Do Response Selection Models Really Know What’s Next? Utterance Manipulation Strategies for Multi-turn Response Selection
نویسندگان
چکیده
In this paper, we study the task of selecting optimal response given a user and system utterance history in retrieval-based multi-turn dialog systems. Recently, pre-trained language models (e.g., BERT, RoBERTa, ELECTRA) showed significant improvements various natural processing tasks. This similar selection tasks can also be solved using such by formulating as dialog--response binary classification Although existing works approach successfully obtained state-of-the-art results, observe that trained manner tend to make predictions based on relatedness candidates, ignoring sequential nature suggests alone is insufficient for learning temporal dependencies between utterances. To end, propose manipulation strategies (UMS) address problem. Specifically, UMS consist several (i.e., insertion, deletion, search), which aid model towards maintaining coherence. Further, are self-supervised methods do not require additional annotation thus easily incorporated into approaches. Extensive evaluation across multiple languages shows highly effective teaching consistency, leads pushing with margins public benchmark datasets.
منابع مشابه
Metabolic models of selection response.
Consequences of directional selection on metabolic flux are explored in models for which variation in flux among individuals is generated by segregation of allelic variants at enzyme activity loci. The pattern of selection response is strongly affected by the presence of genetic dominance and epistasis, which are automatically generated in metabolic systems. The expected magnitudes of dominance...
متن کاملSequential Match Network: A New Architecture for Multi-turn Response Selection in Retrieval-based Chatbots
We study response selection for multiturn conversation in retrieval-based chatbots. Existing work either concatenates utterances in context or matches a response with a highly abstract context vector finally, which may lose relationships among utterances or important contextual information. We propose a sequential matching network (SMN) to address both problems. SMN first matches a response wit...
متن کاملSequential Matching Network: A New Architecture for Multi-turn Response Selection in Retrieval-Based Chatbots
We study response selection for multiturn conversation in retrieval-based chatbots. Existing work either concatenates utterances in context or matches a response with a highly abstract context vector finally, which may lose relationships among utterances or important contextual information. We propose a sequential matching network (SMN) to address both problems. SMN first matches a response wit...
متن کاملA Sequential Matching Framework for Multi-turn Response Selection in Retrieval-based Chatbots
We study the problem of response selection for multi-turn conversation in retrieval-based chatbots. The task requires matching a response candidate with a conversation context, whose challenges include how to recognize important parts of the context, and how to model the relationships among utterances in the context. Existing matching methods may lose important information in contexts as we can...
متن کاملذخیره در منابع من
با ذخیره ی این منبع در منابع من، دسترسی به آن را برای استفاده های بعدی آسان تر کنید
ژورنال
عنوان ژورنال: Proceedings of the ... AAAI Conference on Artificial Intelligence
سال: 2021
ISSN: ['2159-5399', '2374-3468']
DOI: https://doi.org/10.1609/aaai.v35i16.17653