Papers › Multi-hop Selector Network for Multi-turn Response Selection in Retrieval-based Chatbots

Multi-hop Selector Network for Multi-turn Response Selection in Retrieval-based Chatbots

1 Nov 2019IJCNLP 2019 11archive 2025-07-28

Chunyuan Yuan, Wei Zhou, Mingming Li, Shangwen Lv, Fuqing Zhu, Jizhong Han, Songlin Hu

Multi-turn retrieval-based conversation is an important task for building intelligent dialogue systems. Existing works mainly focus on matching candidate responses with every context utterance on multiple levels of granularity, which ignore the side effect of using excessive context information. Context utterances provide abundant information for extracting more matching features, but it also brings noise signals and unnecessary information. In this paper, we will analyze the side effect of using too many context utterances and propose a multi-hop selector network (MSN) to alleviate the problem. Specifically, MSN firstly utilizes a multi-hop selector to select the relevant utterances as context. Then, the model matches the filtered context with the candidate response and obtains a matching score. Experimental results show that MSN outperforms some state-of-the-art methods on three public multi-turn dialogue datasets.

PaperPDFCode

Code

Repository list and official/mentioned flags are the archive's, frozen 2025-07-28. Reachability, where shown, is from one Syntology probe window (2026-09-16 to 2026-09-18); repositories not probed show nothing. GitHub stars are not tracked.

Code Syntology ran Syntology

Not run by Syntology. Nothing on this page verifies that the listed code works.

Tasks

Conversational Response SelectionRetrieval

Results from the paper archive 2025-07-28

TaskDatasetModelMetricValueRank at snapshotLeaderboardReport
Conversational Response Selection Douban MSN MAP 0.587 #11 of 16 Archive leaderboard report
Conversational Response Selection Douban MSN MRR 0.632 #11 of 16 Archive leaderboard report
Conversational Response Selection Douban MSN P@1 0.470 #11 of 16 Archive leaderboard report
Conversational Response Selection Douban MSN R10@1 0.295 #11 of 16 Archive leaderboard report
Conversational Response Selection Douban MSN R10@2 0.452 #11 of 16 Archive leaderboard report
Conversational Response Selection Douban MSN R10@5 0.788 #11 of 16 Archive leaderboard report
Conversational Response Selection E-commerce MSN R10@1 0.606 #12 of 15 Archive leaderboard report
Conversational Response Selection E-commerce MSN R10@2 0.770 #12 of 15 Archive leaderboard report
Conversational Response Selection E-commerce MSN R10@5 0.937 #12 of 15 Archive leaderboard report
Conversational Response Selection RRS MSN MAP 0.550 #5 of 7 Archive leaderboard report
Conversational Response Selection RRS MSN MRR 0.563 #5 of 7 Archive leaderboard report
Conversational Response Selection RRS MSN P@1 0.383 #5 of 7 Archive leaderboard report
Conversational Response Selection RRS MSN R10@1 0.343 #5 of 7 Archive leaderboard report
Conversational Response Selection RRS MSN R10@2 0.498 #5 of 7 Archive leaderboard report
Conversational Response Selection RRS MSN R10@5 0.798 #5 of 7 Archive leaderboard report
Conversational Response Selection Ubuntu Dialogue (v1, Ranking) MSN R10@1 0.800 #13 of 25 Archive leaderboard report
Conversational Response Selection Ubuntu Dialogue (v1, Ranking) MSN R10@2 0.899 #13 of 25 Archive leaderboard report
Conversational Response Selection Ubuntu Dialogue (v1, Ranking) MSN R10@5 0.978 #13 of 25 Archive leaderboard report

Ranks are positions in the archive's leaderboards as they stood at the 2025-07-28 snapshot. Results published since then are not among these rows, so a rank here is not a current standing.

Report a problem or propose a change · a person checks every report against the paper or source before anything changes; decisions are listed on /corrections