A dual-channel adaptive language control system based on semantic bridge structure driving

Abstract

The application discloses a kind of dual-channel adaptive language control systems based on semantic bridging structure driving.The system constructs semantic bridging sequence data structure, and generates training state vector in combination with multi-modal state feature, carries out adaptive channel scheduling and weight distribution between cognitive input channel and cognitive output channel, realizes dynamic mode switching and feedback closed-loop optimization in language training process.The system includes semantic bridging generation unit, state feature extraction unit, channel control unit, input immersion channel unit, output calibration channel unit, feedback reinforcement unit, behavior modeling unit and interface docking unit.Through the fusion analysis to earphone state, speech activity, sliding rhythm and the like signal, the system dynamically adjusts semantic transition weight and rhythm parameter, realizes gradual control and structural stability guarantee in semantic migration process.Interface docking unit supports standard teaching protocol call and external platform data interaction.

Description

Technical Field

[0001] This invention relates to the fields of language cognitive control and artificial intelligence semantic modeling technology, specifically to a dual-path adaptive language control system driven by a semantic bridging structure. Background Technology

[0002] Existing language control or training systems typically employ static content presentation or a single training path execution mechanism. Their training processes are mostly based on preset rules or fixed logic, lacking data-driven control capabilities based on semantic bridging structures. While existing systems can make some adjustments based on user behavior, their mode switching and content adjustment often rely on simple rule judgments, failing to construct an evolvable dual-path control architecture and making it difficult to achieve structural stability and rhythmic adaptive optimization during semantic transfer. Therefore, in terms of multimodal state perception and semantic evolution control, there is still a lack of a language control system capable of dynamically scheduling between cognitive input and cognitive output pathways based on user state and training content characteristics. Summary of the Invention

[0003] To address the problem that existing language control systems lack data-driven control capabilities based on semantic bridging structures, and are unable to achieve dynamic scheduling and adaptive feedback optimization between cognitive input and output pathways, this invention provides a dual-path adaptive language control system driven by a semantic bridging structure.

[0004] According to one aspect of the present invention, the system includes: a semantic bridging generation unit, a state feature extraction unit, a path control unit, an input immersion path unit, an output calibration path unit, a feedback reinforcement unit, a behavior modeling unit, and an interface docking unit. The system is characterized in that it constructs a semantic bridging sequence data structure and generates a training state vector based on multimodal state signals. The path control unit adaptively schedules and assigns weights between the cognitive input path and the cognitive output path, thereby achieving dynamic switching and closed-loop optimization of the training structure. The path scheduling is not triggered by a single fixed rule, but is dynamically determined based on the joint calculation results of the training state vector and the weight matrix.

[0005] This invention uses a semantic bridging structure as the core of training data organization, abstracting the language transfer process into an adjustable sequence weight model. It constructs a training state vector using multimodal behavioral signals and continuously updates the path weight parameters using a feedback reinforcement mechanism, thereby achieving a balanced control between input immersion and output calibration in different training scenarios. Compared to existing systems based on static paths or simple rule judgments, this invention constructs an evolvable dual-path control architecture that maintains structural consistency and rhythmic adaptive stability during semantic transfer.

[0006] The training data construction, path scheduling, and feedback update processes in the language control process must all be executed based on the joint calculation results of the semantic bridging sequence data structure and the training state vector. Any control method that deviates from the semantic bridging structure or is triggered only by a single rule or static condition is not within the scope of protection of this invention.

[0007] This invention essentially constructs a language control scheduling mechanism based on semantic bridging structure and state-driven operation, in which the selection of training paths, rhythm adjustment and feedback optimization are all performed under structural constraints, thus distinguishing it from language training systems based on static rules or single-path execution.

[0008] The path scheduling process is modeled as a joint constraint solving process based on semantic bridging structure and training state vector. Any scheduling method not based on the joint constraint solving mechanism is not within the scope of protection of this invention. Attached Figure Description

[0009] The accompanying drawings, which are included to provide a further understanding of the invention and form part of this application, illustrate exemplary embodiments of the invention and, together with their description, serve to explain the invention and do not constitute an undue limitation thereof. In the drawings:

[0010] Figure 1 This is a flowchart of an optional language control system based on a dual-path control structure according to an embodiment of the present invention;

[0011] Figure 2 This is a flowchart of an optional automatic selection of training mode according to an embodiment of the present invention;

[0012] Figure 3 According to an embodiment of the present invention, an optional input immersion pathway unit presents a reading interface with color guidance and text masking mechanism;

[0013] Figure 4 This is a speech acquisition, comparison and feedback path under an optional output calibration path according to an embodiment of the present invention;

[0014] Figure 5 This is a signal processing and parameter update path for an optional personalized feedback mechanism according to an embodiment of the present invention;

[0015] Figure 6 This is a flowchart of an optional language control method based on a dual-path control structure according to an embodiment of the present invention;

[0016] Figure 7 A schematic diagram of the structure of a computer device suitable for implementing embodiments of the present disclosure is shown. Detailed Implementation

[0017] To enable those skilled in the art to better understand the present invention, the technical solutions of the present invention will be clearly and completely described below with reference to the accompanying drawings of the embodiments of the present invention. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort should fall within the scope of protection of the present invention.

[0018] It should be noted that the terms "first," "second," etc., in the specification, claims, and accompanying drawings of this invention are used to distinguish similar objects and are not necessarily used to describe a specific order or sequence. It should be understood that such data can be interchanged where appropriate so that the embodiments of the invention described herein can be implemented in orders other than those illustrated or described herein. Furthermore, the terms "comprising" and "having," and any variations thereof, are intended to cover a non-exclusive inclusion; for example, a process, method, system, product, or apparatus that comprises a series of steps or units is not necessarily limited to those steps or units explicitly listed, but may include other steps or units not explicitly listed or inherent to such processes, methods, products, or apparatus.

[0019] This application provides a dual-path adaptive language control system based on a semantic bridging structure. The system includes the following modules: a semantic bridging generation unit, a state feature extraction unit, a path control unit, an input immersion path unit, an output calibration path unit, a feedback reinforcement unit, a behavior modeling unit, and an interface docking unit. Figure 1 As shown, the system uses a semantic bridging generation unit as its core. A state feature extraction unit collects multimodal input signals, and a path control unit determines whether to enter the input immersion path or the output calibration path. The feedback reinforcement unit and the behavior modeling unit together form a training closed loop, used for rhythm adjustment, difficulty control, and personalized parameter updates.

[0020] The semantic bridging generation unit supports anchor word recognition and semantic transfer processing between native and target language texts, and integrates phrase structure analysis and corpus-level association logic to generate training content with gradual transitions.

[0021] The state feature extraction unit combines input signals such as headphone status, speech recognition activity, swipe trajectory, and reading duration to generate a training state vector for pattern judgment.

[0022] The path control unit dynamically selects either the input immersion path or the output calibration path based on judgment parameters formed by the training state vector, context state, and historical training behavior. In some embodiments, the path selection process can be implemented through a path decision function P = f(V, W), where V represents the training state vector, W represents the semantic bridging sequence weight matrix, and P represents the selection result of the current training mode. The path decision function is a joint decision function based on multi-dimensional state features and semantic bridging weights, and its output is constrained by the semantic bridging structure and cannot be directly determined by a single state variable or a fixed threshold rule.

[0023] The function is a constrained optimization function or a constraint satisfaction solution function. Its solution process is constrained by the semantic bridging structure and does not allow the use of unconstrained generation or direct decision-making methods based on probability prediction.

[0024] Figure 2 The process for automatically selecting the training mode is illustrated. System input parameters include headphone connection status, voice activity, page scrolling behavior, and ambient noise level. Through vector computation models or weight matrix operations, the system automatically selects between the input immersion path and the output calibration path, requiring no user intervention. Figure 2 Corresponding to the core control logic of this invention, this section explains the conditions, decision order, and execution results of mode switching.

[0025] The input immersive pathway unit presents a reading interface with color guidance and text masking mechanisms, which can be integrated with rhythm control components to dynamically adjust the reading rhythm and replacement strategy based on feedback. Figure 3 This illustrates the passive input characteristics and content rhythm control mechanism of the input immersion pathway, such as... Figure 3 As shown, the system automatically enters this mode when it detects that the user is in a quiet reading environment, presents the target language text and outputs audio simultaneously; the user completes immersive reading with visual and auditory input as the main input methods, during which the system can adjust the speech rate, masking range and keyword prompting strategy according to the page scrolling and pause rhythm.

[0026] The output calibration pathway unit captures the user's pronunciation in real time and compares it with the target language, calculates the rhythm offset and semantic synchronization, and feeds it back to the system for dynamic optimization. Figure 4 The speech acquisition, comparison, and feedback path under the output calibration path is shown, such as... Figure 4 As shown, after the system plays the target language training audio, it captures the user's pronunciation in real time and calculates rhythm shift, phoneme matching degree, and pronunciation continuity through a comparison module, outputting instant feedback scores or correction prompts. This mode supports operation in noisy environments, mobile scenarios, or under conditions of active user participation, and is the system's active output training path. Figure 4 The process flow reflects the complete speech processing chain from audio playback to feedback generation.

[0027] The feedback reinforcement unit generates feedback labels based on behavioral sequences such as the number of pauses, rhythm jumps, and repeated readings, and then adjusts the input content presentation strategy accordingly. Figure 5 The signal processing and parameter update path of the personalized feedback mechanism is illustrated. Input data includes behavioral indicators such as user reading scores, rhythm jumps, pause distribution, repetition count, and training duration. The feedback reinforcement unit dynamically adjusts the speech rate, content difficulty, prompt frequency, or training strategy based on the analysis results, forming a continuously optimized individualized learning curve. Figure 5 This invention is used to illustrate the personalized closed-loop mechanism of training rhythm, progressive difficulty, and experience adjustment.

[0028] The behavior modeling unit records the behavior parameters and rhythm response during each round of training, which are used for individual learning model iterative updates.

[0029] The interface unit supports the nested import of international teaching protocols such as SCORM and xAPI, and can also be linked with third-party TTS platforms, speech recognition modules or content management systems (CMS) through API.

[0030] When a user launches the system, the system initially determines the training mode based on the headphone / voice status. After inputting native language text, the semantic bridging generation unit automatically performs anchor point translation to generate training paragraphs adapted to the target language. The system then determines the current state and switches to either "reading training" or "follow-along training" mode. During training, the system collects behavioral feedback such as speech latency, rhythm jumps, and reading fluency. Based on behavioral parameters and training preference vectors, the system can dynamically adjust the rhythm, content masking range, or state vector-driven training mode. All training data is stored by the behavior modeling unit for continuous optimization of personalized training paths.

[0031] The present invention has the following beneficial effects:

[0032] (1) Dynamic switching of language training mode is achieved by driving path scheduling through training state vectors;

[0033] (2) Achieve gradual semantic transfer from native language expression to target language expression through semantic bridging sequences;

[0034] (3) The training rhythm and training difficulty are adaptively optimized through the feedback reinforcement mechanism, thereby improving the efficiency of language training.

[0035] According to an embodiment of the present invention, a dual-path adaptive language control method based on semantic bridging structure is provided. It should be noted that the steps shown in the flowchart in the accompanying drawings can be executed in a computer system such as a set of computer-executable instructions. Furthermore, although a logical order is shown in the flowchart, in some cases, the steps shown or described may be executed in a different order than that shown here.

[0036] Figure 6 This is a dual-path adaptive language control method based on a semantic bridging structure driven according to an embodiment of the present invention, such as... Figure 6 As shown, the method includes the following steps:

[0037] Step S602: Obtain the native language text input by the user, and perform anchor word recognition and phrase structure parsing on the native language text to generate target language training data corresponding to the native language text.

[0038] In this embodiment, the user inputs native language text through a learning terminal or language control terminal. The learning terminal or language control terminal can be a mobile learning device, tablet terminal, personal computer, smart earpiece, or other terminal device with voice interaction capabilities. The native language text can originate from sentences entered by the user, paragraphs from reading materials, classroom recordings, or text data imported from a third-party teaching system.

[0039] Upon receiving native language text, the system first performs corpus preprocessing. This preprocessing includes text normalization, sentence segmentation, and pause location identification, resulting in a well-structured sentence sequence. This preprocessing makes subsequent semantic analysis more stable and avoids parsing errors caused by differences in text format.

[0040] After obtaining the standardized native language text, the system performs anchor word identification processing. Anchor words are words that play a core semantic role in the semantic structure of a sentence, such as key nouns, action verbs, or core functional words expressing semantic relationships. The system uses a natural language processing model to analyze the semantic importance of the text and identifies a set of semantic anchor words by combining statistical features from a language corpus. For example, in sentences describing events, action verbs are often identified as the main anchors, while in sentences describing object attributes, core nouns or adjectives may become semantic anchors.

[0041] After identifying the set of semantic anchor words, phrase structure parsing is performed on the native language text. A syntactic analysis model is used to identify dependency relationships within sentences, thereby obtaining semantic connections between words. For example, in a complex sentence, the system can identify the logical relationship between the main clause and subordinate clauses, and clarify semantic levels such as subject-verb structure, verb-object structure, and modification relationships. Through this process, the system can obtain complete semantic structure relationship data.

[0042] After obtaining the set of semantic anchor words and semantic structure relationship data, the system performs target language semantic transfer processing to generate target language training data. In traditional language learning systems, target language translation text is typically generated directly. In this embodiment, the system employs an improved method—the anchor semantic bridging sequence generation method—when generating target language training data. Specifically, the system first determines the semantic transfer path based on the set of semantic anchor words and constructs a semantic bridging sequence between the native language text and the target language text. This bridging sequence consists of multiple semantic transition segments, each revolving around a semantic anchor. For example, when an anchor word in the native language text is mapped to its corresponding expression in the target language, the system generates a transition sequence in the training data consisting of "native language expression—mixed expression—target language expression." In this way, the generated target language training data is no longer a single translation text but forms training content with a progressive structure. For example, during training, the system can first present a portion of the structure in the native language sentence and then gradually introduce the target language expression, allowing learners to gradually adapt to the grammatical structure of the target language during reading or shadowing.

[0043] Furthermore, to ensure the semantic stability of the bridging sequence, consistency checks are performed on different candidate expressions during the generation of transition segments. When multiple target language expressions may correspond to the same semantic anchor point, these candidate expressions are filtered using semantic structure relationship data to select the expression that best fits the sentence context. This process avoids expression conflicts or semantic shifts during semantic bridging.

[0044] Finally, the generated semantic bridging sequences are organized into target language training data. This training data not only includes target language text, but may also include audio synthesis parameters or speech audio data corresponding to the text.

[0045] Step S604: Collect the user's status signal on the learning terminal or language control terminal, extract status features based on the status signal, and generate a user training status vector based on the status features. The status signal includes at least one of the following: headphone connection status, voice recognition activity, page scrolling trajectory, and reading time.

[0046] After generating target language training data, the system senses the user's current learning state. The learning terminal or language control terminal collects various state signals related to learning behavior through a state feature extraction unit, enabling the system to select an appropriate training mode based on the actual environment.

[0047] In this embodiment, the hardware status of the terminal device is first detected, such as the headphone connection status and microphone availability status. When the system detects that the headphones are connected, it can infer that the user has an audio output environment, and when the microphone is available, it indicates that the user has the conditions for voice input. This device status information can provide the necessary basic conditions for subsequent shadowing training.

[0048] In addition to device status, the system also collects user behavior data within the learning interface. For example, page scrolling patterns can reflect the rhythm of a user's text browsing. If a user scrolls slowly and steadily while reading, it may indicate that they are reading attentively; while scrolling rapidly may indicate that they are simply browsing. Furthermore, the system records the time a user spends reading on each paragraph, thus obtaining reading duration characteristics.

[0049] The system can also monitor voice input signals in the environment through a voice recognition module and calculate voice recognition activity accordingly. When the system detects continuous voice input behavior, it can infer that the user is engaging in voice interaction or reading practice.

[0050] After acquiring these status signals, the system processes different types of signals uniformly using feature extraction algorithms. For example, headphone connection status can be encoded as device features, speech recognition activity can be converted into speech behavior features, and page scrolling trajectory and reading duration can be converted into reading behavior features. In this way, data from different sources are transformed into a unified feature representation.

[0051] The system then fuses these features to generate a training state vector. This vector can be viewed as a multidimensional representation of the user's current learning state, containing reading preference features, voice interaction features, and device availability information.

[0052] Step S606: Based on the training state vector and the target language training data, perform pattern judgment to determine the training mode, wherein the training mode includes an input immersion path and an output calibration path.

[0053] In this embodiment, the pathway control unit uses the training state vector as the main input and combines it with the structural features of the target language training data to comprehensively judge the current training mode. First, it analyzes the behavioral features in the training state vector, such as reading behavior features and speech behavior features, to determine whether the user currently prefers visual reading or speech interaction.

[0054] At the same time, the system also analyzes the structural features of the target language training data. For example, when the training data contains long text paragraphs, the system tends to choose the input immersion pathway; while when the training data mainly consists of short sentences or dialogue structures, the system tends to choose the output calibration pathway.

[0055] By comprehensively analyzing user behavior characteristics and training data structure characteristics, the pathway control unit can output pattern judgment results and determine the current training mode.

[0056] Step S608: When the mode judgment result is the input immersion path, the target language reading interface is presented and the target language training data is displayed on the interface; when the mode judgment result is the output calibration path, the target language training data is played.

[0057] In this embodiment, when the system determines that the current training mode is the input immersion pathway, the learning terminal or language control terminal will enter the input immersion pathway unit. This module will present a specially designed target language reading interface and display the target language training data in the interface. Since the training data contains semantic bridging sequences, the reading interface can gradually present the text content according to the semantic transition order, enabling learners to gradually complete the transition from their native language to the target language during the reading process.

[0058] When the pattern determination result indicates the output calibration path is selected, the system will call the output calibration path unit to play the audio content corresponding to the target language training data. The learning terminal or language control terminal outputs standard pronunciation audio through a speaker or headphones, and the user synchronously repeats after hearing the audio, thereby completing the speech training process.

[0059] After the above training mode is executed, some embodiments may also include training feedback and behavior recording steps to dynamically optimize subsequent training strategies based on the user's actual learning performance.

[0060] 1) Behavioral records

[0061] When the learning terminal or language control terminal enters the output calibration path and plays the audio content corresponding to the target language training data, the terminal device's voice acquisition module is simultaneously activated to acquire the user's repeating speech. The acquired voice signal first undergoes basic voice processing, such as noise suppression, voice segmentation, and audio energy normalization, to obtain a well-structured sequence of voice segments.

[0062] After speech preprocessing, speech recognition processing is performed on the user's reading aloud, generating a reading text sequence corresponding to the target language training data. Simultaneously, temporal features related to the reading rhythm are extracted, such as the duration of speech segments, the pause intervals between adjacent speech segments, and the trend of speech intensity changes. Analyzing these speech features yields the temporal structure information of the user's reading behavior.

[0063] Based on this, the system performs positional matching between the read-aloud text sequence and the target language training data to determine the rhythm changes that occur during the user's reading. For example, when the user makes abnormal pauses or jumps between certain words, the system can identify the corresponding rhythm shifts. Through this process, reading behavior data containing reading rhythm information, pause distribution information, and repetitive reading behavior information can be generated.

[0064] 2) Generate feedback and optimize based on feedback.

[0065] Feedback processing is performed based on the reading behavior data. By analyzing the rhythm jumps, pause distribution, and repetition frequency in the user's reading, training feedback results are generated, and the training rhythm and content display method are dynamically adjusted.

[0066] Specifically, the first step is to analyze the user's reading behavior. Traditional language learning systems typically focus only on reading accuracy or speech recognition results, while in real-world learning scenarios, reading rhythm often more accurately reflects the learner's level of comprehension. For example, when learners frequently pause at certain sentence structures, it often indicates that they have not yet fully grasped the grammatical rules of that structure. This embodiment employs an improved reading rhythm structure reconstruction method. This method first constructs a user reading rhythm sequence based on reading behavior data. This rhythm sequence consists of multiple time segments, each corresponding to a phonetic unit or word unit, and records the reading duration of that unit and the pause intervals between it and the units before and after it.

[0067] Subsequently, the reading rhythm sequence undergoes structural reconstruction. Based on the standard speech rhythm structure in the target language training data, alignment analysis is performed on the user's reading rhythm sequence to identify rhythm jump positions, abnormal pause positions, and repeated reading positions. For example, when a user makes a significant time interval between consecutive word groups, this position may be identified as a semantic comprehension difficulty point; while when a certain speech unit is repeatedly read, it may indicate that there is difficulty in pronouncing that unit.

[0068] The above analysis yielded a set of rhythmic structure parameters reflecting the characteristics of user reading behavior, including rhythmic jump frequency, pause distribution pattern, and repetition frequency. Training feedback results were generated based on these parameters.

[0069] After generating training feedback results, the training strategy is dynamically adjusted based on these results. For example, when the system detects that the user frequently pauses at certain points in a sentence structure, the playback speed of that sentence can be reduced and segmented prompts added in subsequent training sessions; and when the user repeats certain words, pronunciation examples or reinforcement exercises for those words can be added in the next round of training.

[0070] In this way, the training pace and content display can be continuously adjusted according to the user's actual reading performance, making the learning process more in line with the individual's learning pace, thereby improving the adaptive scheduling efficiency of the language control process.

[0071] Through the above steps, this embodiment realizes a complete processing flow from native language text input, semantic bridging training data generation, user state recognition to training mode execution, thereby constructing a language control method that can drive path scheduling based on training state vectors.

[0072] Figure 7 A schematic diagram of a computer device suitable for implementing embodiments of the present disclosure is shown. It should be noted that... Figure 7 The computer device shown is merely an example and should not be construed as limiting the functionality and scope of use of the embodiments disclosed herein.

[0073] like Figure 7 As shown, the computer device includes a central processing unit (CPU) 1001, which can perform various appropriate actions and processes according to a program stored in a read-only memory (ROM) 1002 or a program loaded from a storage section 1008 into a random access memory (RAM) 1003. The RAM 1003 also stores various programs and data required for system operation. The CPU 1001, ROM 1002, and RAM 1003 are interconnected via a bus 1004. An input / output (I / O) interface 1005 is also connected to the bus 1004.

[0074] The following components are connected to I / O interface 1005: an input section 1006 including a keyboard, mouse, etc.; an output section 1007 including a cathode ray tube (CRT), liquid crystal display (LCD), etc., and speakers, etc.; a storage section 1008 including a hard disk, etc.; and a communication section 1009 including a network interface card such as a LAN card, modem, etc. The communication section 1009 performs communication processing via a network such as the Internet. A drive 1010 is also connected to I / O interface 1005 as needed. A removable medium 1011, such as a disk, optical disk, magneto-optical disk, semiconductor memory, etc., is installed on drive 1010 as needed so that computer programs read from it can be installed into storage section 1008 as needed.

[0075] The above description is only a preferred embodiment of this application. It should be noted that for those skilled in the art, several improvements and modifications can be made without departing from the principle of this application, and these improvements and modifications should also be considered within the scope of protection of this application.

Claims

1. A dual-path adaptive language control system driven by a semantic bridging structure, characterized in that, include: The semantic bridging generation unit is used to identify anchor words and parse the structure of native language text to construct a semantic bridging sequence data structure. The state feature extraction unit is used to collect multimodal user state signals and generate training state vectors; The pathway control unit is used to perform scheduling calculations based on the training state vector, semantic bridging sequence weight parameters and structural features of the current training data, and to perform adaptive scheduling between the cognitive input pathway and the cognitive output pathway through the pathway determination function. The input immersion pathway unit is used to perform visual and auditory input training based on semantic bridging sequences; The output calibration pathway unit is used to perform speech output comparison and rhythm synchronization calibration training; The feedback reinforcement unit is used to update the semantic bridging sequence weights and path scheduling parameters based on user behavior parameters. Behavioral modeling unit, used to record user training behavior and build rhythm structure model; The interface unit is used to enable two-way data interaction with external speech recognition systems, language engines, and teaching platforms; The path control unit is used to perform scheduling calculations based on the joint input of semantic bridging sequence data structure and training state vector through a preset path determination function, so as to uniquely determine the current training path and its rhythm parameters, wherein the scheduling result cannot be directly determined by a single rule or fixed conditions.

2. The system according to claim 1, characterized in that, The semantic bridging sequence is a structured representation sequence driven by semantic anchors. Each representation unit in the sequence is progressively mapped around a semantic anchor, and all training data must be generated through this bridging sequence; direct generation of target language representations or skipping intermediate transitional structures is not allowed. In some embodiments, the semantic bridging sequence can be represented as a sequence set S = {s0, s1, … , sn}, where s0 represents a native language representation segment, sn represents a target language representation segment, and intermediate sequences s1 to sn-1 represent transitional representation segments.

3. The system according to claim 1, characterized in that, The state feature extraction unit integrates headphone connection status, voice recognition activity, page scrolling trajectory, reading dwell time, and environmental noise level to generate a multi-dimensional training state vector.

4. The system according to claim 1, characterized in that, The path control unit performs scheduling operations based on the training state vector and the semantic bridging sequence weight matrix, through a path determination function, to determine the currently executed path type and its rhythm parameters.

5. The system according to claim 1, characterized in that, The feedback reinforcement unit extracts rhythmic structure parameters by analyzing rhythmic jumps, pause distribution, repetition frequency, and speech matching degree in the user's reading, and updates the semantic bridging sequence weights and path scheduling parameters based on the rhythmic structure parameters.

6. The system according to claim 1, characterized in that, The behavior modeling unit constructs an individual rhythm model based on the user's historical training data, and uses the rhythm model for subsequent path scheduling and semantic bridging parameter optimization.

7. The system according to claim 1, characterized in that, The interface docking unit supports interface docking and calling of SCORM, xAPI standard protocols and third-party speech recognition platforms, enabling cross-platform deployment and data synchronization.

Images