Abstract | ||
---|---|---|
Multi-turn conversation understanding is a major challenge for building intelligent dialogue systems. This work focuses on retrieval-based response matching for multi-turn conversation whose related work simply concatenates the conversation utterances, ignoring the interactions among previous utterances for context modeling. In this paper, we formulate previous utterances into context using a proposed deep utterance aggregation model to form a fine-grained context representation. In detail, a self-matching attention is first introduced to route the vital information in each utterance. Then the model matches a response with each refined utterance and the final matching score is obtained after attentive turns aggregation. Experimental results show our model outperforms the state-of-the-art methods on three multi-turn conversation benchmarks, including a newly introduced e-commerce dialogue corpus. |
Year | Venue | DocType |
---|---|---|
2018 | COLING | Conference |
Volume | ISSN | Citations |
abs/1806.09102 | COLING 2018, pages 3740-3752 | 9 |
PageRank | References | Authors |
0.45 | 30 | 5 |
Name | Order | Citations | PageRank |
---|---|---|---|
Zhuosheng Zhang | 1 | 57 | 14.93 |
Jiangtong Li | 2 | 19 | 4.31 |
Pengfei Zhu | 3 | 249 | 31.05 |
Hai Zhao | 4 | 960 | 113.64 |
Gongshen Liu | 5 | 42 | 26.26 |