MuiDial: Improving Dialogue Disentanglement with Intent-Based Mutual Learning

Authors: Ziyou Jiang, Lin Shi, Celia Chen, Fangwen Mu, Yumin Zhang, Qing Wang

IJCAI 2022 | Conference PDF | Archive PDF | Plain Text | LLM Run Details

Reproducibility Variable Result LLM Response
Research Type Experimental Experimental results and in-depth analysis on several benchmark datasets demonstrate the effectiveness and generalizability of our approach. To evaluate the effectiveness of our approach, we conduct an exploratory study on four cross-domain benchmark datasets: IM (Instant Messenger), Reddit, IRC (Ubuntu Internet Relay Chat), and Gitter. The results show that MUIDIAL outperforms SOTA baselines on all the datasets.
Researcher Affiliation Academia Ziyou Jiang1,4 , Lin Shi1,4 , Celia Chen5 , Fangwen Mu1,4 , Yumin Zhang1,4 and Qing Wang1,2,3,4 1Laboratory for Internet Software Technologies, Institute of Software Chinese Academy of Sciences 2State Key Laboratory of Computer Sciences, Institute of Software Chinese Academy of Sciences 3Science&Technology on Integrated Information System Laboratory, Institute of Software Chinese Academy of Sciences 4University of Chinese Academy of Sciences, Beijing, China 5Department of Computer Science, Occidental College, Los Angeles, CA, USA {ziyou2019, shilin, fangwen2020,yumin2020, wq}@iscas.ac.cn, qchen2@oxy.edu
Pseudocode No The paper describes algorithms and methods in prose and equations (e.g., TBT algorithm, mutual optimization process) but does not include a clearly labeled 'Pseudocode' or 'Algorithm' block.
Open Source Code No The paper does not contain any explicit statement about releasing the source code for MUIDIAL, nor does it provide a direct link to a code repository.
Open Datasets Yes We adopt four benchmark datasets for the dialogue disentanglement task. As shown in Table 3, IM and Reddit contain social chats collected from Microsoft Messenger and Reddit forum respectively, whereas IRC [Kummerfeld et al., 2019] and Gitter focus on technical chats collected from Ubuntu IRC chat log and Gitter chat rooms from eight open-source communities respectively.
Dataset Splits Yes The train/valid/test datasets are partitioned by the benchmark datasets. For MUIDIAL: 2,120/190/2,010 for IM, 1,100/125/210 for Reddit, 1,980/134/100 for IRC, and 2,100/300/137 for Gitter.
Hardware Specification Yes The experiment environment is a Windows 10 desktop computer with NVIDIA Ge Force RTX 2060 GPU, Intel Core i7 CPU, and 32GB RAM.
Software Dependencies Yes We first embed utterances with text, heuristic, and user-intent encoders; then we learn its context into the context-aware utterance embedding to obtain a rich representation. We embed it into ui by using pre-trained BERT [Devlin et al., 2019] model. We use a Bi LSTM [Hochreiter and Schmidhuber, 1997] model to encode the utterance information into a deep contextual representation. Adam optimizer [Kingma and Ba, 2015] is used to optimize the parameters.
Experiment Setup Yes The utterance embedding vector and the heuristic feature vector are 768-dimensional. The LSTM hidden size is 256. When training MUIDIAL, the mini-batch size is set to 16. Adam optimizer [Kingma and Ba, 2015] is used to optimize the parameters with the initial learning rate of 5e-4. The final set of the hyperparameters is shown as follows. We follow the existing works to tune η until MUIDIAL achieves the optimal performance, and choose the cutoff-value as η = 0.6.