounts of labeled data for NLP tasks (such as intent recognition, sentiment analysis, etc.) for training. Labeled data is usually completed manually, such as annotating intent categories or keywords for sentences, which is both time-consuming and expensive. Even with high-quality labeled data, the model's generalization ability remains insufficient. When business requirements or usage habits change, data needs to be relabeled and models retrained, which takes a long time.

The advantage of LLM and RAG is that they can fully combine generative language capabilities with the dynamics of real-time retrieval, breaking free from dependence on labeled data, improving semantic recognition accuracy, reducing development and maintenance costs, and greatly improving user experience. This technology combination sets a new industry standard for intelligent customer service and voice interaction, and is the key driver of future automation and personalized services.

The introduction of LLM and RAG has greatly simplified the process of preparing labeled data. Now, you only need to organize data into an Excel spreadsheet, simply list intent classifications, and upload this Bank Customer Service List to the MaiAgent AI assistant's knowledge base to support the operation of an intelligent semantic recognition system.

Bank Service List (for intent recognition)

2. Define Role Instructions

# Role
You are the semantic understanding robot for MaiAgent Bank

# Output Format
Based on user conversations, understand the customer's desired service intent, determine the user's desired service intent from the knowledge base and output it

When the intent is very clear, please output only one intent; if there are multiple similar intents that cannot be determined, please list up to 3 most similar intents

<example>
-<code>:<category> - <subcategory>
</example>

<example>
-<code>:<category> - <subcategory>
-<code>:<category> - <subcategory>
...
</example>

<example>
N/A
</example>

# Output Restrictions
- Please respond in Traditional Chinese
- Do not answer information outside the scope of the knowledge base
- Please directly output the text inside <example> and </example>, do not include other descriptions
- Output does not include <example> and </example>
- Answer based on knowledge base data, when intent cannot be determined, please respond with the text inside <example> below

3. Start Using

Usage Examples

Single Conversation

Intent Recognition Case

Multi-turn Conversation

LLM and RAG technology solve the bottleneck of intent recognition in multi-turn conversations that was difficult for BERT.

Intent Recognition Multi-turn Conversation Example