Enhance word representation for out-of-vocabulary on Ubuntu dialogue corpus

By Jian-xiong Dong, Jim Huang. 2018

24 citations recorded by the source corpus.

Ubuntu dialogue corpus is the largest public available dialogue corpus to make it feasible to build end-to-end deep neural network models directly from the conversation data. One challenge of Ubuntu dialogue corpus is the large number of out-of-vocabulary words. In this paper we proposed an algorithm which combines the general pre-trained word embedding vectors with those generated on the task-specific training set to address this issue. We integrated character embedding into Chen et al's Enhanced LSTM method (ESIM) and used it to evaluate the effectiveness of our proposed method. For the task of next utterance selection, the proposed method has demonstrated a significant performance improvement against original ESIM and the new model has achieved state-of-the-art results on both Ubuntu dialogue corpus and Douban conversation corpus. In addition, we investigated the performance impact of end-of-utterance and end-of-turn token tags.

Subfields: Artificial Intelligence, Natural Language Processing, Dialogue & Conversational AI

How researchers see this paper

The researchers whose trusted networks rank it highest, strongest advocate first. Every researcher gets a different order.

In Artificial Intelligence

1. QUAN LIU
2. Zhenhua Ling
3. Dongsuk Oh
4. Kisu Yang
5. Chanhee Lee
6. Jia-Chen Gu
7. Heuiseok Lim
8. Dongyub Lee

In Natural Language Processing

1. QUAN LIU
2. Dongsuk Oh
3. Kisu Yang
4. Zhenhua Ling
5. Chanhee Lee
6. Jia-Chen Gu
7. Heuiseok Lim
8. Dongyub Lee

In Dialogue & Conversational AI

1. QUAN LIU
2. Zhenhua Ling
3. Jia-Chen Gu
4. Yu-Ping Ruan

Citations in the corpus

• Domain Adaptive Training BERT for Response Selection (2019, methodology): “The response selection task is similar to natural language inference problem; therefore, the most recent works are based on Enhanced Sequential Inference Model (ESIM) (Dong and Huang, 2018; Gu et al., 2019; Chen and Wang, 2019).”
• Building Sequential Inference Models for End-to-End Response Selection (2018, methodology): “To address this issue, we adopt an algorithm (Dong and Huang 2018) which combines the general pre-trained word embedding vectors with those generated on the task-specific training set to enhance word representations.”