A cross-campus campus culture remote real-time interactive display and live broadcast method

Abstract

This invention discloses a method for remote real-time interactive display and live streaming of campus culture across campuses, relating to the fields of educational digitization and multimedia interaction technology. The method comprises six steps: unified access and multimodal standardized preprocessing of cross-campus cultural resources; generation of personalized virtual interactive objects based on campus attributes and live streaming themes; multimodal fusion-based real-time interactive behavior recognition and intelligent guidance across campuses; dynamic information hierarchy and multi-terminal adaptive display based on audience behavior profiles; AI-driven real-time generation and distribution of cross-campus characteristic virtual interactive resources; and multi-dimensional evaluation and closed-loop optimization of cross-campus live streaming interaction effects. This invention enables unified management and collaborative display of cultural resources across campuses, provides personalized virtual interactive experiences and multi-terminal adaptive displays, effectively enhances the depth of cross-campus interaction and the dissemination effect of campus culture, and promotes cultural exchange and integration between different campuses.

Description

Technical Field

[0001] This invention relates to the field of digital education and multimedia interactive technology, and more specifically to a method for remote real-time interactive display and live streaming of campus culture across campuses. Background Technology

[0002] With the deepening of digital transformation in education, the display and dissemination of campus culture is gradually shifting from a single offline format to a digital model that integrates online and offline elements. Currently, campus culture live streams mostly adopt a single-campus independent streaming approach, enabling cross-campus viewing through simple video streaming. Some platforms have introduced basic interactive functions such as bullet comments and likes. However, existing technologies suffer from a core problem of insufficient cross-campus interaction depth: on the one hand, cultural resources from different campuses are scattered and stored in their respective local systems, lacking unified access standards and management mechanisms, making it difficult to achieve collaborative display and real-time linkage of cultural content across campuses, resulting in high-quality cultural resources not being shared across the entire school; on the other hand, the interactive forms are singular and rigid, only supporting basic text bullet comments and general virtual gift sending, unable to generate personalized interactive content based on the unique history, culture, and educational characteristics of each campus, and also unable to provide precise interactive guidance based on the behavioral characteristics of viewers from different campuses, resulting in generally low participation from cross-campus viewers and significantly diminishing the effectiveness of campus culture dissemination. Furthermore, the existing live-streaming methods employ a uniform information layout, which cannot adapt to diverse display terminals such as smart interactive screens, multi-screen splicing systems, and mobile devices across different campuses. It also cannot dynamically adjust the information display hierarchy based on viewer behavior, easily leading to information overload or missing key content. This further impacts the experience of disseminating campus culture and hinders the depth of cross-campus cultural exchange and integration. Therefore, a solution is needed. Summary of the Invention

[0003] To address the shortcomings of existing technologies, this invention provides a method for remote real-time interactive display and live streaming of campus culture across campuses, in order to solve the problems mentioned in the background.

[0004] To achieve the above objectives, the present invention provides the following technical solution: A method for remote real-time interactive display and live streaming of campus culture across campuses, the method comprising the following steps: Step 1: Unified access and multimodal standardized preprocessing of cultural resources across campuses A unified access gateway for cultural resources covering all campuses will be established, supporting each campus to upload various cultural resources through a standardized RESTful interface. These resources include multimodal data such as campus history materials, student artworks, live campus activities, expert lectures, displays of educational achievements, and introductions to featured disciplines. All uploaded resources will be automatically classified and tagged. Optical character recognition technology will be used to extract text information from images, posters, and documents. Automatic speech recognition technology will be used to convert audio and video resources into structured text content. Keyword extraction and semantic analysis technology will be used to generate topic tags, campus tags, content feature tags, and applicable scenario tags for resources. A unified distributed cross-campus cultural resource database will be constructed, which will be structured and stored according to dimensions such as resource type, campus, topic, publication time, and access popularity. A semantic indexing mechanism based on a vector database will be established to support millisecond-level fast retrieval and retrieval based on keywords, natural language semantics, and campus attributes. At the same time, all resources will be standardized and converted to generate multi-resolution and multi-bitrate versions adapted to different display terminals to ensure smooth loading and display of resources on various terminals. Step 2: Generating personalized virtual interactive objects based on campus attributes and live stream theme Before the live stream begins, the system gathers information on the theme, participating campus attributes, and the target audience's basic characteristics. Campus attributes include the campus's history, cultural features, educational philosophy, landmark buildings, distinctive disciplines, and campus symbols. Target audience characteristics include age distribution, grade structure, interests, and past interaction habits. A pre-trained generative AI model is then used to generate virtual interactive objects with corresponding campus cultural characteristics, categorized into virtual guides and virtual audiences. The virtual guides' designs deeply integrate the campus's iconic elements, their voices match the campus's cultural atmosphere, and they have a built-in cultural knowledge base. They can explain the cultural background of each campus and related extended knowledge in real time. The number and type of virtual audiences are dynamically configured based on the size and characteristics of the participating campus's audience, simulating diverse interactive behaviors of real audiences, including sending comments, liking, giving virtual gifts, and initiating discussions. Each virtual interactive object is assigned a globally unique identifier, and a strong association is established between the virtual interactive object and the corresponding campus, facilitating subsequent interaction management and performance statistics. Step 3: Multimodal Fusion Cross-Campus Real-Time Interactive Behavior Recognition and Intelligent Guidance During the live stream, multimodal interaction data from all campus viewers was collected in real time, including bullet comments, voice interactions, likes, gift-giving, interface clicks, and viewer gaze data collected via a smart interactive screen. The collected multimodal data underwent low-latency real-time processing. Natural language processing technology was used to perform semantic analysis on bullet comments and voice content, extracting interactive keywords, sentiment, and topic direction. Eye-tracking technology was used to analyze the dwell time and movement trajectory of viewers' gazes in different areas of the live stream interface, accurately determining viewers' focus and interests. Behavioral analysis technology was used to statistically analyze the frequency, depth, and type of viewer interaction. The system processes multi-dimensional interactive data to identify potential interactive needs of viewers and the overall interactive activity of the live stream in real time. When the overall interactive activity is detected to be below a preset threshold, or when viewers have a clear interactive need but have not received a response, personalized interactive guidance information is generated. This information is closely integrated with the current live stream theme, the cultural content being presented, and the individual behavioral characteristics of the viewers. It guides viewers to interact with virtual interactive objects or real viewers from other campuses. The generated interactive guidance information is displayed as a floating pop-up on the live stream interface of the corresponding campus viewer, and the guidance content is highlighted with a high-brightness border and animation effects to ensure that viewers can clearly understand it. Step 4: Dynamic information hierarchy and multi-terminal adaptation display based on audience behavior profiles A real-time updated behavioral profile is created for each viewer entering the live stream. This profile includes dimensions such as cumulative viewing time, real-time interaction frequency, types of content viewed, type of device used, current network status, and historical viewing history. Based on these real-time behavioral profiles, an approximate KNN algorithm is employed. The algorithm automatically categorizes viewers into five viewing modes: fast browsing, long-stay mode, high interaction mode, low interaction mode, and device / network optimization mode. For each mode, the algorithm dynamically adjusts the information display hierarchy and content density of the live stream interface. For viewers in fast browsing mode, only the core live stream footage, title, and basic interactive controls are displayed, hiding unnecessary auxiliary information and reducing interface elements to avoid information overload. For viewers in long-stay mode, more cultural background information, related resource recommendations, interaction statistics from viewers across different campuses, and discussion topics are displayed. For viewers in high interaction mode, a detailed list of interaction information, real-time interaction dynamics from viewers across campuses, in-depth content analysis, and extended resource links are displayed. For viewers in low interaction mode, appropriate interactive guidance information and engaging cultural content are displayed to gently encourage audience participation. For viewers in device / network optimization mode, the display quality of high-resolution images and videos is automatically reduced, prioritizing the smoothness of live stream audio and core visuals. Furthermore, the algorithm adjusts the display quality based on the type of terminal used by the viewer, including smart interactive screens, multi-screen splicing display systems, and PCs. The system automatically adapts the layout, resolution, and interaction methods of the live broadcast interface to terminals, mobile terminals, and set-top boxes. For multi-screen splicing display systems, different modules such as the main live broadcast screen, interactive information area, cultural resource display area, and cross-campus audience dynamic area are allocated to different screen areas to achieve multi-screen collaborative display and maximize the use of display space. Step 5: Real-time generation and distribution of AI-driven cross-campus themed virtual interactive resources During the live stream, the system acquires real-time data on the content characteristics of the current live stream, the cultural features of participating campuses, and audience interaction behavior. It then utilizes an AI model pipeline comprised of text-to-text, text-to-image, and image-to-video models to generate virtual interactive resources with cross-campus characteristics. These resources include virtual gifts, virtual badges, virtual backgrounds, and interactive effects. The virtual gifts are deeply integrated with each campus's iconic buildings, campus symbols, distinctive academic elements, and cultural symbols. Viewers can send these virtual gifts to the streamer or viewers from other campuses. Virtual badges are automatically generated based on viewer interaction and participation in campus activities, serving as digital mementos of cross-campus cultural interaction. Each generated virtual interactive resource is assigned a globally unique serial number tag. The tag encoding rules include the campus information, generation timestamp, and sequence number, ensuring the uniqueness and traceability of the virtual resources. A real-time distribution mechanism for virtual interactive resources is established, utilizing CDN. The generated resources will be pushed synchronously to all live streaming terminals in participating campuses, allowing viewers to view and use them in real time on the live streaming interface. It also provides personal collection and social sharing functions for virtual resources. Viewers can collect their favorite virtual resources to their personal center or share them to mainstream social platforms, further expanding the reach of campus culture. Step 6: Multi-dimensional evaluation and closed-loop optimization of cross-campus live streaming interaction effects After the live stream ends, all data from the event is automatically collected, including the number of viewers from each campus, peak online viewers, cumulative viewing time, interaction frequency distribution, complete set of bullet comments, virtual resource usage, audience satisfaction survey results, and technical operation logs. A scientific evaluation index system is constructed based on four core dimensions: cultural dissemination effect, interactive participation, content adaptability, and technical stability. The cultural dissemination effect dimension includes the percentage of time each campus displays cultural content, audience attention to different campus cultural content, and the dissemination scope and sharing frequency of virtual resources. The interactive participation dimension includes the total number of cross-campus interactions, the average interaction time of viewers, the percentage of viewers in high-interaction mode, and the interaction percentage between viewers from different campuses. The content adaptability dimension includes the average time viewers spend in different viewing modes. The evaluation criteria included duration, number of times information hierarchy adjustments were triggered, and audience satisfaction with multi-terminal display. Technical stability dimensions included overall live stream stuttering rate, average resource loading speed, interactive response latency, and system failure rate. Based on the evaluation index system, the effectiveness of this live stream was quantitatively evaluated, generating a detailed evaluation report containing data statistics, problem analysis, and improvement suggestions. According to the evaluation results, problems and shortcomings in the live stream were identified, and targeted optimizations were made to the recommendation strategy for cross-campus cultural resources, the generation parameters of virtual interactive objects, the triggering mechanism for interactive guidance, and the hierarchical rules for information display. At the same time, the evaluation data was fed back to the cross-campus cultural resource database to update the tag weights and recommendation priorities of resources, providing more accurate resource support for subsequent live streams and cultural displays, forming a continuous iterative optimization closed-loop mechanism.

[0005] As a preferred embodiment of the present invention: the generation of virtual interactive objects in step 2 supports customized configuration by campus administrators. Each campus administrator can adjust the image characteristics, voice tone and exclusive explanation content of the virtual guide, as well as the frequency and preferences of the virtual audience's interactive behavior according to actual needs. At the same time, the virtual interactive objects can respond to the switching of live broadcast content in real time. When the live broadcast content is switched from one campus to another, it automatically and seamlessly switches to the virtual guide of the corresponding campus, ensuring a high degree of consistency between the explanation content and the display content.

[0006] As a preferred embodiment of the present invention: the generation of interactive guidance information in step 3 is also combined with the real-time status of the virtual interactive object. When the virtual interactive object sends simulated interactive information, feedback guidance information is automatically generated for the simulated information to guide the real audience to respond. In addition, when it is detected that there are common interest topics among the audiences of different campuses, cross-campus topic guidance information is generated to promote in-depth communication and interaction among the audiences of different campuses.

[0007] As a preferred embodiment of the present invention: In step 4, the multi-terminal adaptation display supports cross-terminal synchronous control. The host can uniformly adjust the content layout and display focus of all campus display terminals through the main control terminal, and can also authorize each campus administrator to independently adjust the display content of their own campus terminal. For smart interactive screens, viewers can directly interact with the live broadcast content through touch operation, and realize the zooming, rotation and free switching of cultural resources.

[0008] As a preferred embodiment of the present invention: In step 5, the generation of virtual interactive resources supports audience participation in customization. Audiences can select different element combinations from the characteristic element library of each campus provided by the system to generate personalized virtual gifts and virtual backgrounds. After automated content review, the generated personalized virtual resources are uploaded to the live broadcast interface in real time for all audiences to use. The usage data of virtual resources will be statistically analyzed in real time, serving as an important basis for evaluating the popularity of cultural elements in each campus.

[0009] As a preferred embodiment of the present invention: the weights of the evaluation index system in step 6 can be dynamically adjusted according to the type and objectives of the live broadcast. For live broadcasts with cultural dissemination as the main objective, the weight of the cultural dissemination effect dimension is increased. For live broadcasts with cross-campus communication as the main objective, the weight of the interactive participation dimension is increased. The evaluation report will generate separate analysis results for each participating campus, providing data support for the subsequent cultural activity planning and resource construction of each campus.

[0010] Compared with the prior art, the beneficial effects of the present invention are as follows:

[0011] This technical solution completely solves the problems of scattered cultural resources and difficulty in collaborative utilization in existing technologies by unifying access and standardizing preprocessing of cultural resources across campuses. It achieves centralized management and efficient sharing of high-quality cultural resources across campuses. Personalized virtual interactive objects generated based on campus attributes and live broadcast themes, combined with multimodal interactive behavior recognition and intelligent guidance, break the traditional one-way communication model of live broadcasts, effectively enhancing the depth and fun of cross-campus interactions, and significantly improving audience participation and retention rates. Dynamic information hierarchy based on audience behavior profiles and multi-terminal adaptive display can provide a personalized display experience tailored to the needs of different viewers and the terminals they use, avoiding information overload or missing key content, and greatly improving the dissemination effect of campus culture. AI-driven real-time generation and distribution of cross-campus characteristic virtual interactive resources deeply integrates the unique cultural characteristics of each campus into the interactive process, enhancing the audience's cultural identity and sense of belonging, while expanding the external dissemination scope of campus culture through social sharing functions. A multi-dimensional live broadcast effect evaluation and closed-loop optimization mechanism can continuously improve the quality of live broadcasts and the effect of cultural dissemination, forming a virtuous cycle. This solution not only enables remote real-time interactive display and live streaming of campus culture across campuses, enhancing the digital level of campus culture dissemination, but also more effectively promotes cultural exchange and integration between different campuses, providing an innovative and feasible solution for campus culture construction in the context of digital transformation of education. Attached Figure Description

[0012] Figure 1 This is a flowchart of the present invention. Detailed Implementation

[0013] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. 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 are within the scope of protection of the present invention.

[0014] Implementation Example (Basic Implementation Example)

[0015] 1. System Configuration and Implementation Environment

[0016] This embodiment is based on the actual scenario of three campuses (main campus, east campus, and west campus) of a comprehensive university. The hardware system includes: 12 86-inch smart interactive screens, 3 sets of 3×3 multi-screen splicing display systems, 6 projection control terminals, 3 live streaming servers, and 9 edge computing nodes deployed in each campus. The software system includes: a unified access gateway for cross-campus cultural resources, a distributed vector database (Milvus 2.3), a pre-trained generative large language model (Llama 3-70B), a text-to-graph model (Stable Diffusion XL), a graph-to-video model (AnimateDiff), and a multimodal behavior analysis engine.

[0017] 2. Specific Implementation Steps

[0018] Step 1: Unified access and multimodal standardized preprocessing of cultural resources across campuses

[0019] A unified access gateway based on RESTful API was established to support the uploading of cultural resources from various campuses via web and mobile devices. The more than 12,000 uploaded multimodal resources (including 3,200 historical campus photos, 4,500 student artworks, 2,800 campus activity videos, and 1,500 audio recordings of expert lectures) were processed automatically.

[0020] Text information is extracted from images and documents using OCR technology, with a recognition accuracy rate of 98.7%.

[0021] ASR technology is used to convert audio and video resources into structured text, with a word error rate of less than 2.3%.

[0022] The BERT-based model was used for keyword extraction and semantic analysis, generating 126 types of tags in 4 categories: topic tags, campus tags, content feature tags, and applicable scenario tags.

[0023] Construct a distributed cultural resource database, establish a semantic index based on a vector database, and achieve a retrieval response time of less than 50ms;

[0024] All video resources are transcoded to generate 1080p, 720p, and 480p resolution versions to adapt to different terminals and network environments.

[0025] Step 2: Generating personalized virtual interactive objects based on campus attributes and live stream theme

[0026] The theme of this live stream is "A Century of School History and Campus Charm," featuring the main campus, the east campus, and the west campus. Viewers will gain access to information about each campus: Main Campus (century-old history, red-brick architecture, arts and sciences focus), East Campus (modern architecture, engineering focus, robotics lab), and West Campus (garden-style architecture, medical focus, affiliated hospital).

[0027] The Llama 3-70B model was used to generate virtual tour guides for three campuses, with their images incorporating iconic elements such as the clock tower of the main campus, the robot of the east campus, and the ginkgo leaves of the west campus.

[0028] The virtual tour guide's voice is generated by cloning the voice of alumni from the corresponding campus, and it has a built-in cultural knowledge database of more than 100,000 words from each campus.

[0029] 150 virtual viewers are dynamically configured, including 60 in the main campus, 50 in the east campus, and 40 in the west campus, simulating the behavior of viewers of different ages and interests.

[0030] Assign a globally unique ID to each virtual interactive object, in the format of VOBJ-campus code-timestamp-serial number.

[0031] Step 3: Multimodal Fusion Cross-Campus Real-Time Interactive Behavior Recognition and Intelligent Guidance

[0032] Real-time collection of multimodal interaction data from audiences across all campuses, including bullet comments, voice messages, likes, gift-giving, interface clicks, and eye-tracking data from smart interactive screens:

[0033] By using NLP technology to perform semantic analysis on bullet comments and audio, keywords and sentiment tendencies were extracted, achieving a sentiment classification accuracy of 92.4%.

[0034] The eye-tracking technology was used to analyze the area where the viewer's gaze lingered, with a sampling frequency of 30Hz and a gaze positioning error of less than 1.5°.

[0035] When the overall interaction intensity is detected to be lower than the preset threshold (interactions per minute < 20 times), or when it is identified that the audience has a clear need for interaction but has not responded, personalized guidance information is generated;

[0036] The guidance information is displayed as a floating pop-up window with a highlighted border and a fade-in animation effect, lasting for 5 seconds to ensure that the audience can clearly understand it.

[0037] Step 4: Dynamic information hierarchy and multi-terminal adaptation display based on audience behavior profiles

[0038] A real-time behavioral profile is created for each viewer entering the live stream, and an approximate KNN algorithm (K=5) is used to divide viewers into 5 viewing modes:

[0039] Quick browsing mode: Only the main live stream screen, title, and basic interactive controls are displayed, while unnecessary information is hidden;

[0040] Long-stay mode: Adds cultural background information, relevant resource recommendations, and interaction statistics;

[0041] Highly interactive mode: Displays a detailed list of interactions, cross-campus audience dynamics, and in-depth content analysis;

[0042] Low-interaction mode: Displays moderately guiding information and engaging cultural content;

[0043] Device and network optimization mode: Automatically reduce video resolution to 480p, prioritizing smooth audio playback.

[0044] For multi-screen splicing systems, the main live broadcast screen, interactive area, resource display area, and campus dynamic area are allocated to different screen areas to achieve multi-screen collaboration.

[0045] Step 5: Real-time generation and distribution of AI-driven cross-campus themed virtual interactive resources

[0046] The AI ​​model pipeline is used to generate virtual interactive resources in real time.

[0047] Text-to-text models generate object descriptions, text-to-image models generate resource icons, and image-to-video models generate 3-5 second animation effects.

[0048] The virtual gifts incorporate iconic elements from each campus, such as fireworks at the main campus clock tower, robot dances at the east campus, and falling ginkgo leaves at the west campus.

[0049] Assign a globally unique serial number label to each virtual resource, in the format VRES-Campus Code-Timestamp-Serial Number;

[0050] The generated resources are synchronously pushed to all campus terminals via CDN network, with a loading time of less than 200ms;

[0051] Viewers can select and combine elements from a library of campus-specific elements to generate personalized virtual gifts, which are then automatically reviewed and uploaded in real time.

[0052] Step 6: Multi-dimensional evaluation and closed-loop optimization of cross-campus live streaming interaction effects

[0053] After the live stream ends, all data is automatically collected, and an evaluation indicator system is constructed from four dimensions:

[0054] Cultural dissemination effectiveness: Percentage of content display time in each campus, audience attention, and number of virtual resource shares;

[0055] Interactive engagement: Total number of interactions across campuses, average interaction duration, and percentage of viewers using high-interaction mode;

[0056] Content adaptability: Viewer dwell time and number of times information hierarchy adjustments are triggered in different viewing modes;

[0057] Technical stability: live streaming buffering rate, resource loading speed, and interactive response latency.

[0058] Detailed evaluation reports are generated to optimize resource recommendation strategies, virtual interactive object parameters, and guidance triggering mechanisms in a targeted manner, forming a closed-loop optimization.

[0059] 3. Performance Test Results

[0060] The live stream lasted for 2 hours, with a total of 1,562 viewers and a peak of 896 concurrent viewers. Key performance indicators: 1,286 cross-campus interactions, an average interaction time of 18.7 minutes, 28.3% of viewers in high-interaction mode, 94.2% accuracy in information hierarchy matching, 76.5% average attention to cultural content, an overall stream buffering rate of 0.8%, and a viewer satisfaction rating of 4.6 / 5.0.

[0061] Example 1 (Feature Missing Comparison Group: Virtual Interactive Object Generation Steps Omitted)

[0062] 1. System Configuration and Implementation Environment

[0063] Completely identical to the basic implementation, the test sample consisted of another 500 randomly assigned viewers, with the same live stream topic, content, and duration.

[0064] 2. Specific Implementation Steps

[0065] Except for step 2 (generating personalized virtual interactive objects based on campus attributes and live stream theme), steps 1, 3-6 are completely consistent with the basic implementation. No virtual narrators or virtual audiences are generated during the live stream; only the interactive functions of real audiences are retained.

[0066] 3. Performance Test Results

[0067] Key performance indicators: 324 cross-campus interactions, 7.2 minutes average interaction time, 8.7% of viewers in high-interaction mode, 93.8% accuracy rate in information hierarchy adaptation, 42.3% average attention to cultural content, 0.7% overall live stream stuttering rate, and 3.2 / 5.0 viewer satisfaction rating.

[0068] Comparative Example 1 (corresponding to Example 1)

[0069] 1. System Configuration and Implementation Environment

[0070] Completely consistent with Example 1.

[0071] 2. Specific Implementation Steps

[0072] This is completely consistent with Example 1, that is, the scheme that omits the step of generating virtual interactive objects.

[0073] 3. Comparative Analysis and Conclusions

[0074] Compared with the basic implementation, Comparative Example 1 showed a 74.8% decrease in the total number of cross-campus interactions, a 61.5% decrease in the average interaction time, a 69.3% decrease in the proportion of viewers in the high-interaction mode, a 44.7% decrease in the average attention to cultural content, and a 30.4% decrease in the viewer satisfaction score.

[0075] Conclusion: The generation of personalized virtual interactive objects based on campus attributes and live broadcast themes is an essential technical feature of this invention, which can significantly enhance the depth and fun of cross-campus interactions and strengthen the audience's sense of participation and cultural identity.

[0076] Example 2 (Parameter Out-of-Range Comparison Group: KNN Algorithm K Value Out of Range)

[0077] 1. System Configuration and Implementation Environment

[0078] Completely identical to the basic implementation, the test sample consisted of another 500 randomly assigned viewers, with the same live stream topic, content, and duration.

[0079] 2. Specific Implementation Steps

[0080] Except for step 4, where the K value of the approximate KNN algorithm is set to 1 and 15 respectively, the remaining steps are completely consistent with the basic embodiment. The system performance was tested under the two parameters K=1 and K=15 respectively.

[0081] 3. Performance Test Results

[0082] When K=1: Information hierarchy adaptation accuracy rate is 72.6%, average audience dwell time is 11.3 minutes, information hierarchy adjustment triggers 187 times / hour, and audience satisfaction rating is 3.7 / 5.0;

[0083] When K=15: Information level adaptation accuracy rate is 78.9%, average audience dwell time is 12.5 minutes, information level adjustment is triggered 152 times / hour, and audience satisfaction score is 3.9 / 5.0.

[0084] Comparative Example 2 (corresponding to Example 2)

[0085] 1. System Configuration and Implementation Environment

[0086] Completely consistent with Example 2.

[0087] 2. Specific Implementation Steps

[0088] It is completely consistent with Example 2, that is, using parameter settings of K=1 and K=15.

[0089] 3. Comparative Analysis and Conclusions

[0090] Compared to the basic implementation (K=5, information level adaptation accuracy of 94.2%), the accuracy decreased by 23.0% when K=1 and by 16.2% when K=15. A K value that is too small will cause the viewing mode classification to be too sensitive, resulting in frequent switching of information levels and causing confusion for the audience; a K value that is too large will cause the classification to be too coarse and unable to accurately match the personalized needs of the audience.

[0091] Conclusion: The K=5 specified in this invention is the key parameter for achieving optimal information hierarchy adaptation, and can achieve the best balance between classification accuracy and stability.

[0092] Example 3 (Priority Technology Comparison Group: Traditional Single-Campus Distribution Solution)

[0093] 1. System Configuration and Implementation Environment

[0094] The hardware system is consistent with the basic implementation, while the software system adopts a traditional live streaming platform solution, which only supports independent streaming from a single campus, basic text bullet comments, and general virtual gift giving. It lacks unified cultural resource management, virtual interactive objects, dynamic information hierarchy, and AI-generated virtual resource functions.

[0095] 2. Specific Implementation Steps

[0096] The traditional live streaming process is adopted: each of the three campuses broadcasts independently, and cross-campus viewing is achieved through a simple video link. Viewers can only send text comments and send virtual gifts that are universal to the platform. The live streaming interface adopts a uniform and fixed layout and does not support dynamic information adjustment or multi-terminal adaptation.

[0097] 3. Performance Test Results

[0098] Key performance indicators: 187 cross-campus interactions, 5.8 minutes average interaction time, 5.2% of viewers in high-interaction mode, 31.7% average attention to cultural content, 1.5% overall live stream buffering rate, and 2.9 / 5.0 viewer satisfaction rating.

[0099] Comparative Example 3 (corresponding to Example 3)

[0100] 1. System Configuration and Implementation Environment

[0101] Completely consistent with Example 3.

[0102] 2. Specific Implementation Steps

[0103] Completely consistent with Example 3, namely adopting the traditional single-campus independent live broadcast solution.

[0104] 3. Comparative Analysis and Conclusions

[0105] Compared to the basic implementation, Comparative Example 3 showed an 85.5% decrease in the total number of cross-campus interactions, a 69.0% decrease in the average interaction time, an 81.6% decrease in the proportion of viewers in highly interactive mode, a 58.6% decrease in the average attention to cultural content, and a 37.0% decrease in viewer satisfaction ratings. Traditional solutions suffer from core problems such as dispersed cross-campus resources, limited interaction formats, and rigid presentation methods, failing to meet the needs of in-depth cultural exchange.

[0106] Conclusion: This invention, through technological innovations such as unified management of cross-campus resources, personalized virtual interaction, dynamic information display, and AI-generated distinctive resources, comprehensively surpasses existing technical solutions and can significantly enhance the dissemination effect of campus culture and the depth of cross-campus interaction.

[0107] Core performance comparison data

[0108] Total number of cross-campus interactions 1286 324 1123 1157 187 Average audience interaction time (minutes) 18.7 7.2 14.5 15.2 5.8 Percentage of viewers using the highly interactive mode (%) 28.3 8.7 21.6 23.1 5.2 Information hierarchy adaptation accuracy (%) 94.2 93.8 72.6 78.9 - Average attention rate for cultural content (%) 76.5 42.3 68.7 70.2 31.7 Overall buffering rate of live stream (%) 0.8 0.7 0.9 0.8 1.5 Audience satisfaction rating (out of 5) 4.6 3.2 3.7 3.9 2.9

[0109] Data Validity Statement

[0110] All tests were conducted in the same network environment (gigabit campus network) and under the same hardware conditions to eliminate the influence of system differences;

[0111] Each example and comparative example was repeated three times, and the data were averaged. The statistical significance test showed that P < 0.05, indicating that the results were statistically significant.

[0112] The audience sample consisted of randomly assigned students, with a uniform distribution in age, grade, and major to avoid sample bias.

[0113] Cultural content attention is calculated using eye-tracking data and is defined as the proportion of the total viewing time to the cumulative time the viewer's gaze lingers in the cultural content display area.

[0114] In summary, the cross-campus remote real-time interactive display and live broadcast method for campus culture proposed in this invention systematically solves the key problems existing in the prior art, such as the dispersion of cross-campus cultural resources, insufficient depth of interaction, single display method, and poor personalized experience, through six core technological innovations.

[0115] First, the unified access and multimodal standardized preprocessing mechanism for cultural resources across campuses enables centralized management and campus-wide sharing of high-quality cultural resources from each campus, solving the problem of resource silos and providing a solid data foundation for subsequent personalized displays and interactions. Semantic indexing technology based on vector databases compresses resource retrieval response time to within 50ms, ensuring real-time requirements during live streaming.

[0116] Secondly, the technology for generating personalized virtual interactive objects based on campus attributes and livestream themes is a core breakthrough in enhancing the depth of interaction. By generating virtual guides and virtual audiences with campus cultural characteristics, the traditional one-way communication model of livestreaming is broken, creating an immersive interactive atmosphere for the audience. Comparative data shows that this technology can increase the total number of cross-campus interactions by more than 3 times, and the average interaction time of the audience by 1.6 times, significantly enhancing the audience's sense of participation and cultural identity.

[0117] Third, the multimodal fusion real-time interactive behavior recognition and intelligent guidance technology combines natural language processing, eye tracking, and behavior analysis to accurately identify the audience's potential needs and interaction states, achieving a shift from passive response to proactive guidance. Based on audience behavior profiles, the dynamic information hierarchy and multi-terminal adaptive display technology, using an optimized KNN algorithm (K=5), achieves accurate classification of viewing modes, providing a personalized display experience for different audiences. The information hierarchy adaptation accuracy reaches 94.2%, effectively avoiding information overload or missing key content.

[0118] Fourth, AI-driven technology for real-time generation and distribution of cross-campus virtual interactive resources deeply integrates the unique cultural symbols of each campus into the interactive process, enhancing the fun and cultural connotation of the interaction. A globally unique serial number tagging mechanism ensures the traceability of virtual resources, providing data support for subsequent evaluation of cultural dissemination effectiveness. A multi-dimensional live-streaming effectiveness evaluation and closed-loop optimization mechanism can continuously improve live-streaming quality and cultural dissemination effects, forming a virtuous cycle.

[0119] Compared to traditional single-campus live streaming solutions, this invention increases the total number of cross-campus interactions by 5.9 times, the average audience interaction time by 2.2 times, the average attention to cultural content by 1.4 times, and the audience satisfaction rating by 58.6%. This solution not only enables remote, real-time interactive display and live streaming of campus culture across campuses, enhancing the digitalization of campus culture dissemination, but also effectively promotes cultural exchange and integration between different campuses, providing a practical and innovative solution for campus culture construction in the context of digital transformation in education.

[0120] The foregoing has shown and described the basic principles, main features, and advantages of the present invention. It will be apparent to those skilled in the art that the present invention is not limited to the details of the exemplary embodiments described above, and that the invention can be implemented in other specific forms without departing from its spirit or essential characteristics. Therefore, the embodiments should be considered illustrative and non-limiting in all respects, and the scope of the invention is defined by the appended claims rather than the foregoing description. Thus, all variations falling within the meaning and scope of equivalents of the claims are intended to be included within the present invention. No reference numerals in the claims should be construed as limiting the scope of the claims.

[0121] Finally, the following points should be noted: First, in the description of this application, it should be noted that, unless otherwise specified and limited, the terms "installation", "connection", and "linkage" should be interpreted broadly, and can refer to mechanical or electrical connections, or internal connections between two components, or direct connections. "Up", "down", "left", "right", etc., are only used to indicate relative positional relationships. When the absolute position of the described object changes, the relative positional relationship may change.

[0122] Finally, it should be noted that the above descriptions are merely preferred embodiments of the present invention and are not intended to limit the present invention. Although the present invention has been described in detail with reference to the foregoing embodiments, those skilled in the art can still modify the technical solutions described in the foregoing embodiments or make equivalent substitutions for some of the technical features. Any modifications, equivalent substitutions, improvements, etc., made within the spirit and principles of the present invention should be included within the protection scope of the present invention.

Claims

1. A method for remote real-time interactive display and live streaming of campus culture across campuses, characterized in that: The method includes the following steps: Step 1: Unified access and multimodal standardized preprocessing of cultural resources across campuses A unified access gateway for cultural resources covering all campuses will be established, supporting each campus to upload various cultural resources through a standardized RESTful interface. These resources include multimodal data such as campus history materials, student artworks, live campus activities, expert lectures, displays of educational achievements, and introductions to featured disciplines. All uploaded resources will be automatically classified and tagged. Optical character recognition technology will be used to extract text information from images, posters, and documents. Automatic speech recognition technology will be used to convert audio and video resources into structured text content. Keyword extraction and semantic analysis technology will be used to generate topic tags, campus tags, content feature tags, and applicable scenario tags for resources. A unified distributed cross-campus cultural resource database will be constructed, which will be structured and stored according to dimensions such as resource type, campus, topic, publication time, and access popularity. A semantic indexing mechanism based on a vector database will be established to support millisecond-level fast retrieval and retrieval based on keywords, natural language semantics, and campus attributes. At the same time, all resources will be standardized and converted to generate multi-resolution and multi-bitrate versions adapted to different display terminals to ensure smooth loading and display of resources on various terminals. Step 2: Generating personalized virtual interactive objects based on campus attributes and live stream theme Before the live stream begins, the system gathers information on the theme, participating campus attributes, and the target audience's basic characteristics. Campus attributes include the campus's history, cultural features, educational philosophy, landmark buildings, distinctive disciplines, and campus symbols. Target audience characteristics include age distribution, grade structure, interests, and past interaction habits. A pre-trained generative AI model is then used to generate virtual interactive objects with corresponding campus cultural characteristics, categorized into virtual guides and virtual audiences. The virtual guides' designs deeply integrate the campus's iconic elements, their voices match the campus's cultural atmosphere, and they have a built-in cultural knowledge base. They can explain the cultural background of each campus and related extended knowledge in real time. The number and type of virtual audiences are dynamically configured based on the size and characteristics of the participating campus's audience, simulating diverse interactive behaviors of real audiences, including sending comments, liking, giving virtual gifts, and initiating discussions. Each virtual interactive object is assigned a globally unique identifier, and a strong association is established between the virtual interactive object and the corresponding campus, facilitating subsequent interaction management and performance statistics. Step 3: Multimodal Fusion Cross-Campus Real-Time Interactive Behavior Recognition and Intelligent Guidance During the live stream, multimodal interaction data from all campus viewers was collected in real time, including bullet comments, voice interactions, likes, gift-giving, interface clicks, and viewer gaze data collected via a smart interactive screen. The collected multimodal data underwent low-latency real-time processing. Natural language processing technology was used to perform semantic analysis on bullet comments and voice content, extracting interactive keywords, sentiment, and topic direction. Eye-tracking technology was used to analyze the dwell time and movement trajectory of viewers' gazes in different areas of the live stream interface, accurately determining viewers' focus and interests. Behavioral analysis technology was used to statistically analyze the frequency, depth, and type of viewer interaction. The system processes multi-dimensional interactive data to identify potential interactive needs of viewers and the overall interactive activity of the live stream in real time. When the overall interactive activity is detected to be below a preset threshold, or when viewers have a clear interactive need but have not received a response, personalized interactive guidance information is generated. This information is closely integrated with the current live stream theme, the cultural content being presented, and the individual behavioral characteristics of the viewers. It guides viewers to interact with virtual interactive objects or real viewers from other campuses. The generated interactive guidance information is displayed as a floating pop-up on the live stream interface of the corresponding campus viewer, and the guidance content is highlighted with a high-brightness border and animation effects to ensure that viewers can clearly understand it. Step 4: Dynamic information hierarchy and multi-terminal adaptation display based on audience behavior profiles A real-time updated behavioral profile is created for each viewer entering the live stream. This profile includes dimensions such as cumulative viewing time, real-time interaction frequency, types of content viewed, type of device used, current network status, and historical viewing history. Based on these real-time behavioral profiles, an approximate KNN algorithm is employed. The algorithm automatically categorizes viewers into five viewing modes: fast browsing, long-stay mode, high interaction mode, low interaction mode, and device / network optimization mode. For each mode, the algorithm dynamically adjusts the information display hierarchy and content density of the live stream interface. For viewers in fast browsing mode, only the core live stream footage, title, and basic interactive controls are displayed, hiding unnecessary auxiliary information and reducing interface elements to avoid information overload. For viewers in long-stay mode, more cultural background information, related resource recommendations, interaction statistics from viewers across different campuses, and discussion topics are displayed. For viewers in high interaction mode, a detailed list of interaction information, real-time interaction dynamics from viewers across campuses, in-depth content analysis, and extended resource links are displayed. For viewers in low interaction mode, appropriate interactive guidance information and engaging cultural content are displayed to gently encourage audience participation. For viewers in device / network optimization mode, the display quality of high-resolution images and videos is automatically reduced, prioritizing the smoothness of live stream audio and core visuals. Furthermore, the algorithm adjusts the display quality based on the type of terminal used by the viewer, including smart interactive screens, multi-screen splicing display systems, and PCs. The system automatically adapts the layout, resolution, and interaction methods of the live broadcast interface to terminals, mobile terminals, and set-top boxes. For multi-screen splicing display systems, different modules such as the main live broadcast screen, interactive information area, cultural resource display area, and cross-campus audience dynamic area are allocated to different screen areas to achieve multi-screen collaborative display and maximize the use of display space. Step 5: Real-time generation and distribution of AI-driven cross-campus themed virtual interactive resources During the live stream, the system acquires real-time data on the content characteristics of the current live stream, the cultural features of participating campuses, and audience interaction behavior. It then utilizes an AI model pipeline comprised of text-to-text, text-to-image, and image-to-video models to generate virtual interactive resources with cross-campus characteristics. These resources include virtual gifts, virtual badges, virtual backgrounds, and interactive effects. The virtual gifts are deeply integrated with each campus's iconic buildings, campus symbols, distinctive academic elements, and cultural symbols. Viewers can send these virtual gifts to the streamer or viewers from other campuses. Virtual badges are automatically generated based on viewer interaction and participation in campus activities, serving as digital mementos of cross-campus cultural interaction. Each generated virtual interactive resource is assigned a globally unique serial number tag. The tag encoding rules include the campus information, generation timestamp, and sequence number, ensuring the uniqueness and traceability of the virtual resources. A real-time distribution mechanism for virtual interactive resources is established, utilizing CDN. The generated resources will be pushed synchronously to all live streaming terminals in participating campuses, allowing viewers to view and use them in real time on the live streaming interface. It also provides personal collection and social sharing functions for virtual resources. Viewers can collect their favorite virtual resources to their personal center or share them to mainstream social platforms, further expanding the reach of campus culture. Step 6: Multi-dimensional evaluation and closed-loop optimization of cross-campus live streaming interaction effects After the live stream ends, all data from the event is automatically collected, including the number of viewers from each campus, peak online viewers, cumulative viewing time, interaction frequency distribution, complete set of bullet comments, virtual resource usage, audience satisfaction survey results, and technical operation logs. A scientific evaluation index system is constructed based on four core dimensions: cultural dissemination effect, interactive participation, content adaptability, and technical stability. The cultural dissemination effect dimension includes the percentage of time each campus displays cultural content, audience attention to different campus cultural content, and the dissemination scope and sharing frequency of virtual resources. The interactive participation dimension includes the total number of cross-campus interactions, the average interaction time of viewers, the percentage of viewers in high-interaction mode, and the interaction percentage between viewers from different campuses. The content adaptability dimension includes the average time viewers spend in different viewing modes. The evaluation criteria included duration, number of times information hierarchy adjustments were triggered, and audience satisfaction with multi-terminal display. Technical stability dimensions included overall live stream stuttering rate, average resource loading speed, interactive response latency, and system failure rate. Based on the evaluation index system, the effectiveness of this live stream was quantitatively evaluated, generating a detailed evaluation report containing data statistics, problem analysis, and improvement suggestions. According to the evaluation results, problems and shortcomings in the live stream were identified, and targeted optimizations were made to the recommendation strategy for cross-campus cultural resources, the generation parameters of virtual interactive objects, the triggering mechanism for interactive guidance, and the hierarchical rules for information display. At the same time, the evaluation data was fed back to the cross-campus cultural resource database to update the tag weights and recommendation priorities of resources, providing more accurate resource support for subsequent live streams and cultural displays, forming a continuous iterative optimization closed-loop mechanism.

2. The method for remote real-time interactive display and live streaming of cross-campus campus culture according to claim 1, characterized in that: In step 2, the generation of virtual interactive objects supports customized configuration by campus administrators. Each campus administrator can adjust the virtual guide's appearance, voice tone, and exclusive explanation content, as well as the frequency and preferences of virtual audience interaction, according to actual needs. At the same time, the virtual interactive objects can respond to the switching of live broadcast content in real time. When the live broadcast content switches from one campus to another, it automatically and seamlessly switches to the virtual guide of the corresponding campus, ensuring a high degree of consistency between the explanation content and the displayed content.

3. The method for remote real-time interactive display and live streaming of cross-campus campus culture according to claim 1, characterized in that: In step 3, the generation of interactive guidance information also incorporates the real-time status of the virtual interactive object. When the virtual interactive object sends simulated interactive information, feedback guidance information is automatically generated for that simulated information to guide real viewers to respond. In addition, when it is detected that viewers from different campuses share common topics of interest, cross-campus topic guidance information is generated to promote in-depth communication and interaction between viewers from different campuses.

4. The method for remote real-time interactive display and live streaming of cross-campus campus culture according to claim 1, characterized in that: In step 4, the multi-terminal adaptation display supports cross-terminal synchronous control. The host can uniformly adjust the content layout and display focus of all campus display terminals through the main control terminal, or authorize each campus administrator to independently adjust the display content of their own campus terminal. For smart interactive screens, viewers can directly interact with the live broadcast content through touch operation, and realize the zooming, rotation and free switching of cultural resources.

5. The method for remote real-time interactive display and live streaming of cross-campus campus culture according to claim 1, characterized in that: In step 5, the generation of virtual interactive resources supports audience participation in customization. Audiences can select different element combinations from the library of characteristic elements of each campus provided by the system to generate personalized virtual gifts and virtual backgrounds. After automated content review, the generated personalized virtual resources are uploaded to the live broadcast interface in real time for all viewers to use. The usage data of virtual resources will be statistically analyzed in real time, serving as an important basis for evaluating the popularity of cultural elements in each campus.

6. The method for remote real-time interactive display and live streaming of campus culture across campuses according to claim 1, characterized in that: In step 6, the weights of the evaluation indicators can be dynamically adjusted according to the type and objectives of the live stream. For live streams with cultural dissemination as the main objective, the weight of the cultural dissemination effect dimension will be increased. For live streams with cross-campus exchanges as the main objective, the weight of the interactive participation dimension will be increased. The evaluation report will generate separate analysis results for each participating campus, providing data support for the subsequent planning of cultural activities and resource development for each campus.

Images