Topical-Chat: Towards Knowledge-Grounded Open-Domain Conversations

By Karthik Gopalakrishnan, Behnam Hedayatnia, Qinlang Chen, Anna Gottardi, Sanjeev Kwatra, Anu Venkatesh, Raefer Gabriel, Dilek Z. Hakkani-Tür. 2019

367 citations recorded by the source corpus.

Building socialbots that can have deep, engaging open-domain conversations with humans is one of the grand challenges of artificial intelligence (AI). To this end, bots need to be able to leverage world knowledge spanning several domains effectively when conversing with humans who have their own world knowledge. Existing knowledge-grounded conversation datasets are primarily stylized with explicit roles for conversation partners. These datasets also do not explore depth or breadth of topical coverage with transitions in conversations. We introduce Topical-Chat, a knowledge-grounded human-human conversation dataset where the underlying knowledge spans 8 broad topics and conversation partners don't have explicitly defined roles, to help further research in open-domain conversational AI. We also train several state-of-the-art encoder-decoder conversational models on Topical-Chat and perform automated and human evaluation for benchmarking.

Subfields: Natural Language Processing

How researchers see this paper

The researchers whose trusted networks rank it highest, strongest advocate first. Every researcher gets a different order.

In Natural Language Processing

Citations in the corpus

Retrieval Augmentation Reduces Hallucination in Conversation (2021, background): “Knowledge-grounded dialogue is increasingly becoming a more important topic, with several datasets proposed that attempt to model its occurrence (Dinan et al., 2019b; Ghazvininejad et al., 2018; Gopalakrishnan et al., 2019; Galetzka et al., 2020).”
Internet-Augmented Dialogue Generation (2021, methodology): “Some of these such as Topical-Chat (Gopalakrishnan et al., 2019) and CMU_Dog (Zhou et al., 2018) are constructed given a gold passage of knowledge, and the task analyzes whether the model can use this knowledge in dialogue.”
The Dialogue Dodecathlon: Open-Domain Knowledge and Image Grounded Conversational Agents (2019, background): “For example, there are additional knowledge tasks (Qin et al., 2019; Moghe et al., 2018; Gopalakrishnan et al., 2019) and image-based datasets (Das et al., 2017) one could use.”
Neural Generation Meets Real People: Towards Emotionally Engaging Mixed-Initiative Conversations (2020, methodology): “To achieve this, we finetuned a neural generative model on the TopicalChat dataset (Gopalakrishnan et al., 2019) to obtain a conversational paraphrasing model that adapts external text into a conversational style (Section 5.”
USR: An Unsupervised and Reference Free Evaluation Metric for Dialog Generation (2020, methodology): “It is one of the four metrics used by the creators of the Topical-Chat dataset (Gopalakrishnan et al., 2019), along with perplexity and unique uni-gram/bigram counts.”
Q^{2}: Evaluating Factual Consistency in Knowledge-Grounded Dialogues via Question Generation and Question Answering (2021, methodology): “This is the case in Topical-Chat, in which each dialogue is accompanied by Washington Post articles and fun-facts from Red-dit in addition to pages from Wikipedia; and in the DNLI dataset, which deals with persona and self-consistency of dialogue systems and does not contain any references to Wikipe”
Reason first, then respond: Modular Generation for Knowledge-infused Dialogue (2021, background): “Knowledge-grounded open-domain dialogue datasets (Dinan et al., 2019; Komeili et al., 2021; Zhou et al., 2018; Gopalakrishnan et al., 2019) foster the research and development of knowledge-aware generative dialogue models.”
ProsocialDialog: A Prosocial Backbone for Conversational Agents (2022, methodology): “D IALOG and existing large-scale dialogue datasets: DailyDialog (Li et al., 2017), Topical Chat (Gopalakrishnan et al., 2019), Holl-E (Moghe et al., 2018), PersonaChat (Zhang et al., 2018), Wizard of Wikipedia (Dinan et al., 2018), EmpatheticDia-logues (Rashkin et al., 2019), and BlendedSkillTalk…”
Think Before You Speak: Explicitly Generating Implicit Commonsense Knowledge for Response Generation (2021, background): “Knowledge-grounded RG (Ghazvininejad et al., 2018; Dinan et al., 2019; Gopalakrishnan et al., 2019) addresses this issue, however, most approaches require a knowledge base (KB) to retrieve knowledge for RG (Zhou et al.”
OTTers: One-turn Topic Transitions for Open-Domain Dialogue (2021, neutral): “In addition, we show that existing multi-topic dialogue datasets, such as PersonaChat (Zhang et al., 2018) and TopicalChat (Gopalakrishnan et al., 2019), cannot be easily adapted to this task, due to the different nature of the tasks they were designed for.”
Analyzing the Forgetting Problem in Pretrain-Finetuning of Open-domain Dialogue Response Models (2021, background): “It would be interesting, as future work, to check the forgetting problem for knowledge-grounded datasets such as Topical-chat (Gopalakrishnan et al., 2019).”
Linguistic calibration through metacognition: aligning dialogue agent responses with expected correctness (2020, background): “Dinan et al. (2019) and Gopalakrishnan et al. (2019) have proposed additional conditioning on a knowledge base to address this issue, but success is only partial, so we are far from being able to assume that even a knowledge-conditioned model reliably gives correct answers.”
Towards a Unified Multi-Dimensional Evaluator for Text Generation (2022, methodology): “Topical-Chat (Mehri and Eskenazi, 2020) is a benchmark for knowledge-based dialogue response generation task.”
MMDialog: A Large-scale Multi-turn Dialogue Dataset Towards Multi-modal Open-domain Conversation (2022, background): “Later, researchers begin to incorporate knowledge into conversations (Zhang et al., 2018a; Dinan et al., 2018; Zhou et al., 2018b; Gopalakrishnan et al., 2019; Qin et al., 2019; Moon et al., 2019; Rashkin et al., 2019; Tuan et al., 2019; Song et al., 2020).”
Increasing Faithfulness in Knowledge-Grounded Dialogue with Controllable Features (2021, background): “Knowledge-Grounded Dialogue There has been significant prior work in tasks for designing dialogue agents that are grounded by document knowledge (Dinan et al., 2019; Qin et al., 2019; Ghazvininejad et al., 2018; Tian et al., 2020; Gopalakrishnan et al., 2019; Moghe et al., 2018).”
Commonsense-Focused Dialogues for Response Generation: An Empirical Study (2021, background): “Therefore several datasets have been collected to support such research efforts such as knowledge-grounded dialogues (Ghazvininejad et al., 2018; Gopalakrishnan et al., 2019), Per-sonaChat (Zhang et al., 2018), and Empathet-icDialogues (Rashkin et al., 2019).”
Unsupervised Evaluation of Interactive Dialog with DialoGPT (2020, background): “For under-standable , the Somewhat option was not provided, similar to prior past work (Gopalakrishnan et al., 2019).”