18 results about "Experience base" patented technology

The invention provides an enhancement method for a question answering system during testing based on reinforcement learning, and belongs to the technical field of natural language processing. The method aims at solving the technical problems that an existing large language model (LLM) faces knowledge blind areas, reasoning chain breakage and context length limitation in a question and answer (QA) task, an existing fine adjustment method is high in calculation cost and damaged in generalization ability, and a prompt strategy seriously depends on a limited context window and lacks long-term memory. The method comprises the following steps: firstly, collecting reflection experience through multiple attempts and a failure reflection mechanism, and constructing an experience library; thirdly, performing text embedding, clustering and semantic abstracting on experiences in the experience library, and constructing an external memory library; secondly, a memory selection process is formalized into a Markov decision process (MDP), reinforcement learning (such as a PPO algorithm) is used for training a strategy agent, and the agent is optimized with a current task as a state, with memory item selection as an action and with LLM answer correctness as a reward; finally, in a test time (inference) phase, the agent dynamically selects the most helpful memory entry according to the new task and integrates it into the hint of the LLM to generate a final answer. According to the method, under the condition that LLM internal parameters do not need to be updated, the question and answer accuracy is dynamically improved, the calculation overhead is reduced, and the limitation of a context window is effectively overcome.

One or more embodiments of the present specification provide a dialogue system, a dialogue experience library construction method, and a question answering method. The dialogue system comprises a teacher dialogue model, a student dialogue model and a scheduling module, and in a preprocessing stage, the student dialogue model is used for answering sample questions to obtain corresponding sample answers; the teacher dialogue model is used for evaluating the answering process of the student dialogue model at least based on the sample question and the sample answer to obtain a final evaluation result containing experience information; the scheduling module is used for recording the experience information in a dialogue experience library; in a question answering stage, the scheduling module is used for generating prompt words according to a target question proposed by a user and inputting the prompt words into the student dialogue model; the student dialogue model is used for answering the question according to a target answering strategy indicated by target experience information matched with the target question in the dialogue experience library; and the scheduling module is used for returning answers output by the student dialogue model to the user.

We provide the system. [Solution] An artificial intelligence means for collecting user information and generating an individualized learning plan based on said information, A user interface means that provides the user with an interactive learning experience based on the aforementioned learning plan, A business plan development support tool that analyzes users' business ideas and provides feedback on improvement suggestions, A means of forming a network for collaborating with other users on projects, Resource management tools that provide financial support and mentoring for projects, A system that includes this.

The invention relates to determining activities to be performed by non-player characters (NPC) in an interactive experience, e.g. video game. Activities, e.g. dialogue, are generated using generative machine learning models using a received input activity from a player, an entire established context / lore within the experience, and a set of activity history within the experience, as input. The invention involves updating the entire established context / lore based on a comparison of semantic content of the entire established context and semantic content of the generated activity of the NPC. The invention involves determining subsequent activities to be performed by NPCs based on the updated entire established context / lore, thereby providing consistency within the experience.

The invention provides a GRPO-like self-optimization prompt generation method and system, and the method comprises the steps: carrying out the processing of an input query statement through a large language model, generating a group of candidate outputs, and forming an output group; evaluating each candidate output through a reward model to obtain a reward score corresponding to each candidate output; the big language model performs semantic comparison on the output group based on the reward score corresponding to each candidate output to generate semantic advantage description in a natural language form; converting the semantic advantage description into a modification operation instruction, and updating an external experience library based on the modification operation instruction; injecting the updated strategy experience in the external experience library as a context prompt into the input of the large language model, and guiding the large language model to output a new output group which shifts to a high reward score when processing an input new query statement; according to the method, the adaptive improvement of the model behavior can be realized in the reasoning stage without any gradient update.

This application relates to a method and apparatus for interactive Peking Opera experience based on motion capture, belonging to the field of interactive experience technology. The method includes: acquiring user action data from multiple general users using motion capture; calculating group attention parameters and group learning difficulty parameters for multiple target stylized Peking Opera movements; extracting high-value movement slices from the multiple target stylized Peking Opera movements based on the group attention parameters and group learning difficulty parameters to obtain a movement slice sequence; and dynamically adjusting the experience content of subsequent Peking Opera interactive experiences based on the movement slice sequence. This invention solves the problem that traditional Peking Opera interactive experiences often push fixed content, failing to adapt to the actual interests and learning difficulties of ordinary users, resulting in insufficient exposure to content of interest and a lack of targeted guidance for difficult movements, thereby reducing experience participation and learning effectiveness.

The invention provides a text travel interactive experience data driving system based on virtual-real fusion and 3D game elements, and belongs to the technical field of smart text travel and data interaction.The system comprises a cognitive vector construction module, an instruction acquisition module, a task generation module, a trigger determination module and a parameter optimization module. And generating an optimal narrative action instruction in combination with the target narrative entropy curve, generating a virtual-real fusion interaction task based on the cross-scene component set, determining a credible trigger vector of the sub-virtual event to the real world service, and optimizing the cross-scene component set through user feedback. According to the method, the problems of experience flattening, narrative discretization and the like of an existing system are solved, personalized, coherent and deeply immersive text travel interaction experience is realized, and the interaction suitability and the intelligent level are improved.

A system and method for reading development in children with dyslexia using word-to-graphic transformations are disclosed. The method includes presenting words to the learner and converting the words into corresponding graphical representations that visually depict the meaning of the words. The method further includes associating phonemes with parts of the graphical representation to help the learner connect speech sounds with written letters. Next, the method includes progressively simplifying the graphical representation of words as the learner's reading proficiency improves, transitioning them toward recognizing standard written words. The method also includes tracking the learner's progress through a proficiency tracking module, which dynamically adjusts the complexity of future word transformations based on performance data. Additionally, the method provides real-time auditory and visual feedback during the learning process to reinforce phoneme-grapheme associations. The system operates on an interactive platform, such as a tablet, allowing personalized learning experiences based on the child's individual progress.

We provide the system. [Solution] A means by which an information processing device collects descriptive information related to past eras from a communication network, Means for analyzing the aforementioned explanatory information and generating information representing the three-dimensional structure of the region, A means for generating a virtual reality experience based on the aforementioned three-dimensional structural information, Methods for incorporating educational elements into virtual reality experiences, Means for providing the virtual reality experience to the user device, A system that includes this.

The invention relates to the technical field of problem processing, and discloses a problem solving-based machine learning method, which comprises the steps of event input, problem identification, scheme matching and output, empirical corpus precipitation and updating and iterative optimization. After event phenomenon corpora are collected through an event input port, a user initiative question and AI intelligent mining double-path identification problem is adopted; the AI intelligent agent outputs implicit questions and priorities through corpus preprocessing, feature extraction and question recognition model analysis; matching an identification problem and an experience library problem with a solution corpus, judging a matching result through a semantic similarity algorithm and a dynamic threshold value, and pushing an unmatched problem to an expert for processing; the expert scheme and the problem are bound and stored in the experience library, the process is repeated to realize experience library iteration and AI optimization, and the problem processing efficiency and the experience reuse rate are improved.

An object of a system according to an embodiment is to provide a learning experience personalized on the basis of a user's daily life and emotion.SOLUTION: A system includes a diary collection unit, a life log collection unit, an analysis unit, and a learning content generation unit. The diary collection unit collects a diary of a user. The life log collection unit collects a life log of a user. The analysis part analyzes the data collected by the diary collection part and the life log collection part. The learning content generation unit generates learning content on the basis of the data analyzed by the analysis unit.SELECTED DRAWING: Figure 1

We provide the system. [Solution] A generative artificial intelligence model that generates learning materials based on user information, A communication means for acquiring user information and distributing corresponding learning materials to an information processing device, An analysis means that analyzes the results of user participation in learning materials and generates feedback to show learning progress based on the analysis results, A presentation means that transmits the feedback to an information processing device and presents it to the parent and child visually or audibly, A means of providing safety tests and situational experiences based on a user's past online interaction history, A system that includes this.

An object of the system according to the embodiment is to propose a unique experience on the basis of a user's interest or preference and store the unique experience as a digital time capsule.SOLUTION: A system includes a learning unit, a proposal unit, and a storage unit. The learning unit learns the user's interests, preferences, and dreams. The proposal unit proposes a unique experience or activity on the basis of the information learned by the learning unit. The storage unit stores the experience or activity proposed by the proposal unit as a digital time capsule.SELECTED DRAWING: Figure 1

The invention relates to determining activities to be performed by non-player characters (NPC) in an interactive experience, e.g. video game. Activities, e.g. dialogue, are generated using generative machine learning models using a received input activity from a player, an entire established context / lore within the experience, and a set of activity history within the experience, as input. The invention involves updating the entire established context / lore based on a comparison of semantic content of the entire established context and semantic content of the generated activity of the NPC. The invention involves determining subsequent activities to be performed by NPCs based on the updated entire established context / lore, thereby providing consistency within the experience.

An object of the system according to the embodiment is to accurately grasp the individuality and interest of a child and provide a learning plan and experience based on the individuality and interest.SOLUTION: A system according to an embodiment includes an interaction input unit, a work input unit, an analysis unit, a learning plan providing unit, an experience providing unit, and a parent support unit. The dialog input unit inputs a dialog with the child. The work input unit inputs a work created by a child. An analysis part analyzes the individuality and interest of the child on the basis of information obtained from the interaction input part and the work input part. The study plan providing unit provides a customized study plan based on the analysis result obtained by the analysis unit. The experience providing unit provides a customized experience based on the analysis result. The parent support unit provides information for understanding the talent and growth of the child to the parent based on the analysis result.SELECTED DRAWING: Figure 1

An object of a system according to an embodiment is to efficiently and effectively learn the contents of a history textbook.SOLUTION: A system according to an embodiment includes a text analysis unit, a summary generation unit, a question answering unit, and an interactive learning unit. The text analysis unit analyzes a text of a history textbook. The summary generation unit extracts an important point from the text analyzed by the text analysis unit and generates a summary. The question answering unit provides an appropriate answer to the user's question based on the summary generated by the summary generation unit. The interactive learner provides an interactive learning experience to the user based on the answer provided by the question answering unit.SELECTED DRAWING: Figure 1

The invention relates to a Beijing Opera interactive experience method and device based on motion capture, and relates to the technical field of interactive experience, and the method comprises the steps: obtaining user motion data of a plurality of generic experience users through combination with motion capture; calculating group attention parameters and group learning difficulty parameters of the plurality of target Beijing Opera stylized actions; according to the group attention parameters and the group learning difficulty parameters, high-value action slices are extracted from the multiple target Beijing opera stylized actions, and an action slice sequence is obtained; and dynamically adjusting the experience content of the follow-up Beijing Opera interactive experience based on the action slice sequence. According to the method and the device, the problem that the experience participation degree and the learning effect are reduced due to the fact that the user is in insufficient contact with interested contents and lacks targeted guidance for actions difficult to master because most traditional Beijing Opera interactive experience is fixed content pushing and cannot be matched with learning obstacles in combination with actual interests and preferences of a common user is solved.

We provide the system. [Solution] A generative artificial intelligence means that collects user information and personalizes the learning experience based on that information, A means of providing vocational experience using virtual reality technology, A means to automatically set daily goals and manage their progress, A user interface that allows parents to check their child's learning progress, A means of encrypting user data using blockchain technology and storing it securely, A system that includes this.