Financial English learning method based on AI multi-modal evaluation and virtual simulation

Abstract

The invention discloses a financial English learning method based on AI multi-modal evaluation and virtual simulation. The financial English learning method is suitable for financial English teaching scenes. A real-time interaction scene between the learner and the AI virtual character is generated through a virtual simulation technology, and multi-modal data, including language, action and emotional characteristics, of the learner is collected; and in combination with the standard characteristics of the virtual scene, a multi-modal score is calculated by using an integral formula, and a comprehensive score is optimized through a dynamic weight adjustment algorithm. According to the method, specific task completion degree parameters of financial English are introduced, the task completion condition of a learner is evaluated, and a learning path is dynamically adjusted based on a comprehensive score. The method mainly comprises the steps of virtual simulation scene initialization, multi-modal feature extraction, multi-modal score calculation, dynamic weight adjustment and comprehensive scoring and learning path optimization. According to the invention, comprehensive evaluation, real-time feedback and personalized path design are realized, the problem of lack of dynamic adaptation and scene reality sense in traditional teaching is solved, and the financial English learning effect and evaluation precision are significantly improved.

Description

Technical Field

[0001] The present invention relates to the fields of virtual simulation technology and artificial intelligence, and specifically to a financial English learning method based on AI multimodal evaluation and virtual simulation. Background Art

[0002] With the deepening of globalization, financial English, as a key skill in business and finance, has become an important part of vocational education. However, traditional financial English teaching has the following problems: 1. The lack of real financial scenarios makes it difficult for students to apply what they have learned into practice; 2. The single assessment method makes it difficult to comprehensively measure learners' language, movement, and emotional abilities; 3. Personalized teaching is difficult to achieve, and learning content cannot be dynamically adjusted according to the performance of different students.

[0003] To address these issues, virtual simulation and artificial intelligence technologies have been introduced into financial English teaching. Existing approaches primarily use virtual simulation to construct static scenarios, and artificial intelligence primarily for language scoring. These approaches lack real-time interactivity and multimodal assessment mechanisms. Furthermore, the ability to dynamically adjust learning paths to match learner performance is also limited. Therefore, a financial English learning method that combines dynamic interaction with virtual characters and adaptive multimodal parameters is urgently needed to address these issues. Summary of the Invention

[0004] The objective of this invention is to propose a financial English learning method based on AI multimodal assessment and virtual simulation, combining the task completion parameters unique to financial English, and using dynamic interaction and multimodal evaluation technology to achieve real-time learning feedback and path optimization.

[0005] The main innovative features of the present invention include: Dynamic interaction of virtual characters: In virtual simulated financial scenarios, multimodal data, including language, action, and emotional characteristics, is collected through real-time interaction between AI virtual characters and learners.

[0006] Multimodal parameter adaptive weighting algorithm: Based on the learner's real-time performance, the weights of multimodal evaluation dimensions are dynamically adjusted to optimize the comprehensive score.

[0007] Financial English task completion parameter: By calculating the learners' completion status in specific tasks, it is included as a scoring parameter in the comprehensive assessment.

[0008] The present invention can be widely used in financial English teaching, providing functions of real-scene simulation, comprehensive evaluation and personalized learning path optimization, and solving the limitations of traditional teaching models. BRIEF DESCRIPTION OF THE DRAWINGS

[0009] Figure 1 : Main flow chart of a financial English learning method based on AI multimodal assessment and virtual simulation.

[0010] Figure 2 : Detailed flowchart of point scoring and dynamic weight adjustment. DETAILED DESCRIPTION

[0011] The present invention comprises the following main steps: Step S100: Initialization of virtual simulation financial scenario Generate AI virtual characters in a virtual simulation environment and learner roles .

[0012] Set up financial English scenarios (such as business negotiations, financial reports, cross-border e-commerce live broadcasts) and related tasks .

[0013] Collect multimodal data from learners ,include: 1. Language data: learner’s speech signals ; 2. Action data: learner’s action characteristics ; 3. Emotional data: learners’ facial expressions and voice intonation .

[0014] Step S200: Multimodal feature extraction The following features are extracted from the language data:

[0015]

[0016] Extracting motion features from motion data .

[0017] Extracting emotional features from emotional data .

[0018] Extracting standard features from virtual scenes As a comparison benchmark.

[0019] Step S300: Multimodal score calculation Calculate the difference score between the learner and the standard features , using the integral formula:

[0020] in, It is an adjustment parameter used to control the attenuation of feature weights.

[0021] Discrete integration implementation:

[0022] Step S400: Dynamic weight adjustment Calculate the average of the scores for each dimension :

[0023] Dynamically adjust weights :

[0024] in, is the score increment, is the initial weight.

[0025] Weight normalization:

[0026] Step S500: Comprehensive scoring and learning path optimization Calculate the overall score :

[0027] in, Rate task completion:

[0028] Dynamically select learning paths :

[0029] Beneficial effects 1. This invention collects learners' multimodal data through dynamic interaction of virtual characters, and combines it with a dynamic weight adjustment algorithm to comprehensively evaluate the learners' language, movement and emotional characteristics, thereby improving the evaluation accuracy.

[0030] 2. A comprehensive scoring mechanism based on the Financial English task completion parameters ensures that the assessment results are closely aligned with the learning objectives.

[0031] 3. Provide real-time feedback and personalized learning path adjustment functions, effectively solving the problem of dynamic adaptation in traditional teaching.

[0032] in conclusion This invention innovatively combines AI multimodal assessment and virtual character dynamic interaction technology in the virtual simulation teaching of financial English learning, optimizes the learning path through integral scoring and dynamic weight adjustment, and has broad application prospects and practical value.

Claims

1. A financial English learning method based on dynamic interaction of virtual characters, characterized in that: The following steps are involved: Step S100: Generate AI virtual character A in a virtual financial scenario avatar and learner role R learner And collect learners' multimodal data X learner (t), the data includes pronunciation data, action data and emotion data; Step S200: collect the data X learner (t) Perform feature extraction, including: Extract pronunciation features MFCC learner (t) and fundamental frequency curve P learner (t); Extract action features M learner (t) and emotional characteristics E learner (t); Step S300: Based on the standard feature X in the virtual simulation scene standard，i (t), use the following integral formula to calculate the learner's difference score S on multimodal dimension iii i : Among them, λ i is the integral decay parameter. Step S400: Dynamically adjust the weight w of the scoring dimension i , the adjustment formula is: in, is the average of the multimodal ratings. Step S500: Calculate the learner's total score S total : The total score is used to dynamically adjust the learning path L next The path includes a basic training module, an intensive training module and an advanced scenario simulation module.

2. A financial English learning method based on dynamic interaction of virtual characters according to claim 1, characterized in that: The integral formula is implemented by the following discretization: Where K is the total number of discrete time points, Δt = t k+1 -t k .

3. The financial English learning method based on dynamic interaction of virtual characters according to claim 1, characterized in that: Dynamic weight w i The adjustment is done by normalization, the formula is: Where n is the number of multimodal dimensions.

Images