An Artificial Intelligence-Based Campus Smart Freshman Welcome System and Method

Abstract

The present application discloses a campus intelligent freshman welcoming system and method based on artificial intelligence. After a freshman creates a personal account and enters information, it receives the text descriptions of the student's hobbies and professional needs input by the student, and uses natural language processing technology based on deep learning to perform semantic parsing and feature fusion on the student's hobbies and professional needs to obtain a joint representation of the student's personalized needs. Furthermore, it extracts the club activity information of the school, and through fine-grained semantic interaction query between the club activity information and the student's personalized needs, to reveal the matching degree between the two, so as to realize the intelligent recommendation of club activities. In this way, suitable club activities can be recommended to students according to their hobbies and professional inclinations, thereby improving student participation and satisfaction and promoting the all-round development of students.

Description

Technical Field

[0001] This application relates to the field of intelligent freshmen welcoming, and more specifically, to a campus intelligent freshmen welcoming system and method based on artificial intelligence. Background Art

[0002] With the rapid development of information technology, higher education institutions are facing the challenge of better meeting the needs of students in the new era and providing personalized services. The traditional way of welcoming freshmen in colleges and universities often focuses on the guidance and arrangement of administrative processes and mainly relies on paper materials or simple online information entry. This way is not only inefficient but also ignores the diversity of freshmen's personal interests and needs, making it difficult to effectively promote the rapid integration and personalized development of freshmen.

[0003] For freshmen, college life is not limited to academic learning but also involves a wide range of social activities and the development of personal interests. Currently, club activities, as an important part of college life, play an important role in the construction of freshmen's social networks and interest cultivation. However, for freshmen, the choice of club activities mainly depends on random recommendations or the advice of senior students, lacking systematic guidance and personalized matching. As a result, freshmen often have difficulty finding club activities that match their interests, hobbies, and professional needs, leading to low participation rates and even missing opportunities to develop personal potential and expand interpersonal networks. Summary of the Invention

[0004] To solve the above technical problems, this application is proposed. The embodiments of this application provide a campus intelligent freshmen welcoming system and method based on artificial intelligence. After a freshman creates a personal account and enters information, it receives the text descriptions of the interests and hobbies and professional needs input by the student, and uses natural language processing technology based on deep learning to perform semantic parsing and feature fusion on the student's interests and hobbies and professional needs to obtain a joint representation of the student's personalized needs. Furthermore, it extracts the club activity information of the school, and through fine-grained semantic interaction queries between the club activity information and the student's personalized needs, it reveals the matching degree between the two, thereby realizing the intelligent recommendation of club activities. In this way, suitable club activities can be recommended to students according to their interests and professional inclinations, thereby improving student participation and satisfaction and promoting the all-round development of students.

[0005] According to one aspect of this application, a campus intelligent freshmen welcoming system based on artificial intelligence is provided, which includes: a student registration module for registering a student account;

[0006] An interest and hobby collection module for receiving the text description of the interests and hobbies input by the target student object; a professional need collection module for receiving the text description of the professional needs input by the target student object;

[0007] The intelligent freshman welcoming module is used to recommend club activities based on the text description of hobbies and the text description of professional needs;

[0008] Among them, the intelligent freshman welcoming module includes: a student information semantic feature extraction unit, which is used to extract the joint semantic features of the text description of hobbies and the text description of professional needs to obtain an interest-hobby - professional-need semantic concatenated representation vector; a club activity information acquisition unit, which is used to acquire the text description of the first alternative club activity; a club activity information semantic encoding unit, which is used to perform semantic encoding on the text description of the first alternative club activity to obtain a first alternative club activity semantic encoding feature vector; a semantic interaction encoding unit, which is used to perform fine-grained feature interaction based on external knowledge guidance on the interest-hobby - professional-need semantic concatenated representation vector and the first alternative club activity semantic encoding feature vector to obtain a student-demand - club-activity fine-grained semantic interaction feature vector; a club activity recommendation unit, which is used to determine whether to recommend the first alternative club activity to the target student object based on the student-demand - club-activity fine-grained semantic interaction feature vector.

[0009] According to another aspect of the present application, there is provided an artificial intelligence-based campus intelligent freshman welcoming method, which includes:

[0010] Register a student account;

[0011] Receive the text description of hobbies input by the target student object;

[0012] Receive the text description of professional needs input by the target student object;

[0013] Recommend club activities based on the text description of hobbies and the text description of professional needs.

[0014] Compared with the prior art, the artificial intelligence-based campus intelligent freshman welcoming system and method provided by the present application, after a freshman creates a personal account and enters information, receives the text description of hobbies and the text description of professional needs input by the student, and uses deep learning-based natural language processing technology to perform semantic parsing and feature fusion on the student's hobbies and professional needs to obtain a joint representation of the student's personalized needs. Furthermore, it extracts the club activity information of the school, and through fine-grained semantic interaction query between the club activity information and the student's personalized needs, reveals the matching degree between the two, so as to realize the intelligent recommendation of club activities. In this way, suitable club activities can be recommended to students according to their hobbies and professional inclinations, thereby improving student participation and satisfaction and promoting the all-round development of students. BRIEF DESCRIPTION OF THE DRAWINGS

[0015] The above and other objects, features, and advantages of the present application will become more apparent by describing the embodiments of the present application in more detail with reference to the accompanying drawings. The drawings are used to provide a further understanding of the embodiments of the present application and constitute a part of the specification. Together with the embodiments of the present application, they are used to explain the present application and do not constitute a limitation to the present application. In the drawings, the same reference numerals generally represent the same components or steps.

[0016] Figure 1 FIG. is a block diagram of an AI-based campus intelligent freshmen welcoming system according to an embodiment of the present application;

[0017] Figure 2 FIG. is a schematic diagram of data flow of an AI-based campus intelligent freshmen welcoming system according to an embodiment of the present application;

[0018] Figure 3 FIG. is a block diagram of an intelligent freshmen welcoming module in an AI-based campus intelligent freshmen welcoming system according to an embodiment of the present application;

[0019] Figure 4 FIG. is a flowchart of an AI-based campus intelligent freshmen welcoming method according to an embodiment of the present application. Detailed Embodiments

[0020] Hereinafter, exemplary embodiments according to the present application will be described in detail with reference to the drawings. Obviously, the described embodiments are only a part of the embodiments of the present application, rather than all the embodiments of the present application. It should be understood that the present application is not limited by the exemplary embodiments described herein.

[0021] As shown in the present application and the claims, unless the context clearly indicates an exception, words such as "a", "an", "one", and / or "the" are not specifically singular and may also include plural. Generally speaking, the terms "including" and "comprising" only indicate the inclusion of the steps and elements that have been clearly identified, and these steps and elements do not constitute an exclusive list. The method or device may also include other steps or elements.

[0022] Although the present application makes various references to certain modules in the system according to the embodiments of the present application, however, any number of different modules can be used and run on the user terminal and / or the server. The modules are only illustrative, and different aspects of the system and method can use different modules.

[0023] Flowcharts are used in the present application to illustrate the operations performed by the system according to the embodiments of the present application. It should be understood that the operations before or below do not necessarily need to be executed precisely in sequence. On the contrary, as needed, various steps can be executed in reverse order or simultaneously. At the same time, other operations can also be added to these processes, or one or several steps can be removed from these processes.

[0024] Next, exemplary embodiments according to the present application will be described in detail with reference to the accompanying drawings. Obviously, the described embodiments are only a part of the embodiments of the present application, rather than all embodiments of the present application. It should be understood that the present application is not limited by the exemplary embodiments described herein.

[0025] In the technical solution of the present application, a campus intelligent freshman enrollment system based on artificial intelligence is proposed. Figure 1 FIG. is a flowchart of a wind-solar power generation energy storage management method according to an embodiment of the present application. Figure 2 FIG. is a system architecture diagram of a wind-solar power generation energy storage management method according to an embodiment of the present application. As Figure 1 and Figure 2 shown, the artificial intelligence-based campus intelligent freshman enrollment system 300 according to an embodiment of the present application includes: a student registration module 310 for registering student accounts; an interest and hobby collection module 320 for receiving text descriptions of interests and hobbies input by a target student object; a professional need collection module 330 for receiving text descriptions of professional needs input by the target student object; and an intelligent freshman enrollment module 340 for recommending club activities based on the text descriptions of interests and hobbies and the text descriptions of professional needs.

[0026] Specifically, the student registration module 310 is used to register student accounts. In the artificial intelligence-based campus intelligent freshman enrollment system, the process of registering student accounts is the first step of the entire process, which lays the foundation for subsequent student information collection, personalized need analysis, and club activity recommendation.

[0027] In one example, when freshmen are about to enter college life, they first need to access the registration page of the intelligent freshman enrollment system through the official platform designated by the school. This platform can be the school's official website or a mobile application specifically designed for freshman enrollment. To ensure a good experience on all devices, the platform adopts a responsive design, and freshmen can easily complete the registration whether on a computer, tablet, or mobile phone.

[0028] After freshmen access the registration page, they will immediately be guided to perform identity verification. This is a key step to ensure the legitimacy of freshmen's identities. The system usually requires providing the admission notice number, ID number, or other unique identifiers to confirm the identity. To increase security, the system may introduce two-factor authentication (2FA), such as sending a one-time verification code via text message or an email verification link, to ensure that only real freshmen can create accounts and continue the registration process.

[0029] After successfully passing the authentication, freshmen will be guided to a form page where they need to fill in a series of basic information, such as name, gender, date of birth, contact information, etc. These information constitute an important part of the freshmen's personal profiles, helping the school better understand the basic situation of each freshman. At the same time, these information will also be used in subsequent student management and service provision. To simplify the operation, some information can be automatically filled from the electronic files provided when freshmen enroll, reducing the workload of manual input and the possibility of errors.

[0030] Next, freshmen need to set up a username and password for themselves. These are the key credentials for future system logins. The system should give clear guidance on how to create a strong password, such as including uppercase and lowercase letters, numbers, and special characters, to enhance the security of the account. In addition to the basic login credentials, freshmen can also be encouraged to set up security questions and answers as a way to retrieve passwords.

[0031] Before submitting any information, the system will prompt freshmen to read and agree to the relevant service terms and privacy policies. This is an important step because it clarifies how user data will be collected, used, and protected. In this way, the school demonstrates respect for personal privacy rights to freshmen and complies with the requirements of relevant laws and regulations. At the same time, this is also an important step in building a trust relationship, making freshmen feel that their personal information is properly handled.

[0032] When all required fields have been completed, freshmen click the "Submit" button. At this time, the system will conduct a preliminary check on the filled information to ensure there are no obvious errors or omissions. Once the check passes, the information will be submitted to the back-end management system for review. If the system has the ability to automate the review, this process can be completed within a few seconds and the account will be activated immediately; for cases that require manual intervention, freshmen will be notified of the review results within a certain period of time. In either case, the review time should be minimized as much as possible to improve the user experience.

[0033] Once the review is passed, the freshmen's accounts will be activated and they can start using other functions of the intelligent freshman enrollment system. At this time, the system will automatically send a confirmation email or text message to the freshmen, which contains the login link and initial guidance information. This email not only serves as a notice but also as a backup measure. In case freshmen forget the login address or encounter technical problems, they can regain access through the guidance in the email.

[0034] After successful registration, the system usually directly guides or prompts freshmen to go to the next step - entering text descriptions of hobbies and professional needs.

[0035] Specifically, the hobby collection module 320 is used to receive the text description of hobbies input by the target student object. In one example, to ensure that freshmen can easily and accurately express their hobbies, the system needs to provide an intuitive and user-friendly interface. This interface should include clear instructions and examples to help freshmen understand how to describe their hobbies. For example, the system can provide a series of prompt words or tags for common hobby categories (such as music, sports, reading, etc.), as well as short example texts to let freshmen know what kind of description is valid. Considering that freshmen may be unfamiliar with some fields or unsure how to describe them, the system can also adopt a question-and-answer form to assist freshmen in expressing themselves more comprehensively. In the actual data collection process, the system provides multiple ways to adapt to the different needs and technical levels of freshmen. First, the system will provide one or more free text input boxes, allowing freshmen to freely describe their hobbies in natural language. This method gives freshmen the greatest flexibility to express their love for various things in their own words. At the same time, to avoid information being too scattered, there should be clear prompts next to each input box to tell freshmen to describe around specific topics, such as academic interests, artistic creation, physical exercise, etc. In addition, the system can also be designed in the form of multiple-choice questions or rating items, allowing students to select the content that best matches their interests from the preset options and score according to importance. This structured data collection method not only simplifies the operation difficulty for users but also facilitates subsequent data analysis and processing. For example, for each listed hobby category, freshmen can choose "very like", "like generally", or "dislike", and sort them according to the intensity of preference. If freshmen have a ready-made list of hobbies or other relevant documents, the system can support the file upload function and use optical character recognition (OCR) technology to automatically extract the text content therein. This can not only save time but also ensure the accuracy of information. However, it should be noted that when using OCR, it is necessary to ensure that the uploaded file format is compatible and provide necessary format conversion tools.

[0036] Specifically, the professional need collection module 330 is used to receive the text description of professional needs input by the target student object. To ensure that freshmen can accurately express their professional needs, the system needs to provide an intuitive and user-friendly interface. This interface should not only include clear instructions and examples to help freshmen understand how to describe their professional interests, academic goals, and future career plans, but also take into account that freshmen may not be clear about their major choices or unsure how to specify their needs. Therefore, the system can assist freshmen in expressing themselves more comprehensively through a question-and-answer form. For example, some guiding questions can be asked: "What are your expectations for your future career?", "In which fields do you have special interests or skills?", "What do you hope to gain through university study?", and so on. These questions can help freshmen think and specify their professional needs, thus providing more valuable information for subsequent analysis.

[0037] In the actual data collection process, the system provides multiple ways to adapt to the different needs and technical levels of freshmen. First, the system will provide one or more free text input boxes, allowing freshmen to freely describe their professional needs in the form of natural language. This method gives freshmen the greatest flexibility, enabling them to express their interests in specific disciplines, research directions, or career goals in their own words. At the same time, to avoid information being too scattered, clear prompts should be provided next to each input box, telling freshmen to describe around specific topics, such as academic interests, research directions, career skills, etc. In addition, the system can also be designed in the form of multiple-choice questions or rating items, allowing students to select the content that best meets their professional needs from the preset options and score according to importance. Such a structured data collection method not only simplifies the operation difficulty for users but also facilitates subsequent data analysis and processing. For example, for each listed major category, freshmen can choose "very interested", "moderately interested", or "not interested" and sort them according to the intensity of interest. In addition, multiple-choice questions can be set to allow freshmen to select multiple relevant fields to reflect their broad and complex academic interests. If freshmen have ready-made professional planning documents or other relevant materials, the system can support the file upload function and use optical character recognition (OCR) technology to automatically extract the text content therein. This can not only save time but also ensure the accuracy of information. However, it should be noted that when using OCR, it is necessary to ensure that the uploaded file format is compatible and provide necessary format conversion tools to ensure that all freshmen can use this function smoothly.

[0038] Once the text descriptions of the freshmen's professional needs are collected, the next step is to perform semantic parsing and feature extraction on them, which are the core steps for realizing intelligent recommendation. First, clean the received text, remove irrelevant symbols and stop words (such as high-frequency but meaningless words like "de" and "le"), and split long sentences to better capture individual professional need points. This cleaning work helps improve the efficiency and accuracy of subsequent processing. Then, apply the TF-IDF algorithm, TextRank algorithm, or other advanced keyword extraction techniques to extract the most representative words from the text as professional labels. These labels will serve as the basis for subsequent matching. Keyword extraction can not only condense text information but also help the system quickly locate the freshmen's main professional interest areas. For professional descriptions with subjective evaluation nature, such as "very eager to engage in artificial intelligence research", the sentiment analysis model can be used to judge its positive or negative sentiment tendency, so as to more accurately reflect the freshmen's true attitude. Sentiment analysis enables the system to distinguish different levels of demand intensity and further refine the recommendation results.

[0039] Specifically, the intelligent freshman enrollment module 340 is used to recommend club activities based on the hobby text description and the professional need text description. That is, after the freshmen complete the registration process of the student account, the system will guide them to enter text descriptions about their personal hobbies and professional needs through the user interface, so as to match and recommend club activities that fit their interests and professional needs. Specifically, in a specific example of the present application, as Figure 3 shown, the intelligent freshman enrollment module 340 includes: a student information semantic feature extraction unit 341, which is used to extract the joint semantic features of the hobby text description and the professional need text description to obtain a hobby-professional need semantic concatenated representation vector; a club activity information acquisition unit 342, which is used to acquire the text description of the first alternative club activity; a club activity information semantic encoding unit 343, which is used to perform semantic encoding on the text description of the first alternative club activity to obtain a first alternative club activity semantic encoding feature vector; a semantic interaction encoding unit 344, which is used to perform fine-grained feature interaction based on external knowledge guidance on the hobby-professional need semantic concatenated representation vector and the first alternative club activity semantic encoding feature vector to obtain a student demand-club activity fine-grained semantic interaction feature vector; a club activity recommendation unit 345, which is used to determine whether to recommend the first alternative club activity to the target student object based on the student demand-club activity fine-grained semantic interaction feature vector.

[0040] Specifically, the student information semantic feature extraction unit 341 is used to extract the joint semantic features of the hobby text description and the major requirement text description to obtain the hobby-major requirement semantic concatenated representation vector. That is, in the technical solution of this application, first, considering that the simple keyword matching technology lacks in-depth understanding of students' hobbies and major requirements, it is often difficult to accurately capture students' personalized needs. Therefore, in the technical solution of this application, in order to improve the accuracy of club activity recommendations, natural language processing technology is adopted to perform semantic analysis on the hobby text description and the major requirement text description input by students, so as to achieve in-depth understanding of students' hobbies and major requirements. In a specific example of this application, a semantic encoder based on the BERT model is used to perform semantic encoding on the hobby text description and the major requirement text description to obtain the hobby semantic encoding feature vector and the major requirement semantic encoding feature vector. It should be noted that the core idea of BERT is to learn the context-related representation of text through a deep bidirectional Transformer architecture. Different from previous unidirectional language models, BERT can utilize both left and right context information, thus more accurately capturing the relationships between words. Specifically, BERT uses two strategies, namely "Masked Language Model" (MLM) and "Next Sentence Prediction" (NSP), for pre-training, enabling the model to perform better in understanding the relationships within and between sentences.

[0041] Next, the hobby semantic encoding feature vector and the major requirement semantic encoding feature vector are concatenated to obtain the hobby-major requirement semantic concatenated representation vector. Here, considering that hobbies and major requirements are two different but important aspects of students. Hobbies reflect students' preferences in extracurricular activities, while major requirements reflect their pursuits in academic or career development. If these two aspects are processed separately, the potential correlation between them may be ignored. Therefore, in this application, through the concatenation operation, information fusion is performed on the hobby semantic encoding feature vector and the major requirement semantic encoding feature vector to form the hobby-major requirement semantic concatenated representation vector, thereby achieving a comprehensive representation of students' personalized needs and providing a more comprehensive and accurate basis for subsequent club activity recommendations.

[0042] Specifically, the club activity information acquisition unit 342 is used to acquire the text description of the first alternative club activity. In a specific example of the present application, the text description of the first alternative club activity includes, but is not limited to, information such as the cultural concept of the club, activity content, skill fields involved, time arrangement, activity location, etc. Here, for the sake of simplicity in description, in the technical solution of the present application, only the data processing and matching recommendation process for a single alternative club activity is elaborated in detail. However, it should be understood that in actual applications, similar processing can be performed on each club activity in the school, so as to achieve traversal query and intelligent recommendation of club activities.

[0043] Specifically, the club activity information semantic encoding unit 343 is used to perform semantic encoding on the text description of the first alternative club activity to obtain the first alternative club activity semantic encoding feature vector. In order to achieve semantic interaction and matching between club activities and students' personalized needs, first of all, it is necessary to perform semantic encoding on the text description of the first alternative club activity to convert it into a machine-understandable format. In a specific example of the present application, the semantic encoder based on the BERT model is also used to perform semantic encoding on the text description of the first alternative club activity. It should be understood that in the technical solution of the present application, by adopting the same semantic encoding process, it helps to ensure the semantic comparability between club activity information and students' personalized needs, reduce semantic deviations introduced by inconsistent encoding processes, and thus improve the accuracy of semantic matching.

[0044] Specifically, the semantic interaction encoding unit 344 is used to perform fine-grained feature interaction based on external knowledge guidance on the interest-hobby - major requirement semantic concatenated representation vector and the first alternative club activity semantic encoding feature vector to obtain the student requirement - club activity fine-grained semantic interaction feature vector. In order to further improve the semantic matching accuracy between club activity information and students' personalized needs, the present application proposes a fine-grained feature interaction method based on external knowledge guidance. By introducing external knowledge, such as evaluations and opinions of former club members, participation data of former club members, etc., it optimizes the semantic interaction process between the interest-hobby - major requirement semantic concatenated representation vector and the first alternative club activity semantic encoding feature vector, reveals the potential semantic connection between the two, and thus more accurately understands the matching degree between club activities and students' personalized needs.

[0045] Specifically, the specific process of performing fine-grained feature interaction guided by external knowledge on the interest-hobby - professional-need semantic concatenated representation vector and the first alternative club activity semantic encoding feature vector includes: First, based on external knowledge, attention optimization is performed on the fine-grained interaction features between the interest-hobby - professional-need semantic concatenated representation vector and the first alternative club activity semantic encoding feature vector to obtain an externally knowledge-optimized student-need - club-activity semantic feature fine-grained interaction matrix. That is, in the embodiments of the present application, first, the interest-hobby - professional-need semantic concatenated representation vector and the first alternative club activity semantic encoding feature vector are input into a fine-grained feature interaction network to obtain a student-need - club-activity semantic feature fine-grained interaction matrix; through the fine-grained feature interaction network, the semantic relevance between the two is understood at the micro level, the fine-grained semantic interaction information between the two is captured, and a student-need - club-activity semantic feature fine-grained interaction matrix is generated. Then, the student-need - club-activity semantic feature fine-grained interaction matrix is input into an attention unit based on external knowledge to obtain the externally knowledge-optimized student-need - club-activity semantic feature fine-grained interaction matrix. That is, the generated student-need - club-activity semantic feature fine-grained interaction matrix is fed into the attention unit based on external knowledge, and external knowledge is used to further optimize the student-need - club-activity semantic feature fine-grained interaction matrix. In specific implementation, by introducing external knowledge, such as feedback from club members on club activities, participation rate, satisfaction, professional statistics of club members, etc., to train the parameter matrix of the attention unit, so that the attention unit can more accurately identify and strengthen club activity features highly relevant to students' needs, while suppressing the interference of irrelevant or unimportant features, thereby enhancing the recognition ability of the matching degree between club activities and students' personalized needs.

[0046] More specifically, based on external knowledge, the following attention optimization formula is used to perform attention optimization on the fine-grained interaction features between the interest-hobby - professional-need semantic concatenated representation vector and the first alternative club activity semantic encoding feature vector to obtain an externally knowledge-optimized student-need - club-activity semantic feature fine-grained interaction matrix; where the attention optimization formula is:

[0047]

[0048] where, V a represents the interest-hobby - professional-need semantic concatenated representation vector, V b represents the first alternative club activity semantic encoding feature vector, (·) T represents the transpose of the vector, represents matrix multiplication operation, M p represents the student-need - club-activity semantic feature fine-grained interaction matrix, Mk and M v represents the learnable memory parameter matrix of the external knowledge-based attention unit, norm(·) represents the normalization function, and M y represents the fine-grained interaction matrix of external knowledge-optimized student needs-club activity semantic features.

[0049] Subsequently, based on the fine-grained interaction matrix of external knowledge-optimized student needs-club activity semantic features, the interest-hobby-major requirement semantic concatenated representation vector and the first alternative club activity semantic encoding feature vector are respectively subjected to feature modulation optimization to obtain an optimized interest-hobby-major requirement semantic concatenated representation vector and an optimized first alternative club activity semantic encoding feature vector. That is, in the embodiment of the present application, the interest-hobby-major requirement semantic concatenated representation vector is linearly transformed to obtain a first query feature vector and a first value feature vector, and the fine-grained interaction matrix of external knowledge-optimized student needs-club activity semantic features is used as the key matrix. The first query feature vector, the first value feature vector, and the key matrix are input into the fine-grained modulation module based on the Transformer structure to obtain the optimized interest-hobby-major requirement semantic concatenated representation vector; and the first alternative club activity semantic encoding feature vector is linearly transformed to obtain a second query feature vector and a second value feature vector, and the fine-grained feature interaction matrix optimized by external knowledge is used as the key matrix. The second query feature vector, the second value feature vector, and the key matrix are input into the fine-grained modulation module based on the Transformer structure to obtain the optimized first alternative club activity semantic encoding feature vector. Here, by using the fine-grained interaction matrix of student needs-club activity semantic features optimized by external knowledge as the key matrix, and simultaneously constructing a first query feature vector and a first value feature vector based on the interest-hobby-major requirement semantic concatenated representation vector, and constructing a second query feature vector and a second value feature vector based on the first alternative club activity semantic encoding feature vector, the information exchange and integration between the features inside and external knowledge are realized through the self-attention mechanism of the Transformer structure, so as to ensure that the interest-hobby-major requirement semantic concatenated representation vector and the first alternative club activity semantic encoding feature vector can obtain a deeper semantic understanding from external knowledge to improve the quality of their feature expressions.

[0050] More specifically, based on the external knowledge, optimize the fine-grained interaction matrix of student needs - club activity semantic features, and perform feature modulation optimization on the semantic concatenation representation vector of hobby - major needs and the semantic encoding feature vector of the first alternative club activity respectively using the following feature modulation optimization formula to obtain the optimized semantic concatenation representation vector of hobby - major needs and the optimized semantic encoding feature vector of the first alternative club activity; where the feature modulation optimization formula is:

[0051]

[0052] where, W 1q and W 1v respectively represent the first query embedding matrix and the first value embedding matrix, V 1q and V 1v respectively represent the first query feature vector and the first value feature vector, W 2q and W 2v respectively represent the second query embedding matrix and the second value embedding matrix, V 2q and V 2v respectively represent the second query feature vector and the second value feature vector, b 1q 、b 1v 、b 2q and b 2v respectively represent different bias terms, softmax(·) represents the normalized exponential function, d represents the feature scale value of the external knowledge optimized fine-grained feature interaction matrix, V a ′ and V b ′ respectively represent the optimized semantic concatenation representation vector of hobby - major needs and the optimized semantic encoding feature vector of the first alternative club activity.

[0053] Furthermore, perform per-position semantic response encoding on the optimized semantic concatenation representation vector of hobby - major needs and the optimized semantic encoding feature vector of the first alternative club activity to obtain the fine-grained semantic interaction feature vector of student needs - club activities. That is, in the embodiments of the present application, calculate the division of the optimized semantic concatenation representation vector of hobby - major needs and the optimized semantic encoding feature vector of the first alternative club activity at each position to obtain the fine-grained semantic interaction feature vector of student needs - club activities. Here, by performing division at each position on the optimized semantic concatenation representation vector of hobby - major needs and the optimized semantic encoding feature vector of the first alternative club activity, per-position semantic query response encoding is performed to obtain the fine-grained semantic interaction feature vector of student needs - club activities, thereby more accurately reflecting the semantic matching information between students' personalized needs and club activities.

[0054] More specifically, the optimized interest - major demand semantic - level concatenated representation vector and the optimized first alternative club activity semantic coding feature vector are subjected to per - position semantic response coding according to the following per - position semantic response coding formula to obtain the student demand - club activity fine - grained semantic interaction feature vector; where the per - position semantic response coding formula is:

[0055]

[0056] where, V i represents the student demand - club activity fine - grained semantic interaction feature vector.

[0057] Specifically, the club activity recommendation unit 345 is configured to determine whether to recommend the first alternative club activity to the target student object based on the student demand - club activity fine - grained semantic interaction feature vector. In the technical solution of this application, the student demand - club activity fine - grained semantic interaction feature vector is input into a recommendation result generator based on an SVM model to obtain a recommendation result, and the recommendation result is used to indicate whether to recommend the first alternative club activity to the target student object. It should be understood that the SVM model has good generalization ability and classification performance, can handle high - dimensional data and effectively find the optimal decision boundary. In the technical solution of this application, the SVM model is trained to identify the matching relationship between student demands and club activities. By using the known student - club activity pairings to construct a training data set, the SVM is trained to learn a decision boundary that can correctly judge the pairing relationship between student demands and club activities. After training is completed, the trained SVM model performs probability calculation on the input student demand - club activity fine - grained semantic interaction feature vector to obtain the probability that it belongs to the positive class (should be recommended) or the negative class (should not be recommended), thereby giving a recommendation result indicating whether the club activity should be recommended to the target student.

[0058] When the interest - hobby - major - requirement semantic - level cascaded representation vector and the first alternative club - activity semantic - coding feature vector respectively represent the semantic - joint - coding features of the interest - hobby text description and major - requirement text description of the target student object and the text - description semantic - coding features of the first alternative club activity, when performing feature - fine - grained interaction based on external knowledge modulation, the introduction of external knowledge will improve the accuracy and comprehensiveness of semantic - fine - grained interaction. However, it also makes the student - requirement - club - activity fine - grained semantic - interaction feature vector contain the semantic expression of external knowledge in its internal semantic structure, resulting in the diversity of semantic - distribution expressions in the feature distribution of the student - requirement - club - activity fine - grained semantic - interaction feature vector. Therefore, when the student - requirement - club - activity fine - grained semantic - interaction feature vector is input into the recommendation - result generator based on the SVM model for classification probability regression, it will affect the accuracy of the recommendation result.

[0059] Based on this, when the student - requirement - club - activity fine - grained semantic - interaction feature vector is input into the recommendation - result generator based on the SVM model for classification probability regression, first, optimize the student - requirement - club - activity fine - grained semantic - interaction feature vector. The optimization process specifically includes:

[0060] Determine the mean and variance of the feature set composed of the eigenvalues of the student - requirement - club - activity fine - grained semantic - interaction feature vector, and divide the mean by the variance to obtain the student - requirement - club - activity fine - grained semantic - interaction probability - weight value:

[0061]

[0062] where μ represents the mean of the feature set composed of the eigenvalues of the student - requirement - club - activity fine - grained semantic - interaction feature vector, σ represents the variance of the feature set composed of the eigenvalues of the student - requirement - club - activity fine - grained semantic - interaction feature vector, and θ represents the student - requirement - club - activity fine - grained semantic - interaction probability - weight value:

[0063] Multiply the student - requirement - club - activity fine - grained semantic - interaction feature vector by the reciprocal of the maximum eigenvalue of the student - requirement - club - activity fine - grained semantic - interaction feature vector to obtain the student - requirement - club - activity fine - grained semantic - interaction probability - constraint vector:

[0064] V1 = V⊙v nxx -1

[0065] where V represents the student - requirement - club - activity fine - grained semantic - interaction feature vector, v max -1represents the reciprocal of the maximum eigenvalue of the fine-grained semantic interaction feature vector of the student needs - club activities, ⊙ represents dot product, and V1 represents the fine-grained semantic interaction probability constraint vector of the student needs - club activities;

[0066] Dot-add the fine-grained semantic interaction probability constraint vector of the student needs - club activities with the fine-grained semantic interaction probability weight value of the student needs - club activities, and take the logarithm with base 2 to obtain the fine-grained semantic interaction information interaction vector of the student needs - club activities:

[0067]

[0068] where, represents dot-addition, log represents the logarithmic operation with base 2, and V2 represents the fine-grained semantic interaction information interaction vector of the student needs - club activities;

[0069] After subtracting the fine-grained semantic interaction probability constraint vector of the student needs - club activities from the fine-grained semantic interaction probability weight value of the student needs - club activities, calculate the reciprocal of each eigenvalue to obtain the fine-grained semantic interaction sequence constraint vector of the student needs - club activities:

[0070]

[0071] where, represents subtraction, and V3 represents the fine-grained semantic interaction sequence constraint vector of the student needs - club activities;

[0072] Finally, dot-add the fine-grained semantic interaction information interaction vector of the student needs - club activities with the fine-grained semantic interaction sequence constraint vector of the student needs - club activities to obtain the optimized fine-grained semantic interaction feature vector of the student needs - club activities.

[0073] That is, considering that when the weight matrix in the classification probability regression acts on the student demand-club activity fine-grained semantic interaction feature vector, the uncertainty of the weight stimulation parameters of the weight matrix caused by the distribution diversity of the student demand-club activity fine-grained semantic interaction feature vector results in the lack of posterior probability density inference of the student demand-club activity fine-grained semantic interaction feature vector via the action of the weight matrix, which will affect the accuracy of the probability regression result. The above optimization process uses the probability statistical features of the student demand-club activity fine-grained semantic interaction feature vector to simulate the mesoscale interaction structure under probability constraints between the eigenvalue scale and the eigenvector scale of the student demand-club activity fine-grained semantic interaction feature vector, constructs a bidirectional latent variable motif based on the mesoscale short sequence relative to the probability statistical value for mesoscale bidirectional migration, and performs posterior recovery based on the sequence constraint of the interaction information, improving the convergence effect in the probability density domain of the student demand-club activity fine-grained semantic interaction feature vector and improving the accuracy of the recommendation result obtained by its input to the recommendation result generator based on the SVM model.

[0074] As described above, the artificial intelligence-based campus intelligent freshman welcoming system 300 according to the embodiments of the present application can be implemented in various wireless terminals, such as a server with an artificial intelligence-based campus intelligent freshman welcoming algorithm. In a possible implementation manner, the artificial intelligence-based campus intelligent freshman welcoming system 300 according to the embodiments of the present application can be integrated into the wireless terminal as a software module and / or a hardware module. For example, the artificial intelligence-based campus intelligent freshman welcoming system 300 can be a software module in the operating system of the wireless terminal, or can be an application program developed for the wireless terminal; of course, the artificial intelligence-based campus intelligent freshman welcoming system 300 can also be one of the many hardware modules of the wireless terminal.

[0075] Alternatively, in another example, the artificial intelligence-based campus intelligent freshman welcoming system 300 and the wireless terminal can also be separate devices, and the artificial intelligence-based campus intelligent freshman welcoming system 300 can be connected to the wireless terminal through a wired and / or wireless network and transmit interaction information in accordance with a predefined data format.

[0076] Furthermore, an artificial intelligence-based campus intelligent freshman welcoming method is also provided.

[0077] Figure 4 It is a flowchart of the artificial intelligence-based campus intelligent freshman welcoming method according to the embodiments of the present application. As Figure 4As shown, the AI-based intelligent campus freshman orientation method according to an embodiment of the present application includes the steps of: S1, registering a student account; S2, receiving a text description of hobbies input by a target student object; S3, receiving a text description of professional needs input by the target student object; S4, recommending club activities based on the text description of hobbies and the text description of professional needs.

[0078] In summary, the AI-based intelligent campus freshman orientation method according to an embodiment of the present application is elucidated. After freshmen create personal accounts and input information, it receives the text description of hobbies and the text description of professional needs input by students, and uses natural language processing technology based on deep learning to perform semantic parsing and feature fusion on students' hobbies and professional needs to obtain a joint representation of students' personalized needs. Furthermore, it extracts club activity information from the school, and through fine-grained semantic interaction query between the club activity information and students' personalized needs, reveals the matching degree between the two, thereby realizing intelligent recommendation of club activities. In this way, suitable club activities can be recommended to students according to their hobbies and professional inclinations, thereby improving students' participation and satisfaction and promoting the all-round development of students.

[0079] The embodiments of the present disclosure have been described above. The above description is exemplary and not exhaustive, and is not limited to the disclosed embodiments. Many modifications and variations are obvious to those of ordinary skill in the art in the technical field without departing from the scope and spirit of the described embodiments. The choice of terms used herein is intended to best explain the principles of the embodiments, practical applications, or improvements to technologies in the market, or to enable other ordinary skill in the art in the technical field to understand the embodiments disclosed herein.

Claims

1. An intelligent campus freshmen welcoming system based on artificial intelligence, characterized in that, Including: A student registration module for registering student accounts; An interest and hobby collection module for receiving text descriptions of interests and hobbies input by a target student object; A professional need collection module for receiving text descriptions of professional needs input by the target student object; A smart freshman welcome module for recommending club activities based on the text descriptions of interests and hobbies and the text descriptions of professional needs; Among them, the smart freshman welcome module includes: a student information semantic feature extraction unit for extracting the joint semantic features of the text descriptions of interests and hobbies and the text descriptions of professional needs to obtain an interest-hobby-professional need semantic concatenated representation vector; a club activity information acquisition unit for acquiring text descriptions of first alternative club activities; a club activity information semantic encoding unit for semantically encoding the text descriptions of the first alternative club activities to obtain a first alternative club activity semantic encoding feature vector; a semantic interaction encoding unit for performing fine-grained feature interaction guided by external knowledge on the interest-hobby-professional need semantic concatenated representation vector and the first alternative club activity semantic encoding feature vector to obtain a student need-club activity fine-grained semantic interaction feature vector; a club activity recommendation unit for determining whether to recommend the first alternative club activity to the target student object based on the student need-club activity fine-grained semantic interaction feature vector; When inputting the student need-club activity fine-grained semantic interaction feature vector into a recommendation result generator based on an SVM model for classification probability regression, the student need-club activity fine-grained semantic interaction feature vector is optimized, and the optimization process specifically includes: Determining the mean and variance of the feature set composed of the eigenvalues of the student need-club activity fine-grained semantic interaction feature vector, and dividing the mean by the variance to obtain a student need-club activity fine-grained semantic interaction probability weight value: Dot-multiplying the student need-club activity fine-grained semantic interaction feature vector by the reciprocal of the maximum eigenvalue of the student need-club activity fine-grained semantic interaction feature vector to obtain a student need-club activity fine-grained semantic interaction probability constraint vector; Dot-adding the student need-club activity fine-grained semantic interaction probability constraint vector and the student need-club activity fine-grained semantic interaction probability weight value, and taking the logarithm with base 2 to obtain a student need-club activity fine-grained semantic interaction information interaction vector; After subtracting the student need-club activity fine-grained semantic interaction probability constraint vector from the student need-club activity fine-grained semantic interaction probability weight value, calculating the reciprocal of each eigenvalue to obtain a student need-club activity fine-grained semantic interaction sequence constraint vector; Dot-adding the student need-club activity fine-grained semantic interaction information interaction vector and the student need-club activity fine-grained semantic interaction sequence constraint vector to obtain an optimized student need-club activity fine-grained semantic interaction feature vector.

2. The campus intelligent freshmen welcoming system based on artificial intelligence according to claim 1, characterized in that, The student information semantic feature extraction unit includes: A semantic feature extraction subunit, configured to perform semantic encoding on the hobby text description and the professional requirement text description to obtain a hobby semantic encoding feature vector and a professional requirement semantic encoding feature vector; A semantic feature concatenation subunit, configured to concatenate the hobby semantic encoding feature vector and the professional requirement semantic encoding feature vector to obtain the hobby-professional requirement semantic concatenated representation vector.

3. The campus intelligent freshmen welcoming system based on artificial intelligence according to claim 2, characterized in that, The semantic feature extraction subunit is configured to: Use a semantic encoder based on the BERT model to perform semantic encoding on the hobby text description and the professional requirement text description to obtain the hobby semantic encoding feature vector and the professional requirement semantic encoding feature vector.

4. The campus intelligent freshmen welcoming system based on artificial intelligence according to claim 3, characterized in that, The semantic interaction encoding unit includes: A feature fine-grained interaction optimization subunit, configured to perform attention optimization on the fine-grained interaction features between the hobby-professional requirement semantic concatenated representation vector and the first alternative club activity semantic encoding feature vector based on external knowledge to obtain an externally knowledge-optimized student requirement-club activity semantic feature fine-grained interaction matrix; A feature-guided modulation subunit, configured to perform feature modulation optimization on the hobby-professional requirement semantic concatenated representation vector and the first alternative club activity semantic encoding feature vector respectively based on the externally knowledge-optimized student requirement-club activity semantic feature fine-grained interaction matrix to obtain an optimized hobby-professional requirement semantic concatenated representation vector and an optimized first alternative club activity semantic encoding feature vector; A per-position semantic interaction encoding subunit, configured to perform per-position semantic response encoding on the optimized hobby-professional requirement semantic concatenated representation vector and the optimized first alternative club activity semantic encoding feature vector to obtain the student requirement-club activity fine-grained semantic interaction feature vector.

5. The campus intelligent freshman enrollment system based on artificial intelligence according to claim 4, characterized in that, The feature fine-grained interaction optimization subunit is configured to: Input the hobby-professional requirement semantic concatenated representation vector and the first alternative club activity semantic encoding feature vector into a fine-grained feature interaction network to obtain a student requirement-club activity semantic feature fine-grained interaction matrix; Input the student requirement-club activity semantic feature fine-grained interaction matrix into an attention unit based on external knowledge to obtain the externally knowledge-optimized student requirement-club activity semantic feature fine-grained interaction matrix.

6. The campus intelligent freshman orientation system based on artificial intelligence according to claim 5, wherein The feature-guided modulation subunit is configured to: Perform a linear transformation on the hobby-professional requirement semantic concatenated representation vector to obtain a first query feature vector and a first value feature vector, and use the externally knowledge-optimized student requirement-club activity semantic feature fine-grained interaction matrix as a key matrix, and input the first query feature vector, the first value feature vector, and the key matrix into a fine-grained modulation module based on the Transformer structure to obtain the optimized hobby-professional requirement semantic concatenated representation vector; Perform a linear transformation on the first alternative club activity semantic encoding feature vector to obtain a second query feature vector and a second value feature vector, and use the external knowledge-optimized fine-grained feature interaction matrix as the key matrix. Input the second query feature vector, the second value feature vector, and the key matrix into the fine-grained modulation module based on the Transformer structure to obtain the optimized first alternative club activity semantic encoding feature vector.

7. The campus intelligent freshmen welcoming system based on artificial intelligence according to claim 6, characterized in that, The per-position semantic interaction encoding sub-unit is used for: Calculate the per-position point division between the optimized hobby-major requirement semantic concatenated representation vector and the optimized first alternative club activity semantic encoding feature vector to obtain the student requirement-club activity fine-grained semantic interaction feature vector.

8. The campus intelligent freshman welcoming system based on artificial intelligence according to claim 7, characterized in that, The club activity recommendation unit is used for: Input the student requirement-club activity fine-grained semantic interaction feature vector into a recommendation result generator based on the SVM model to obtain a recommendation result, which is used to indicate whether to recommend the first alternative club activity to the target student object.

9. A campus intelligent freshman welcoming method based on artificial intelligence, which is executed by the campus intelligent freshman welcoming system according to any one of claims 1 to 8, characterized in that, It includes: Register a student account; Receive the text description of hobbies input by the target student object; Receive the text description of major requirements input by the target student object; Recommend club activities based on the text description of hobbies and the text description of major requirements.

Images