Diversifying Neural Conversation Model with Maximal Marginal Relevance
نویسندگان
چکیده
Neural conversation systems, typically using sequence-to-sequence (seq2seq) models, are showing promising progress recently. However, traditional seq2seq suffer from a severe weakness: during beam search decoding, they tend to rank universal replies at the top of the candidate list, resulting in the lack of diversity among candidate replies. Maximum Marginal Relevance (MMR) is a ranking algorithm that has been widely used for subset selection. In this paper, we propose the MMR-BS decoding method, which incorporates MMR into the beam search (BS) process of seq2seq . The MMR-BS method improves the diversity of generated replies without sacrificing their high relevance with the userissued query. Experiments show that our proposed model achieves the best performance among other comparison methods.
منابع مشابه
ery-adaptive Video Summarization viaality-aware Relevance Estimation
Although the problem of automatic video summarization has recently received a lot of aention, the problem of creating a video summary that also highlights elements relevant to a search query has been less studied. We address this problem by posing queryrelevant summarization as a video frame subset selection problem, which lets us optimise for summaries which are simultaneously diverse, repres...
متن کاملEmulating Human Conversations using Convolutional Neural Network-based IR
Conversational agents (“bots”) are beginning to be widely used in conversational interfaces. To design a system that is capable of emulating human-like interactions, a conversational layer that can serve as a fabric for chat-like interaction with the agent is needed. In this paper, we introduce a model that employs Information Retrieval by utilizing convolutional deep structured semantic neural...
متن کاملRanking Responses Oriented to Conversational Relevance in Chat-bots
For automatic chatting systems, it is indeed a great challenge to reply the given query considering the conversation history, rather than based on the query only. This paper proposes a deep neural network to address the context-aware response ranking problem by end-to-end learning, so as to help to select conversationally relevant candidate. By combining the multi-column convolutional layer and...
متن کاملConcept Based Tie-breaking and Maximal Marginal Relevance Retrieval in Microblog Retrieval
There are enormous tweets posted on any given day, and the number keeps increasing. As a result, the needs of effectively retrieving tweets depending upon user’s information need, and summarizing tweets pertaining to a given topic have become increasingly important. In this paper, Wikipedia concepts [1] was introduced in tie-breaking to perform ad-hoc microblog retrieval. The Maximal Marginal R...
متن کاملQuery-Focused Multidocument Summarization Based on Hybrid Relevance Analysis and Surface Feature Salience
Query-focused multidocument summarization is to synthesize from a set of topic-related documents a brief, well-organized, fluent summary for the purpose of answering an information need that cannot be met by just stating a name, date, quantity, etc. In this paper, the task is essentially treated as a sentence retrieval task. We propose a hybrid relevance analysis to evaluate the relevance of a ...
متن کامل