Abstract:Prediction over event sequences is critical for many real-world applications in Information Retrieval and Natural Language Processing. Future Event Generation (FEG) is a challenging task in event sequence prediction because it requires not only fluent text generation but also commonsense reasoning to maintain the logical coherence of the entire event story. In this paper, we propose a novel explainable FEG framework, Coep. It highlights and integrates two types of event knowledge, sequential knowledge of direct event-event relations and inferential knowledge that reflects the intermediate character psychology between events, such as intents, causes, reactions, which intrinsically pushes the story forward. To alleviate the knowledge forgetting issue, we design two modules, Im and Gm, for each type of knowledge, which are combined via prompt tuning. First, Im focuses on understanding inferential knowledge to generate commonsense explanations and provide a soft prompt vector for Gm. We also design a contrastive discriminator for better generalization ability. Second, Gm generates future events by modeling direct sequential knowledge with the guidance of Im. Automatic and human evaluation demonstrate that our approach can generate more coherent, specific, and logical future events.

What problem does this paper attempt to address?

### Problems the paper attempts to solve The paper aims to solve the logical coherence problem in the Future Event Generation (FEG) task. Specifically, the FEG task requires predicting the next event based on the given previous events. This requires not only the ability to generate fluent text, but also the ability of common - sense reasoning to maintain the logical coherence of the entire event story. However, although the existing FEG research can generate fluent sentences, it usually fails to capture the logical coherence between events, and the generated events are either too general or irrelevant to the previous events. ### Main contributions of the paper 1. **Proposed a new framework Coep**: - The Coep framework integrates two types of common - sense knowledge: direct event - sequence knowledge and inferential knowledge (such as the mental states, intentions, reactions, etc. of characters), which can promote the development of the story. - By introducing inferential knowledge, Coep can generate more coherent, specific and logically reasonable future events. 2. **Provided the interpretability of FEG**: - Based on the generated explanations, the paper analyzes the influence of different types of character mental states on the development of the story, providing inspiration for future research. 3. **Experimental verification**: - Extensive experiments were carried out on public data sets, and both automatic evaluation and manual evaluation proved the effectiveness of Coep. Further ablation studies showed that Coep has consistency, specificity and logicality in the generation process. ### Method overview - **Common - sense reasoning module (Im)**: - Based on the pre - trained BART model, by fine - tuning on the ATOMIC data set, Im can infer the common - sense explanations of the input events and describe the mental states or causal factors of the characters in the event script. - A contrastive discriminator was introduced to evaluate the consistency between the decoded common - sense explanations and the input events, thereby enhancing the generalization ability of the model. - **Future event generation module (Gm)**: - Another module based on the pre - trained BART model. By fine - tuning on the ConceptNet sequential knowledge graph, Gm can generate future events. - Use the inferential knowledge generated by Im as a soft prompt to guide Gm to generate future events. - **Prompt training strategy**: - Use implicit continuous representations as soft prompts to guide the generation of future events. The collection of prompts is supervised by a contrastive discriminator to ensure the consistency between the generated explanations and the input events. ### Conclusion The paper proposes a new FEG framework Coep by integrating inferential knowledge and sequential knowledge, which can maintain logical coherence and interpretability when generating future events. The experimental results show that Coep is superior to existing methods in multiple aspects and provides new ideas for future research.

What Makes the Story Forward? Inferring Commonsense Explanations as Prompts for Future Event Generation

Principled Knowledge Extrapolation with GANs.

What Happens Next? Event Prediction Using a Compositional Neural Network Model

Go Back in Time: Generating Flashbacks in Stories with Event Temporal Prompts

EventGround: Narrative Reasoning by Grounding to Eventuality-centric Knowledge Graphs

Generating Commonsense Explanation by Extracting Bridge Concepts from Reasoning Paths.

What Would Happen Next? Predicting Consequences from An Event Causality Graph

x-ECON: Explainable Event-Consumption Commonsense Reasoning

Local Explanation of Dialogue Response Generation

KGR^4: Retrieval, Retrospect, Refine and Rethink for Commonsense Generation

OpenEP: Open-Ended Future Event Prediction

Reasoning before Responding: Integrating Commonsense-based Causality Explanation for Empathetic Response Generation

Constructing Narrative Event Evolutionary Graph for Script Event Prediction

KGExplainer: Towards Exploring Connected Subgraph Explanations for Knowledge Graph Completion

A Pluggable Common Sense-Enhanced Framework for Knowledge Graph Completion

Contextualized Hybrid Prompt-Tuning for Generation-Based Event Extraction.

From Simple to Complex: A Progressive Framework for Document-level Informative Argument Extraction

Object-Centric Cross-Modal Knowledge Reasoning for Future Event Prediction in Videos

Robust Stochastic Graph Generator for Counterfactual Explanations

Knowledgeable Storyteller: A Commonsense-Driven Generative Model for Visual Storytelling

Cooperative Explanations of Graph Neural Networks.