5 results about "Levenshtein distance" patented technology

This invention discloses an iterative chart code generation method and apparatus based on a chart and a large-scale code correction model, belonging to the field of chart code generation technology. The method includes: based on a code error classification system, performing data mutation using a multimodal large-scale model based on the original chart code pairs, and constructing an original dataset; based on the code error classification system and the row-level Levenshtein distance algorithm, adding correction prompts to the original dataset, and constructing a training dataset; based on the LLaMA-Factory fine-tuning framework and the GRPO training strategy, performing two-stage training on the large-scale code correction model based on the training dataset to obtain an optimized large-scale code correction model; and based on the optimized large-scale code correction model and the large-scale chart code generation model, iteratively generating code based on a target reference chart to obtain the target generated code. This invention is a high-precision, interpretable, and iterative chart code generation method based on charts.

The application relates to the technical field of industrial manufacturing, and discloses a method and system for improving the completion parameters of understanding user input, which comprises the following steps: inputting an industrial standardized file, generating a semantic label, reasoning through the industrial standardized file and the semantic label, obtaining an optimized industrial standardized file, and obtaining an industrial rule set and a text parameter set input by a user. The application realizes accurate semantic label generation based on a WordNet synonym set, solves the polysemy ambiguity problem, calculates the similarity between a candidate parameter and a user conversation through an improved Levenshtein distance algorithm and a compensation mechanism, accurately captures the potential demand of the user, improves the consistency with the actual input intention of the user, multiplies the initial weight value, the click rate and the time decay factor to sort the candidate parameters, balances the instant production demand and the historical experience in the industrial scene, and significantly improves the accuracy, adaptability, efficiency and interpretability in four dimensions.

The application provides a text difference degree calculation method and system, and the method comprises the following steps: obtaining the length of a first text and the length of a second text; in the case that the length of the first text and the length of the second text satisfy a first preset condition, calculating the difference degree between the first text and the second text according to a target Levenshtein distance between the first text and the second text, the length of the first text and the length of the second text. The application calculates the difference degree between the texts to be compared based on the length of the texts to be compared and the target Levenshtein distance, solves the problems of slow manual comparison speed and complicated operation in the case that the amount of text data is very large, improves the calculation efficiency of the difference degree of the texts to be compared, and enables the user to quickly understand the difference degree between the texts to be compared.

The application discloses a kind of cross-system document intelligent coding method and system based on dynamic rule engine, by obtaining the metadata information of to-be-coded document, call dynamic rule engine from configurable rule base and match coding rule, rule engine supports multi-dimensional condition matching and priority dynamic calculation, and provide visual configuration interface to realize the real-time adjustment and version management of rule;Adopt three-level mapping strategy to convert the metadata of PDMS, BIM, SCADA and other heterogeneous systems into unified equipment and facility list ID without loss, combined with improved Levenshtein distance algorithm to calculate the similarity of equipment name, form the dual conflict detection mechanism of equipment ID accurate matching and equipment name similarity calculation, trigger artificial review when conflict is detected, review result is fed back to knowledge base to realize self-learning optimization, while establishing coding life cycle management;The application realizes the automation, intelligentization and traceability of cross-system document coding, significantly improves coding consistency and standard adaptation efficiency.

FIELD: computer technology.SUBSTANCE: method for multilingual training of language models using reinforcement learning based on human feedback using synthetic gaze trajectories, comprising the steps of feeding multilingual text to a gaze prediction model based on a multilingual eye movement corpus and a multilingual BERT model to generate a fixation sequence, computing visual attention features including first-pass regression rate, skip rate, first-pass and total fixation counts, and normalized Levenshtein distance, generating a gaze-aware reward model by projecting features into a latent space through a fully connected network, distributing rewards to tokens proportional to fixation probabilities, and optimizing the policy using PPO or GRPO algorithms with a modified advantage that takes into account the Levenshtein distance.EFFECT: multilingual training in thirteen languages without collecting real eye tracking data, achieving the accuracy of reproducing the features of visual attention.6 cl