Conversational Neuro-Symbolic Commonsense Reasoning

Arabshahi, Forough; Lee, Jennifer; Gawarecki, Mikayla; Mazaitis, Kathryn; Azaria, Amos; Mitchell, Tom M.

doi:10.48550/arxiv.2006.10022

Cited by 6 publications

(6 citation statements)

References 27 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…They then utilize a neural module to learn continuous vectors for symbolic representations to deal with the uncertainty of data [75,76,20]. Specifically, for commonsense and multi-hop reasoning, the symbolic module can be designed to identify relevant knowledge and multi-hop reasoning chain, respectively, and then encode them into the neural module to match the answer [77,78,79,80]. However, how to generate high-quality programs under weak supervision and extract symbolic reasoning units in an unsupervised manner remains an elusive challenge.…”

Section: B Advanced Methods Of Complex Reasoningmentioning

confidence: 99%

From LSAT: The Progress and Challenges of Complex Reasoning

Wang¹,

Liu²,

Zhong³

et al. 2021

Preprint

View full text Add to dashboard Cite

Complex reasoning aims to draw a correct inference based on complex rules. As a hallmark of human intelligence, it involves a degree of explicit reading comprehension, interpretation of logical knowledge and complex rule application. In this paper, we take a step forward in complex reasoning by systematically studying the three challenging and domain-general tasks of the Law School Admission Test (LSAT), including analytical reasoning, logical reasoning and reading comprehension. We propose a hybrid reasoning system to integrate these three tasks and achieve impressive overall performance on the LSAT tests. The experimental results demonstrate that our system endows itself a certain complex reasoning ability, especially the fundamental reading comprehension and challenging logical reasoning capacities. Further analysis also shows the effectiveness of combining the pre-trained models with the task-specific reasoning module, and integrating symbolic knowledge into discrete interpretable reasoning steps in complex reasoning. We further shed a light on the potential future directions, like unsupervised symbolic knowledge extraction, model interpretability, few-shot learning and comprehensive benchmark for complex reasoning.

show abstract

Section: B Advanced Methods Of Complex Reasoningmentioning

confidence: 99%

From LSAT: The Progress and Challenges of Complex Reasoning

Wang¹,

Liu²,

Zhong³

et al. 2021

Preprint

View full text Add to dashboard Cite

show abstract

“…But these neural methods may not be easily generalized to our desired propositional logical schema without annotations and still perform an implicit inference. To circumvent these limitations and utilize the superior performance of neural models, we take inspiration from neuro-symbolic reasoning (Wang et al, 2018;Arabshahi et al, 2020) to integrate symbolic inference and neural representation. We design an explicit three-step logical reasoning paradigm and propose a logic-driven reasoning system to generically identify the logical structure and perform interpretable logical inference in a symbolic module while taking a pre-trained model as the backbone.…”

Section: Related Workmentioning

confidence: 99%

Logic-Driven Context Extension and Data Augmentation for Logical Reasoning of Text

Wang

Zhong

Tang

et al. 2022

Findings of the Association for Computational Linguistics: ACL 2022

View full text Add to dashboard Cite

Logical reasoning of text requires identifying critical logical structures in the text and performing inference over them. Existing methods for logical reasoning mainly focus on contextual semantics of text while struggling to explicitly model the logical inference process. In this paper, we not only put forward a logic-driven context extension framework but also propose a logic-driven data augmentation algorithm. The former follows a three-step reasoning paradigm, and each step is respectively to extract logical expressions as elementary reasoning units, symbolically infer the implicit expressions following equivalence laws and extend the context to validate the options. The latter augments literally similar but logically different instances and incorporates contrastive learning to better capture logical information, especially logical negative and conditional relationships. We conduct experiments on two benchmark datasets, ReClor and LogiQA. The results show that our method achieves state-of-the-art performance on both datasets, and even surpasses human performance on the ReClor dataset. 1

show abstract

“…Commonsense for Dialogue Generation: Commonsense knowledge resources (Speer et al, 2017b;Malaviya et al, 2020) have been used in dialogue response generation for tasks such as personagrounded dialogue (Majumder et al, 2020) and open-domain dialogue generation (Ghazvininejad et al, 2018;Hedayatnia et al, 2020;Zhou et al, 2021c). Zhou et al (2021a) created a dataset focusing on social commonsense inferences in dialogue and Arabshahi et al (2020) designed a theorem prover for if-then-because reasoning. A concurrent work (Zhou et al, 2021b) proposed to train a model to explicitly generate implicit knowledge and use this knowledge to generate a response.…”

Section: Related Workmentioning

confidence: 99%

Target-Guided Dialogue Response Generation Using Commonsense and Data Augmentation

Gupta¹,

Jhamtani²,

Bigham³

2022

Findings of the Association for Computational Linguistics: NAACL 2022

View full text Add to dashboard Cite

Target-guided response generation enables dialogue systems to smoothly transition a conversation from a dialogue context toward a target sentence. Such control is useful for designing dialogue systems that direct a conversation toward specific goals, such as creating nonobtrusive recommendations or introducing new topics in the conversation. In this paper, we introduce a new technique for target-guided response generation, which first finds a bridging path of commonsense knowledge concepts between the source and the target, and then uses the identified bridging path to generate transition responses. Additionally, we propose techniques to re-purpose existing dialogue datasets for target-guided generation. Experiments reveal that the proposed techniques outperform various baselines on this task. Finally, we observe that the existing automated metrics for this task correlate poorly with human judgement ratings. We propose a novel evaluation metric that we demonstrate is more reliable for target-guided response evaluation. Our work generally enables dialogue system designers to exercise more control over the conversations that their systems produce. 1 Target: my puppy is called georgie. GPT-2: My mom likes the water. Multigen: My pet is the gecko. CODA: My dog walks along the beach with sand. CODA-Path: sand is at location beach belongs to walk is desired by puppy Context: my favorite city is seattle. Target: i ride my bicycle everywhere. GPT2: Seattle is my favorite city to go to Multigen: So what do you do when you go to the seattle CODA: I bought my bicycle from a bike shop in seattle. CODA-Path: favorite city is the location which has bicycle shop is a dependency of ride bicycle Context: i am a server at a food place. Target: i eat greasy foods. GPT-2: I eat healthy foods at restaurants. Multigen: I hate my food. CODA: I am a server, but I don't want to eat too much. CODA-Path: server is a person not desires eat greasy food

show abstract

Conversational Neuro-Symbolic Commonsense Reasoning

Cited by 6 publications

References 27 publications

From LSAT: The Progress and Challenges of Complex Reasoning

From LSAT: The Progress and Challenges of Complex Reasoning

Logic-Driven Context Extension and Data Augmentation for Logical Reasoning of Text

Target-Guided Dialogue Response Generation Using Commonsense and Data Augmentation

Contact Info

Product

Resources

About