DuanzAI: Slang-Enhanced LLM with Prompt for Humor Understanding

Yesian Rohn

DOI: https://doi.org/10.48550/arXiv.2405.15818

2024-05-23

Abstract:Language's complexity is evident in the rich tapestry of slang expressions, often laden with humor and cultural nuances. This linguistic phenomenon has become increasingly prevalent, especially in digital communication. However, existing AI models, including ChatGPT-3.5, face challenges in comprehending these nuances, particularly in Chinese slang. In this study, we present DuanzAI, an innovative approach enhancing Large Language Models (LLMs) with deep Chinese slang comprehension. Leveraging curated datasets and advanced techniques, DuanzAI bridges the gap between human expression and AI comprehension, enabling contextually relevant responses. Our experiments contrast LLMs' performance with a custom Punchline Entity Recognition (PER) system, integrating phonetic matching and pinyin2hanzi techniques. Applying these insights, we developed ChatDAI, an advanced chatbot and released our code at \url{<a class="link-external link-https" href="https://github.com/YesianRohn/DuanzAI" rel="external noopener nofollow">this https URL</a>}.

Computation and Language

What problem does this paper attempt to address?

The paper aims to address the current shortcomings of large language models in understanding humor, especially Chinese slang humor. Specifically, the paper proposes DuanzAI, an innovative approach to enhance large language models' understanding of Chinese slang, bridging the gap between human expression and AI comprehension. The research mainly focuses on the following points:

1. **Proposing the DuanzAI System**: By carefully curated datasets and advanced techniques, the system enhances large language models' ability to understand Chinese slang, enabling the model to recognize, interpret diverse slang expressions, and provide contextually relevant responses.

2. **Building the Punchline Entity Recognition (PER) System**: A system specifically designed to identify and extract punchline entities in Chinese slang expressions. By combining phonetic matching and pinyin-to-character conversion techniques, it achieves accurate retrieval of original vocabulary.

3. **Experimental Validation**: Through comparative experiments, the performance of different models (such as ChatGLM-6B and ChatGPT-3.5) in recognizing punchlines was evaluated. Additionally, two methods, Clue Provided and 5-Shot, were introduced to enhance the models' understanding of humor.

4. **Practical Application**: Based on the above research findings, the ChatDAI chatbot was developed, utilizing the Spark large model API for practical application, demonstrating the model's ability to understand and engage in Chinese slang humor conversations.

In summary, the core objective of the paper is to improve the performance of language models in handling Chinese slang and humorous expressions through technical means, thereby enhancing the authenticity and fun of human-computer interaction.