CN116549967A

CN116549967A - Interactive reasoning game system for dynamically generating scenario and role based on AI

Info

Publication number: CN116549967A
Application number: CN202310600890.9A
Authority: CN
Inventors: 王浩臣
Original assignee: Individual
Current assignee: Individual
Priority date: 2023-05-25
Filing date: 2023-05-25
Publication date: 2023-08-08

Abstract

The invention relates to an interactive reasoning game system for dynamically generating scenario and role based on AI, which utilizes AI natural language processing technology to create dynamically generated collusion and killing reasoning game scene. Game backgrounds, characters, events, and timelines are all generated in real-time by AI, providing a unique and highly re-playable storyline and character interaction. The system has the function of dynamically communicating with players, generating rich storylines according to player selection, and creating complex roles with individuality, background stories and motivations by using AI. Players can collect clues through natural language conversations and character interactions, and the system provides context-aware puzzles and clues that need to be resolved to advance the game. The system can adjust the difficulty according to the player performance and adapt to users with different skill levels. Supporting a multi-person mode, a collaborative or competitive puzzle solution is possible. And iterating through player feedback and player interaction to refine the model.

Description

Interactive reasoning game system for dynamically generating scenario and role based on AI

Technical Field

The invention belongs to the field of computer game design and artificial intelligence, and particularly relates to an AI technology-based reasoning game system.

Background

The inference game is a form of game of lively education, which can exercise a player's logical thinking, inference ability and judgment. However, the existing reasoning games are usually single in scenario, fixed in character setting, and lack of fun and challenge. In addition, existing gaming systems are unable to dynamically generate scenarios and characters based on player behavior and preferences, which limits game playability and attractiveness.

The key links of character allocation, scenario generation, clue distribution and the like in the existing system often depend on manual addition. This not only easily results in cumbersome operations, inefficiency, but may also affect the fairness of the game. In addition, the intelligent degree of key links such as role identification, scenario generation and the like in the conventional scenario killing game system is low, and real automatic generation and individuation are difficult to realize. For example, character recognition and scenario generation in existing systems is typically based on pre-written scripts and fixed character settings, lacking sufficient flexibility and innovativeness.

The prior art has weaker self-adaptive capacity in aspects of scenario clue generation, role interaction, accidental event triggering and the like, and cannot be adjusted in real time according to actual behaviors of players and game progress. This limits to some extent the development and innovation of games.

With the development of computer science, artificial intelligence, and natural language processing technologies, these technologies have played an important role in a variety of applications, including computer game design. In particular, large natural language processing technology (BERT, GPT) can enable AI to understand and generate human language, and bring more abundant and realistic player experience to game design. Aiming at the limitations of the prior art, the patent provides an AI-based interactive reasoning game system for dynamically generating the scenario and the roles, which aims to overcome the defects of the prior art and provide a richer, interesting and immersive game experience for players.

Disclosure of Invention

The invention aims to provide an interactive reasoning game system for dynamically generating scenario and roles based on AI.

In a first aspect, the system includes a game engine, a realistic character interaction module, a context-aware puzzle and cue module, an adaptive difficulty module, and a multiplayer game mode. The system can utilize the large pre-trained natural language model AI technology to generate different storylines, roles, events and timelines in real time according to the selection of players, the game progress and other factors, and continuously optimize the model according to player feedback and the game times.

With reference to the first aspect, in a first possible implementation manner of the first aspect, the game engine uses other similar mature big data natural language processing technologies such as BERT, GPT, and uses the preprocessing scenario to generate dynamic scenarios, characters, events, and timelines by repeatedly training and fine tuning the preprocessing scenario. The game engine will generate a unique and rich storyline based on the player's selections and the progress of the game. In addition, the game engine creates characters with their own personalities, background stories, and motivations from the existing scenario.

With reference to the first aspect, in a second possible implementation manner of the first aspect, the realistic character interaction module uses a natural language processing technology to allow a player to conduct a natural language dialogue with a game character. In this way, the player can collect clues to drive the progress of the game.

With reference to the first aspect, in a third possible implementation manner of the first aspect, the context-aware puzzle and cue module generates relevant puzzles and cues according to a progress of a player in a game. This increases the challenges of the game and helps the player make progress in the game.

With reference to the first aspect, in a fourth possible implementation manner of the first aspect, the adaptive difficulty module uses AI technology to automatically adjust the difficulty of the game according to the performance of the player in the game. In this way, all players, regardless of their skill level, are able to get a game experience that suits them.

With reference to the first aspect, in a fifth possible implementation manner of the first aspect, the multiplayer game mode allows players to cooperatively solve the puzzle together, and also can compete with each other to see who first solves the puzzle. The multi-player game mode makes the game more interactive and competitive, and increases the interest of the game.

With reference to the first aspect, in a sixth possible implementation manner of the first aspect, the optimizing the model and providing a better game experience are further performed through feedback of the player and interaction between the AI and the player in each game.

Through the aspects and possible implementation manners, the invention provides the interactive reasoning game system for dynamically generating the scenario and the roles based on the AI, which can generate different storylines, roles, events and timelines according to the preference and the game progress of the player, and provides a highly immersive and personalized game experience for the player. The realistic role interaction module enables a player to conduct natural language dialogue with a game role, the context-aware puzzle and clue module generates relevant puzzles and clues according to the game progress, the self-adaptive difficulty module adjusts game difficulty according to the performance of the player in a game, and the multi-player game mode provides diversified game experience of cooperation and competition.

The specific implementation details of the AI-based dynamic generation scenario and character interactive reasoning game system of the present invention may vary, but various modifications and changes may be made thereto without departing from the spirit and scope of the invention.

Drawings

FIG. 1 is a schematic diagram of an embodiment of an AI-based dynamic scenario and character generation interactive reasoning game system of the present invention.

Detailed Description

For a better understanding of the present invention, a detailed description will now be given of a specific embodiment of an AI-based interactive reasoning game system for dynamically generating scenarios and characters. The particular components and features illustrated in the drawings and described herein are merely exemplary and should not be considered limiting of the present invention. It will be appreciated by those skilled in the art that modifications and variations can be made to the invention without departing from its spirit or scope.

The following describes in detail the specific embodiment of the interactive reasoning game system based on the dynamic generation scenario and role of AI by taking the material of "West tour memory" as an example. The present embodiment includes the steps of: the method comprises the steps of (1) game starting, (2) level selection, (3) map loading, (4) role loading, (5) murder and victim selection, (6) murder timeline generation, (7) other role timeline generation, (8) overall adjustment, (9) game progress, (10) puzzle solving, (11) unexpected event triggering, (12) murder confirmation, (13) game settlement, (14) ending/replay/selection of other levels and (15) game feedback and perfection.

Game start (1): the game engine is based on a pre-trained language model (such as BERT, GPT and the like) and trains and fine-tunes the western pleasure and hierarchical data, so that the model can better understand the storyline and generate the storyline close to the storyline. In addition, in the whole reasoning game model engine, a large number of true cases and novel cases of the trial class are trained, so that the model is more diversified in the design of reasoning drama.

Checkpoint selection (2): the system selects the next level or repeatedly selects the level that has been cleared according to the player's game progress. Upon selection of the level, the game will load an initial game background story.

Map loading (3): the system selects a corresponding map according to the level and adds the map data, the scenes appearing in the map, and the background story of each scene to the game background.

Role loading (4): the number of participants is determined based on the corresponding difficulties. The participants may be AI-controlled characters and players. The player can select his favorite character and the system will assign the remaining characters to the AI system based on the number of players. Thus, each level will have a different combination of roles, increasing the diversity and complexity of the game.

Murder and victim choice (5): the murder may be a plurality of persons or a single person, and the victim may be one or more. This choice will be chosen randomly by the system.

Murder timeline generation (6): a pre-trained language model (such as GPT-4) is used, and a time line, a task and a composition tool are generated according to the characteristics of the characters of the murder characters, map information and background information by combining a condition generation technology. The system can be proved out of the field for the design of the murder, so that the murder is more challenging to find.

Other roles timeline generation (7): based on the murder timeline, a pre-trained language model (e.g., GPT-4) is used to generate timelines, tasks, and composition tools for other roles.

Overall adjustment (8): the generated scenario, character setting and time line are integrally adjusted by using reinforcement learning (Reinforcement Learning), so that the method is more reasonable and interesting, and the continuity of game experience is ensured. Meanwhile, the interaction of all roles is rich, and the reasoning process is full of challenges.

Game play (9): real-time conversational systems use a background trained language model (e.g., GPT-4) to interact with players in natural language, generating reasonable actions and motivations for the characters based on their characteristics and identity to help or prevent the players from solving puzzles. The relationship between the character and the event is captured using a self-attention mechanism (attention mechanism in a Transformer), ensuring that the character's behavior remains consistent with the event. During game play, the player may communicate with other players and AI-controlled characters. They may ask questions, share clues, or discuss possible solutions for a killing.

Puzzle solution (10): puzzle answers are generated from player inputs and game states using deep learning models, such as Recurrent Neural Networks (RNNs) and Transformer structures. Based on the speed and accuracy with which the player solves the puzzle, the system will automatically adjust the puzzle difficulty, making the game more challenging.

Unexpected event trigger (11): an incident is generated using a pre-trained language model (e.g., GPT-4). The model is fine-tuned to generate incidents related to game context and character settings. Event difficulty is optimized using reinforcement learning techniques (Q-learning or Actor-Critic algorithms). By defining the appropriate bonus function, the lead model generates events of appropriate difficulty based on the skill level of the player and the progress of the game.

Murder confirmation (12): the NLP technology based on emotion analysis and intention recognition can help AI roles to behave more naturally when answering questions, and improve game challenges. The AI character will answer based on the story background, its own timeline, whether it is a murder or not, and the player's question. If the AI character is a murder, it will hide itself as much as possible while answering the question. The answer to the AI character will be different in each game due to the dynamics of the game.

Game settlement (13): after the game is finished, the system provides a multi-disc function. The player can see the complete timeline and can ask the AI character questions again. This time, the AI character will fully describe the facts, helping the player to understand the full view of the event. And generating popular and easily understood event descriptions by combining Natural Language Generation (NLG) technology, and helping players to know the full view of the event.

Ending/replay/select other checkpoints (14): new levels and game content are recommended based on player behavior and preferences using reinforcement learning and collaborative filtering techniques. Players may also choose to replay the same level, but each game will have a different experience and outcome due to the dynamic generation capabilities of the system.

Game feedback and perfection (15): by collecting player feedback and analyzing player interactions with AI characters in each game, the pre-training model is continuously optimized and better game experience is provided. The system may also adjust the generated storyline and character settings through the player's experience of success and failure, making them more challenging.

Through the specific implementation mode, the interactive reasoning game system based on the AI dynamically generated scenario and the characters can be created, and the reasoning game experience of players under the western game materials is provided. Such gaming systems are highly playable, versatile, and challenging and can attract a wide population of players.

Claims

1. An AI-based interactive reasoning game system for dynamically generating scenarios and characters, the system comprising: the dynamic scenario generation module, the role generation module, the dialogue interaction module, the clue and puzzle generation module and the difficulty adjustment module; the dynamic scenario generation module generates a story line according to player selection; the character generation module creates a complex character with personality, background stories and motivations; the dialogue interaction module realizes natural language interaction between players and roles; the clue and puzzle generation module provides context-aware puzzles and clues; the difficulty adjusting module adjusts game difficulty according to player performance; and continuously optimizing the model through player feedback and game times.

2. The interactive inferential gaming system of claim 1, wherein the system supports a multiplayer model, allowing players to collaborate or compete to solve a puzzle of reasoning.

3. The interactive reasoner gaming system of claim 1, wherein said system is operable across platforms, including smartphones, tablet computers and personal computers.

4. The interactive inferential gaming system of claim 1, wherein feedback from players of the system and interaction with players continues to optimize the model.

5. The interactive reasoning game system of claim 1, wherein the dynamic scenario generation module and character generation module are implemented using large natural language processing techniques.

6. The interactive reasonement game system according to claim 1, wherein the dialogue interaction module supports a player to collect clues through natural language dialogue.

7. The interactive inferential game system of claim 1, wherein the context-aware puzzles and cues provided by the cues and puzzle generation module need to be resolved to advance the game.

8. The interactive inferential gaming system of claim 1, wherein the difficulty adjustment module is adaptable to users of different skill levels, ensuring that all users have a pleasant gaming experience.

9. The interactive inferential gaming system of claim 1, wherein the model continues to be optimized and provides a better gaming experience through feedback from the players and interaction with the players in each game.