Learning Symmetric Collaborative Dialogue Agents with Dynamic Knowledge Graph Embeddings
By He He, Anusha Balakrishnan, Mihail Eric, Percy Liang. 2017
210 citations recorded by the source corpus.
We study a symmetric collaborative dialogue setting in which two agents, each with private knowledge, must strategically communicate to achieve a common goal. The open-ended dialogue state in this setting poses new challenges for existing dialogue systems. We collected a dataset of 11K human-human dialogues, which exhibits interesting lexical, semantic, and strategic elements. To model both structured knowledge and unstructured language, we propose a neural model with dynamic knowledge graph embeddings that evolve as the dialogue progresses. Automatic and human evaluations show that our model is both more effective at achieving the goal and more human-like than baseline neural and rule-based models.
Subfields: Artificial Intelligence, Natural Language Processing, LLM Agents & Tool Use
How researchers see this paper
The researchers whose trusted networks rank it highest, strongest advocate first. Every researcher gets a different order.
In Artificial Intelligence
- Mohit Bansal
- Jianfeng Gao
- Percy Liang
- Luke S. Zettlemoyer
- Yejin Choi
- Yoshua Bengio
- Graham Neubig
- Noah A. Smith
In Natural Language Processing
- Mohit Bansal
- Jianfeng Gao
- Yejin Choi
- Luke S. Zettlemoyer
- Jason Aaron Edward Weston
- Graham Neubig
- Percy Liang
- Noah A. Smith
In LLM Agents & Tool Use
- Mohit Bansal
- Tong Niu
- Michael Lewis
- Denis Yarats
- Jason Aaron Edward Weston
- Jianfeng Gao
- Mitesh M. Khapra
- Percy Liang
Citations in the corpus
- QuAC: Question Answering in Context (2018, background): “and item guessing (He et al., 2017) have also been explored, but the language is more constrained than in .”
- Towards Exploiting Background Knowledge for Building Conversation Systems (2018, methodology): “works such as (He et al., 2017; Lewis et al., 2017; Krause et al., 2017) which have used crowdsourcing for creating conversation datasets. In fact, our data collection method is inspired by the work of Krause et al. (2017) where the authors use self-”
- A Knowledge-Grounded Neural Conversation Model (2017, background): “Similarly, (He et al. 2017) employ a knowledge graph to embed side information into dialog systems.”
- Few-Shot NLG with Pre-Trained Language Model (2019, background): “Some examples are task-oriented dialog, question answering (He et al. 2017; Su et al. 2016; Saha et al. 2018; Yin et al. 2016) and interdisciplinary applications such as medicine (Hasan and Farri 2019; Cawsey, Webber, and Jones 1997) and health-care (Hasan and Farri 2019; DiMarco et al. 2007),…”
- Learning Rewards from Linguistic Feedback (2020, background): “…(Djalali et al. 2011; Djalali, Lauer, and Potts 2012; Potts 2012; Ilinykh, Zarrieß, and Schlangen 2019; Suhr et al. 2019), reference pragmatics (He et al. 2017; Udagawa and Aizawa 2019), navigation (Thomason et al. 2019), and “Wizard of Oz” experiments (Kim et al. 2009; Allison, Luger, and…”
- OpenDialKG: Explainable Conversational Reasoning with Attention-based Walks over Knowledge Graphs (2019, background): “…(2018) propose to explicitly augment input text with concepts expanded via 1-hop relations (where KG triples are represented in the sentence embeddings space), and He et al. (2017) propose a system which iteratively updates KG embed-dings and attends over connected entities for response generation.”
- CaSiNo: A Corpus of Campsite Negotiation Dialogues for Automatic Negotiation Systems (2021, unknown): “Several other related efforts have explored problems between task-oriented and open-domain scenarios, such as persuasion for a charity (Wang et al., 2019), anti-scam (Li et al., 2020), collecting cards in a maze (Potts, 2012), and searching for a mutual friend (He et al., 2017).”
- Hierarchical Text Generation and Planning for Strategic Dialogue (2017, methodology): “Future work should apply our model to other dialogue settings, such as cooperative strategic dialogue games (He et al., 2017), or multi-sentence generation tasks, such as long document language modelling (Merity et al., 2016).”
- Do Neural Dialog Systems Use the Conversation History Effectively? An Empirical Study (2019, methodology): “Recurrent Models We trained a seq2seq (seq2seq lstm) model where the encoder and decoder are parameterized as LSTMs (Hochreiter and Schmidhuber, 1997).”
- Learning from History: Modeling Temporal Knowledge Graphs with Sequential Copy-Generation Networks (2020, background): “2018), natural language understanding (Chen, Chen, and Yu 2019; He et al. 2017), recommender systems (Koren, Bell, and Volinsky 2009) and healthcare (Hao et al.”
- Decoupling Strategy and Generation in Negotiation Dialogues (2018, methodology): “Recently, end-to-end neural models (Lewis et al., 2017; He et al., 2017) are used to simultaneously learn dialogue strategy and language realization from human-human di-alogues, following the trend of using neural network models on both goal-oriented dialogue (Wen et al., 2017a; Dhingra et al.,…”
- Modeling Multi-turn Conversation with Deep Utterance Aggregation (2018, background): “…et al., 2018d; Qin et al., 2017; Zhang et al., 2018b; Bai and Zhao, 2018), developing an intelligent dialogue system becomes realizable, which means training machines to converse with human in natural languages (Williams et al., 2017; He et al., 2017; Dhingra et al., 2017; Zhang et al., 2018a).”
- Learning Attention-based Embeddings for Relation Prediction in Knowledge Graphs (2019, background): “KGs find uses in a wide variety of applications such as semantic search (Be-rant et al., 2013; Berant and Liang, 2014), dialogue generation (He et al., 2017; Keizer et al., 2017), and question answering (Zhang et al., 2016; ∗ Equal Contribution Diefenbach et al., 2018), to name a few.”
- The Rapidly Changing Landscape of Conversational Agents (2018, background): “Example demonstrating how DynoNet augments the conversation [21]”
- Recent advances in deep learning based dialogue systems: a systematic survey (2021, background): “MedDialog(EN,CN) Large scale dataset in medical domain consisting of conversations between doctors and patients Task-oriented Dialogue He et al. (2020b) CamRest It consists of human-human multiturn dialogues in restaurant domain. Task-oriented Dialogue Wen et al. (2016a)”
- Multi-Hop Knowledge Graph Reasoning with Reward Shaping (2018, background): “Large-scale knowledge graphs (KGs) support a variety of downstream NLP applications such as semantic search (Berant et al., 2013) and dialogue generation (He et al., 2017).”
- Recommendation as a Communication Game: Self-Supervised Bot-Play for Goal-oriented Dialogue (2019, background): “Traditional recommendation systems factorize users’ historical data (i.e., ratings on movies) to extract common preference patterns (Koren et al., 2009; He et al., 2017b).”
- Adversarial Language Games for Advanced Natural Language Intelligence (2019, background): “…based on the descriptions from another agent (Lazaridou et al., 2017; Havrylov and Titov, 2017; Bouchacourt and Ba-roni, 2018; Kharitonov et al., 2019), or two agents with incomplete private information communicate to achieve a common goal (Vogel et al., 2013; He et al., 2017; Khani et al., 2018).”