Learning resource recommendation method and device based on machine learning, terminal and medium

Abstract

The invention provides a learning resource recommendation method and device based on machine learning, a terminal and a medium, and relates to the technical field of artificial intelligence, and the method comprises the steps: recording the visiting behavior data of tourists in a venue, including the residence time, the number of visited exhibition areas and the interaction behavior with exhibits; calculating an interest concentration index, an interaction depth index, an exploration breadth index and an exhibition area preference index of the tourist, generating a tourist feature index based on the interest concentration index and the interaction depth index, training a tourist portrait prediction model based on a deep learning network, and performing tourist portrait prediction by using the tourist feature index as input; and finally, based on the combination of the prediction result of the tourist portrait and the knowledge graph, recommending corresponding learning resources for the tourist portrait.

Description

Technical Field

[0001] The present invention relates to the field of artificial intelligence technology, and specifically to a learning resource recommendation method, device, terminal and medium based on machine learning. Background Art

[0002] With the rapid development of social informatization, venues such as science and technology museums, cultural centers, and museums have gradually become important places for people to learn and explore knowledge. These venues usually contain a large number of exhibition areas and exhibits. Therefore, it is very important to recommend relevant learning resources during the visit of tourists to improve the efficiency of the visit and the interest of tourists. Traditional static recommendation methods usually make recommendations based on fixed classifications of exhibition areas or exhibits. Due to the significant differences in interests and needs of different tourists, such methods are difficult to meet the personalized learning needs of tourists and cannot tap into the potential interests of tourists. In addition, traditional recommendation methods usually lack in-depth analysis of the interactive behavior of tourists and exhibits, making it difficult to accurately capture the interest preferences of tourists, resulting in insufficient pertinence and accuracy in the recommendation of learning resources. Therefore, there is an urgent need for a personalized learning resource recommendation method that can combine tourist behavior data and venue resource characteristics to improve tourists' tour experience while promoting knowledge dissemination.

[0003] In the prior art, publication number CN118210982A discloses a learning resource recommendation method, device and electronic device based on machine learning, which obtains the identity information, skill requirement information and learning behavior information of the target user; extracts text feature vectors of the skill requirement information; determines the semantic feature representation of the text feature vector set; uses the semantic feature representation as input, and generates a target learning resource category to be recommended by a pre-trained learning resource recommendation model; determines at least one of a plurality of pre-established resource pools as a target resource pool based on the identity information and learning behavior information, determines the learning resources in the target resource pool that match the target learning resource category as the target learning resource, and returns the resource information of the target learning resource to the target user.

[0004] The main problems with the above scheme are: it is mainly based on text semantic analysis of skill requirement information, lacks feedback support for dynamic real-time behavior, and is difficult to adapt to the immediate changes in user needs; and it makes recommendations by extracting semantic feature representations from the text feature vectors of skill requirement information, pays little attention to behavioral data, and lacks quantification of user behavior patterns, resulting in insufficient accuracy and personalization of recommendations.

[0005] The above information disclosed in this Background section is only for enhancement of understanding of the background of the present disclosure and therefore it may contain information that does not constitute the prior art that is already known to one of ordinary skill in the art. Summary of the invention

[0006] The purpose of the present invention is to provide a learning resource recommendation method, device, terminal and medium based on machine learning to solve the problems raised in the above background technology.

[0007] To achieve the above object, the present invention provides the following technical solutions: A learning resource recommendation method based on machine learning, the specific steps include: Step 1: During the opening hours of the venue, identify each visitor entering the venue, record the total time each visitor spends in the venue, the number of exhibition areas visited in the venue and the length of time they stay in different exhibition areas, the number and duration of interactions between visitors and exhibits in the exhibition areas, and the total number and duration of interactions between each exhibit and all visitors; Step 2: Generate an interest concentration index based on the time tourists stay in different exhibition areas and the total time they visit the venue; generate an interaction depth index based on the total number of exhibits that tourists interact with and the number and duration of interaction with each exhibit; generate an exploration breadth index based on the number of exhibition areas visited by tourists and the total number of exhibition areas in the venue; normalize the interest concentration index and the interaction depth index; and generate an exhibition area preference index based on the normalized interest concentration index and the normalized interaction depth index. The exhibition area corresponding to the highest exhibition area preference index is the most preferred exhibition area for the tourist; Step 3: Build a knowledge graph based on the theme and specific resources of the venue, generate a visitor characteristic index based on the highest exhibition area preference index and exploration breadth index, build a model based on a deep learning network, use the visitor characteristic index as input and the visitor portrait as a label, and train a visitor portrait prediction model; Step 4: Input the tourist characteristic index of the tourist to be recommended learning resources into the tourist portrait prediction model to generate a tourist portrait, and recommend learning resources to tourists based on the tourist portrait and knowledge graph.

[0008] Furthermore, the method for counting the number of interactive behaviors is: touching the display cabinet or control panel of the exhibit for more than 3 seconds is recorded as one interactive behavior, reading the exhibit introduction for more than 5 seconds is recorded as one interactive behavior, moving around the exhibit within a 1m area around the exhibit for more than 10 seconds or experiencing the exhibit function once is recorded as one interactive behavior. The interactive time is the sum of the time of all interactive behaviors between the visitor and the corresponding exhibit.

[0009] Furthermore, the formula for generating the interest concentration index is: ; in, Indicates that tourist k has The interest concentration index of each exhibition area, Indicates the index of the exhibition area. It means that tourist k is in the The duration of stay in each exhibition area, represents the total visiting time of tourist k in the venue, k is the index of the tourist, and , It indicates the total number of visitors during the opening hours of the venue in a week; The formula used to generate the Engagement Depth Index is: ; in, Indicates that tourist k has The interactive depth index of each exhibition area, Indicates The number of exhibits in each exhibition area, Indicates The number of exhibits that visitor k interacted with in each exhibition area, represents the index of the exhibits in the exhibition area, and , Indicates that tourist k and The exhibition area Interactive time for each exhibit, indicates that tourist k and The exhibition area The number of interactive behaviors for each exhibit, Indicates The exhibition area The interactive weight of the exhibits is The exhibition area The ratio of the product of the total number of interactive behaviors and the total interactive time of each exhibit to the product of the total number of interactive behaviors and the total interactive time of all exhibits; The formula used to generate the exploration breadth index is: ; in, represents the exploration breadth index of tourist k, represents the number of exhibition areas visited by visitor k, Indicates the total number of exhibition areas in the venue; The formula for generating the exhibition area preference index is: ; in, Indicates that tourist k has The exhibition area preference index of each exhibition area, represents the number of visitors k to the first The interest concentration index of each exhibition area, Represents the weight coefficient of the interest concentration index after normalization, represents the number of visitors k to the first The interactive depth index of each exhibition area, Represents the weight coefficient of the interaction depth index after normalization. and .

[0010] Furthermore, the principles for constructing the knowledge graph are: The entities in the knowledge graph include exhibition areas, exhibits, learning resources, and visitors; The relationships between entities include exhibits belonging to exhibition areas, learning resources belonging to learning resource collections, exhibition areas being associated with learning resource collections, exhibits being associated with learning resources, visitors visiting exhibition areas, visitors interacting with exhibits, and learning resources being recommended to visitors; According to the relationship between entities, generate "entity-relationship-entity" triples and import them into the database; The hierarchical structure of the knowledge graph is: Exhibition area layer: includes different exhibition areas in the venue; Exhibit layer: contains exhibits displayed in different exhibition areas; Learning resource layer: contains learning resource collections and learning resources corresponding to different exhibition areas and exhibits; Visitor interaction layer: connected to the exhibition area node through the visit relationship, connected to the exhibit node through the interaction relationship, and connected to the learning resource node through the recommendation relationship; Furthermore, the formula for generating the tourist characteristic index is: ; in, represents the tourist characteristic index of tourist k, represents the highest exhibition area preference index of visitor k among all exhibition areas; when hour, , the exhibition area preference index is dominant, and the corresponding tourist profile is the deep exploration type for their most preferred exhibition area, that is, they have specific goals and explore and learn about their most preferred exhibition area; when hour, , the exploration breadth index is dominant, and the corresponding tourist profile is extensive exploration, that is, there is no specific goal and the exploration range is wide; when When the number of tourists increases, the corresponding tourist portrait is a balanced exploration type, who has a favorite exhibition area while also exploring other exhibition areas.

[0011] Furthermore, the principles for recommending learning resources to tourists are as follows: When the tourist profile of the tourist to be recommended learning resources is a deep exploration type for a certain exhibition area, learning resources are recommended to the tourist based on the tourist's most preferred exhibition area and the learning resource set of the exhibition area in the knowledge graph; When the tourist profile of the tourist to be recommended learning resources is the extensive exploration type, based on the interaction weights of the exhibits in the venue, the relevant learning resources of the top three exhibits with the highest interaction weights are recommended to the tourist; When the tourist profile of the tourist to whom learning resources are recommended is of the balanced exploration type, learning resources are recommended to the tourist in a 1:1 ratio based on the tourist's most preferred exhibition area and the set of learning resources in the exhibition area in the knowledge graph, as well as the top three exhibits with the highest interaction weight in the venue.

[0012] The present invention also provides a learning resource recommendation device based on machine learning, which is used to implement the above-mentioned learning resource recommendation method based on machine learning, and specifically includes: The data collection module is used to identify each visitor entering the venue during the opening hours of the venue, record the total visiting time of each visitor in the venue, the number of exhibition areas visited in the venue and the stay time in different exhibition areas, the number and duration of interaction between the visitor and the exhibits in the exhibition area, and the total number and duration of interaction between each exhibit and all visitors; A visitor preference module is used to generate an interest concentration index based on the time that visitors stay in different exhibition areas and the total time that visitors visit the venue, generate an interaction depth index based on the total number of exhibits that visitors interact with and the number of interaction behaviors and interaction time with each exhibit, generate an exploration breadth index based on the number of exhibition areas visited by visitors and the total number of exhibition areas in the venue, normalize the interest concentration index and the interaction depth index, and generate an exhibition area preference index based on the normalized interest concentration index and the normalized interaction depth index. The exhibition area corresponding to the highest exhibition area preference index is the most preferred exhibition area of ​​the visitor; The feature prediction module is used to build a knowledge graph based on the theme and specific resources of the venue, generate a visitor feature index based on the highest exhibition area preference index and exploration breadth index, build a model based on a deep learning network, use the visitor feature index as input and the visitor portrait as a label, and train the visitor portrait prediction model; The comprehensive output module is used to input the tourist characteristic index of the tourists to be recommended learning resources into the tourist portrait prediction model to generate tourist portraits, and recommend learning resources to tourists based on the tourist portraits and knowledge graphs.

[0013] The present invention also provides a terminal, which includes at least one processor and a memory connected to the processor, wherein: the memory is used to store computer programs or instructions, and the processor is used to execute the computer program or instructions so that the electronic device implements the above-mentioned learning resource recommendation method based on machine learning.

[0014] The present invention also provides a medium, which stores a computer program, and when the computer program is running, the medium is controlled to execute the above-mentioned learning resource recommendation method based on machine learning.

[0015] Compared with the prior art, the present invention has the following beneficial effects: The present invention identifies each visitor entering the venue and records detailed personalized visit behavior data, so that the recommendation of learning resources is more targeted, and can dynamically capture the changes in tourists' interests and adjust the recommendation strategy in time; by combining multiple indicators including visit time and interactive behavior, the recommendation accuracy is improved; by generating an interest concentration index, an interaction depth index and an exploration breadth index, the tourists' behavior is converted into quantifiable indicators, thereby more intuitively reflecting the tourists' interests and preferences, reducing recommendation errors, and ensuring the tourists' acceptance of recommended learning resources.

[0016] The present invention also systematizes information such as exhibition content, exhibit features, and visitor preferences by constructing a knowledge graph based on venue themes and specific resources, providing rich background knowledge for understanding and reasoning, and improving the efficiency of information retrieval and recommendation; combining the exhibition area preference index and the exploration breadth index to generate a visitor characteristic index, it can more carefully depict the interests and behavior patterns of tourists, provide more personalized services and recommendations, and then determine the tourist portrait according to the deep learning network, and recommend learning resources to different passengers based on different tourist portraits, thereby achieving efficient matching of tourists' interest needs with venue resources, improving the personalization and scientificity of learning resource recommendations, and effectively improving the utilization efficiency of venue resources. BRIEF DESCRIPTION OF THE DRAWINGS

[0017] Figure 1 This is a schematic diagram of a method flow of an embodiment of the present invention; Figure 2 Schematic diagram of a device module according to an embodiment of the present invention. DETAILED DESCRIPTION

[0018] In order to make the objectives, technical solutions and advantages of the present invention more clearly understood, the present invention is further described in detail below in conjunction with specific embodiments.

[0019] It should be noted that, unless otherwise defined, the technical terms or scientific terms used in the present invention should be understood by people with ordinary skills in the field to which the present invention belongs. The words "first", "second" and similar words used in the present invention do not indicate any order, quantity or importance, but are only used to distinguish different components. "Include" or "comprise" and similar words mean that the elements or objects appearing before the word include the elements or objects listed after the word and their equivalents, without excluding other elements or objects. "Connect" or "connected" and similar words are not limited to physical or mechanical connections, but may include electrical connections, whether direct or indirect. "Up", "down", "left", "right" and the like are only used to indicate relative positional relationships. When the absolute position of the described object changes, the relative positional relationship may also change accordingly.

[0020] Example: See also Figure 1 , the present invention provides a technical solution: A learning resource recommendation method based on machine learning, the specific steps include: Step 1: During the opening hours of the venue, identify each visitor entering the venue, record the total time each visitor spends in the venue, the number of exhibition areas visited in the venue and the length of time they stay in different exhibition areas, the number and duration of interactions between visitors and exhibits in the exhibition areas, and the total number and duration of interactions between each exhibit and all visitors; In this embodiment, the method for counting the number of interactive behaviors is: touching the display cabinet or control panel of the exhibit for more than 3 seconds is recorded as one interactive behavior, reading the exhibit introduction for more than 5 seconds is recorded as one interactive behavior, and moving around the exhibit within a 1m area around the exhibit for more than 10 seconds or experiencing the exhibit function once is recorded as one interactive behavior. The interactive time is the sum of the time of all interactive behaviors between the visitor and the corresponding exhibit.

[0021] Equip visitors entering the venue with RFID tags, install RFID readers at the entrances and exits of the venue and exhibition areas, and record the time when visitors enter and exit the venue and exhibition areas, so as to obtain the total time of visitors in the venue, the number of exhibition areas visited in the venue, and the length of time they stay in different exhibition areas; for each exhibit, use infrared sensors and video surveillance to collect the interaction between visitors and exhibits in the exhibition area, so as to obtain the number and duration of interaction between visitors and exhibits in the exhibition area.

[0022] Step 2: Generate an interest concentration index based on the time tourists spend in different exhibition areas and their total visit time in the venue; generate an interaction depth index based on the total number of exhibits that tourists interact with and the number and duration of interaction with each exhibit; generate an exploration breadth index based on the number of exhibition areas visited by tourists and the total number of exhibition areas in the venue; normalize the interest concentration index and interaction depth index, and generate an exhibition area preference index based on the normalized interest concentration index and interaction depth index. The exhibition area corresponding to the highest exhibition area preference index is the most preferred exhibition area for the tourist; In this embodiment, the formula for generating the interest concentration index is: ; in, Indicates that tourist k has The interest concentration index of each exhibition area, Indicates the index of the exhibition area. It means that tourist k is in the The duration of stay in each exhibition area, represents the total visiting time of tourist k in the venue, k is the index of the tourist, and , It indicates the total number of visitors during the opening hours of the venue in a week; The interest concentration index reflects the degree of preference of tourists for exhibition areas. The higher the proportion of tourists' stay in a certain exhibition area to their total visiting time in the venue, the higher the tourists' preference for the corresponding exhibition area and the higher the interest concentration index.

[0023] The formula used to generate the Engagement Depth Index is: ; in, Indicates that tourist k has The interactive depth index of each exhibition area, Indicates The number of exhibits in each exhibition area, Indicates The number of exhibits that visitor k interacted with in each exhibition area, represents the index of the exhibits in the exhibition area, and , indicates that tourist k and The exhibition area Interactive time for each exhibit, indicates that tourist k and The exhibition area The number of interactive behaviors for each exhibit, Indicates The exhibition area The interactive weight of the exhibits is The exhibition area The ratio of the product of the total number of interactive behaviors and the total interactive time of each exhibit to the product of the total number of interactive behaviors and the total interactive time of all exhibits; The interaction depth index reflects the degree of interaction between visitors and different exhibits in the exhibition area. The higher the interaction depth index, the more visitors focus on interacting with a few exhibits in the exhibition area. The interaction depth index is directly proportional to the interaction time and number of interactions between visitors and exhibits in the exhibition area, and inversely proportional to the number of exhibits that visitors interact with in the venue. The inverse of the ratio of the number of exhibits interacted by visitor k to the total number of exhibits in the exhibition area, reflecting the breadth of visitor k's exploration in the exhibition area. The higher the value, the smaller the exploration range of visitor k in the exhibition area, and the more visitor k focuses on in-depth interaction with some exhibits; The interaction number, interaction time and interaction weight are combined to measure the interaction between visitor k and the exhibits. The more the total number of interactions and total interaction time between the exhibits and visitors, the higher the interaction depth index.

[0024] The formula used to generate the exploration breadth index is: ; in, represents the exploration breadth index of tourist k, represents the number of exhibition areas visited by visitor k, Indicates the total number of exhibition areas in the venue; The exploration breadth index is used to reflect whether visitors tend to visit multiple exhibition areas in a comprehensive manner or focus on visiting some specific exhibition areas. The number of exhibition areas visited by visitors Indicates the exhibition area in which visitor k has at least one interactive behavior, the total number of exhibition areas in the venue It is a fixed constant. The higher the exploration breadth index is, the wider the scope of tourists' visits is, and they tend to have a comprehensive understanding of the exhibition content of the venue. The lower the exploration breadth index is, the more selective the tourists' tour behavior is, and the more focused their interests are.

[0025] The formula for generating the exhibition area preference index is: ; in, Indicates that tourist k has The exhibition area preference index of each exhibition area, represents the number of visitors k to the first The interest concentration index of each exhibition area, Represents the weight coefficient of the interest concentration index after normalization, represents the number of visitors k to the first The interactive depth index of each exhibition area, Represents the weight coefficient of the interaction depth index after normalization. and .

[0026] The interest concentration index and interaction depth index are normalized using the maximum and minimum normalization method, with the maximum value being the highest among all tourists. The maximum interest concentration index and interaction depth index of each exhibition area, the minimum value among all visitors is The minimum interest concentration index and interaction depth index of each exhibition area.

[0027] The exhibition area preference index is based on the interest concentration index and the interaction depth index, and is proportional to the interest concentration index and the interaction depth index. It reflects the preference of tourists for different exhibition areas during the tour. The exhibition area with the highest exhibition area preference index corresponds to the exhibition area that tourists prefer most during the tour. The exhibition area preference index focuses on evaluating whether tourists have a clear learning purpose when visiting the exhibition area, such as experiencing a certain exhibit. Therefore, the weight coefficient of the interaction depth index is higher than that of the interest concentration index. , .

[0028] Step 3: Build a knowledge graph based on the theme and specific resources of the venue, generate a visitor characteristic index based on the highest exhibition area preference index and exploration breadth index, build a model based on a deep learning network, use the visitor characteristic index as input and the visitor portrait as a label, and train a visitor portrait prediction model; In this embodiment, the principle for constructing the knowledge graph is: The entities in the knowledge graph include exhibition areas, exhibits, learning resources, and visitors; The relationships between entities include exhibits belonging to exhibition areas, learning resources belonging to learning resource collections, exhibition areas being associated with learning resources, exhibits being associated with learning resources, visitors visiting exhibition areas, visitors interacting with exhibits, and learning resources being recommended to visitors; Obtain the name, theme and exhibits of the exhibition area in the venue, obtain detailed information of the exhibits, and associate the existing resources in the venue with the corresponding exhibition area and exhibits; According to the relationship between entities, generate "entity-relationship-entity" triples and import them into the database; The "entity-relationship-entity" triples specifically include: The exhibit belongs to the exhibition area: (Exhibit A, belongs to, exhibition area X); Learning resources belong to the learning resource collection: (Learning Resources , belongs to, learning resources L) Exhibition area associated learning resource set: (exhibition area X, associated, learning resource set L); Exhibit-related learning resources: (Exhibit A, related, learning resources ); Visitors visit exhibition areas: (visitor V, visit, exhibition area X); Visitors interact with exhibits: (visitor V, interaction, exhibit A); Learning resources are recommended to visitors: (Learning resources ,recommended,tourist V); Fill all the resources in the venue according to triples and import them into the database; The hierarchical structure of the knowledge graph is: Exhibition area layer: includes different exhibition areas in the venue; Exhibit layer: contains exhibits displayed in different exhibition areas; Learning resource layer: contains learning resource collections and learning resources corresponding to different exhibition areas and exhibits; Visitor interaction layer: connected to the exhibition area node through the visit relationship, connected to the exhibit node through the interaction relationship, and connected to the learning resource node through the recommendation relationship; The formula used to generate the tourist characteristics index is: ; in, represents the tourist characteristic index, represents the highest exhibition area preference index of visitor k among all exhibition areas; when hour, , the exhibition area preference index is dominant, and the corresponding tourist profile is the deep exploration type for their most preferred exhibition area, that is, they have specific goals and explore and learn about their most preferred exhibition area; when hour, , the exploration breadth index is dominant, and the corresponding tourist profile is extensive exploration, that is, there is no specific goal and the exploration range is wide; when When the number of tourists increases, the corresponding tourist portrait is a balanced exploration type, who has a favorite exhibition area while also exploring other exhibition areas.

[0029] The purpose of the tourist characteristics index is to reflect the exploration breadth index and the highest exhibition area preference index relationship, The corresponding exhibition area is the most preferred exhibition area of ​​tourist k during this visit. When it is high, it means that tourists have a clear preference for an exhibition area during this visit, and have limited attention and preference for other exhibition areas, and the exhibition area with a clear preference is The corresponding exhibition area; When it is high, it means that tourist k is not purposeful enough in this tour and is more likely to conduct a comprehensive tour. It means that tourists have a certain purpose in the visit and have a preferred exhibition area, and also show interest in visiting other exhibition areas.

[0030] The structure of the deep learning network is: Input layer: contains 1 neuron, which is used to input tourist feature index; The first hidden layer: contains 128 neurons, activated by the ReLU function; The second hidden layer: contains 64 neurons, activated by the ReLU function; The third hidden layer: contains 32 neurons, activated by the ReLU function; Output layer: contains 2I+1 neurons, which are used to output tourist portrait types, including deep exploration with a goal, extensive exploration, and balanced exploration with a goal. Indicates the total number of exhibition areas in the venue.

[0031] Step 4: Input the tourist characteristic index of the tourist to be recommended learning resources into the tourist portrait prediction model to generate a tourist portrait, and recommend learning resources to tourists based on the tourist portrait and knowledge graph.

[0032] In this embodiment, the principle for recommending learning resources to visitors to whom learning resources are recommended is as follows: When the tourist profile of the tourist to be recommended learning resources is a deep exploration type for a certain exhibition area, learning resources are recommended to the tourist based on the tourist's most preferred exhibition area and the learning resource set of the exhibition area in the knowledge graph; When the tourist profile of the tourist to be recommended learning resources is the extensive exploration type, based on the interaction weights of the exhibits in the venue, the relevant learning resources of the top three exhibits with the highest interaction weights are recommended to the tourist; When the tourist profile of the tourist to whom learning resources are recommended is of the balanced exploration type, learning resources are recommended to the tourist in a 1:1 ratio based on the tourist's most preferred exhibition area and the set of learning resources in the exhibition area in the knowledge graph, as well as the top three exhibits with the highest interaction weight in the venue.

[0033] When the tourist portrait is a balanced exploratory type, this type of tourist has a clear learning purpose on the one hand, and will also pay attention to other exhibition areas in the venue on the other hand. Therefore, when recommending learning resources to them, we must take into account their preferred exhibition areas and popular exhibition areas in the venue. Based on the tourist’s most preferred exhibition area and the set of learning resources in the exhibition area in the knowledge graph, as well as the top three exhibits in terms of interaction weight in the venue, learning resources are recommended to the tourist in a 1:1 ratio.

[0034] See also Figure 2The present invention also provides a learning resource recommendation device based on machine learning, which is used to implement the above-mentioned learning resource recommendation method based on machine learning, and specifically includes: The data collection module is used to identify each visitor entering the venue during the opening hours of the venue, record the total visiting time of each visitor in the venue, the number of exhibition areas visited in the venue and the stay time in different exhibition areas, the number and duration of interaction between the visitor and the exhibits in the exhibition area, and the total number and duration of interaction between each exhibit and all visitors; A visitor preference module is used to generate an interest concentration index based on the time that visitors stay in different exhibition areas and the total time that visitors visit the venue, generate an interaction depth index based on the total number of exhibits that visitors interact with and the number of interaction behaviors and interaction time with each exhibit, generate an exploration breadth index based on the number of exhibition areas visited by visitors and the total number of exhibition areas in the venue, normalize the interest concentration index and the interaction depth index, and generate an exhibition area preference index based on the normalized interest concentration index and the normalized interaction depth index. The exhibition area corresponding to the highest exhibition area preference index is the most preferred exhibition area of ​​the visitor; The feature prediction module is used to build a knowledge graph based on the theme and specific resources of the venue, generate a visitor feature index based on the highest exhibition area preference index and exploration breadth index, build a model based on a deep learning network, use the visitor feature index as input and the visitor portrait as a label, and train the visitor portrait prediction model; The comprehensive output module is used to input the tourist characteristic index of the tourists to be recommended learning resources into the tourist portrait prediction model to generate tourist portraits, and recommend learning resources to tourists based on the tourist portraits and knowledge graphs.

[0035] The present invention also provides a terminal, which includes at least one processor and a memory connected to the processor, wherein: the memory is used to store computer programs or instructions, and the processor is used to execute the computer program or instructions so that the electronic device implements the above-mentioned learning resource recommendation method based on machine learning.

[0036] The present invention also provides a medium, which stores a computer program, and when the computer program is running, the medium is controlled to execute the above-mentioned learning resource recommendation method based on machine learning.

[0037] The above formulas are all dimensionless and numerical calculations. The formula is a formula for the most recent real situation obtained by collecting a large amount of data and performing software simulation. The preset parameters in the formula are set by technicians in this field according to actual conditions.

[0038] The above embodiments may be implemented in whole or in part by software, hardware, firmware or any other combination thereof. When implemented by software, the above embodiments may be implemented in whole or in part in the form of a computer program product. Those skilled in the art may appreciate that the units and algorithm steps of each example described in conjunction with the embodiments disclosed herein may be implemented by electronic hardware, or a combination of computer software and electronic hardware. Whether these functions are performed by hardware or software methods depends on the specific application and design constraints of the technical solution.

[0039] The units described as separate components may or may not be physically separated, and the components shown as units may or may not be physical units, and may be located in one place or distributed on multiple network units. Some or all of the units may be selected according to actual needs to achieve the purpose of the solution of this embodiment.

[0040] The above description is only a specific implementation manner of the present application, but the protection scope of the present application is not limited thereto. Any technician familiar with the technical field can easily think of changes or substitutions within the technical scope disclosed in the present application, which should be included in the protection scope of the present application.

Claims

1. A learning resource recommendation method based on machine learning, characterized in that: The specific steps include: Step 1: During the opening hours of the venue, identify each visitor entering the venue, record the total time each visitor spends in the venue, the number of exhibition areas visited in the venue and the length of time they stay in different exhibition areas, the number and duration of interactions between visitors and exhibits in the exhibition areas, and the total number and duration of interactions between each exhibit and all visitors; Step 2: Generate an interest concentration index based on the time tourists stay in different exhibition areas and the total time they visit the venue; generate an interaction depth index based on the total number of exhibits that tourists interact with and the number and duration of interaction with each exhibit; generate an exploration breadth index based on the number of exhibition areas visited by tourists and the total number of exhibition areas in the venue; normalize the interest concentration index and the interaction depth index; and generate an exhibition area preference index based on the normalized interest concentration index and the normalized interaction depth index. The exhibition area corresponding to the highest exhibition area preference index is the most preferred exhibition area for the tourist; Step 3: Build a knowledge graph based on the theme and specific resources of the venue, generate a visitor characteristic index based on the highest exhibition area preference index and exploration breadth index, build a model based on a deep learning network, use the visitor characteristic index as input and the visitor portrait as a label, and train a visitor portrait prediction model; Step 4: Input the tourist characteristic index of the tourist to be recommended learning resources into the tourist portrait prediction model to generate a tourist portrait, and recommend learning resources to tourists based on the tourist portrait and knowledge graph.

2. The method for recommending learning resources based on machine learning according to claim 1, characterized in that: The counting method of the number of interactive behaviors in step 1 is: touching the display cabinet or control panel of the exhibit for more than 3 seconds is recorded as one interactive behavior, reading the exhibit introduction for more than 5 seconds is recorded as one interactive behavior, moving around the exhibit in an area of ​​1m around the exhibit for more than 10 seconds or experiencing the function of the exhibit once is recorded as one interactive behavior, and the interactive time is the sum of the time of all interactive behaviors between the visitor and the corresponding exhibit.

3. The method for recommending learning resources based on machine learning according to claim 1, characterized in that: The formula for generating the interest concentration index in step 2 is: ； in, Indicates that tourist k has The interest concentration index of each exhibition area, Indicates the index of the exhibition area. Indicates that tourist k is in the The duration of stay in each exhibition area, represents the total visiting time of tourist k in the venue, k is the index of the tourist, and , It indicates the total number of visitors during the opening hours of the venue in a week; The formula used to generate the Engagement Depth Index is: ； in, Indicates that tourist k has The interactive depth index of each exhibition area, Indicates The number of exhibits in each exhibition area, Indicates The number of exhibits that visitor k interacted with in each exhibition area, represents the index of the exhibits in the exhibition area, and , Indicates that tourist k and The exhibition area Interactive time for each exhibit, Indicates that tourist k and The exhibition area The number of interactive behaviors for each exhibit, Indicates The exhibition area The interactive weight of the exhibits is The exhibition area The ratio of the product of the total number of interactive behaviors and the total interactive time of each exhibit to the product of the total number of interactive behaviors and the total interactive time of all exhibits; The formula used to generate the exploration breadth index is: ； in, represents the exploration breadth index of tourist k, represents the number of exhibition areas visited by visitor k, Indicates the total number of exhibition areas in the venue; The formula for generating the exhibition area preference index is: ； in, Indicates that tourist k has The exhibition area preference index of each exhibition area, represents the number of visitors k to the first The interest concentration index of each exhibition area, Represents the weight coefficient of the interest concentration index after normalization, represents the number of visitors k to the first The interactive depth index of each exhibition area, Represents the weight coefficient of the interaction depth index after normalization. and .

4. The method for recommending learning resources based on machine learning according to claim 1, characterized in that: The principle for constructing the knowledge graph in step 3 is: The entities in the knowledge graph include exhibition areas, exhibits, learning resources, and visitors; The relationships between entities include exhibits belonging to exhibition areas, learning resources belonging to learning resource collections, exhibition areas being associated with learning resource collections, exhibits being associated with learning resources, visitors visiting exhibition areas, visitors interacting with exhibits, and learning resources being recommended to visitors; According to the relationship between entities, generate "entity-relationship-entity" triples and import them into the database; The hierarchical structure of the knowledge graph is: Exhibition area layer: includes different exhibition areas in the venue; Exhibit layer: contains exhibits displayed in different exhibition areas; Learning resource layer: contains learning resource collections and learning resources corresponding to different exhibition areas and exhibits; Visitor interaction layer: connected to the exhibition area node through the visit relationship, connected to the exhibit node through the interaction relationship, and connected to the learning resource node through the recommendation relationship.

5. The method for recommending learning resources based on machine learning according to claim 3, characterized in that: The formula for generating the tourist characteristic index in step 3 is: ； in, represents the tourist characteristic index of tourist k, represents the highest exhibition area preference index of visitor k among all exhibition areas; when hour, , the exhibition area preference index is dominant, and the corresponding tourist profile is the deep exploration type for their most preferred exhibition area, that is, they have specific goals and explore and learn about their most preferred exhibition area; when hour, , the exploration breadth index is dominant, and the corresponding tourist profile is extensive exploration, that is, there is no specific goal and the exploration range is wide; when When the number of tourists increases, the corresponding tourist portrait is a balanced exploration type, who has a favorite exhibition area while also exploring other exhibition areas.

6. The method for recommending learning resources based on machine learning according to claim 3, characterized in that: The principle for recommending learning resources to the visitors to be recommended learning resources in step 4 is: When the tourist profile of the tourist to be recommended learning resources is a deep exploration type for a certain exhibition area, learning resources are recommended to the tourist based on the tourist's most preferred exhibition area and the learning resource set of the exhibition area in the knowledge graph; When the tourist profile of the tourist to be recommended learning resources is the extensive exploration type, based on the interaction weights of the exhibits in the venue, the relevant learning resources of the top three exhibits with the highest interaction weights are recommended to the tourist; When the tourist profile of the tourist to whom learning resources are recommended is of the balanced exploration type, learning resources are recommended to the tourist in a 1:1 ratio based on the tourist's most preferred exhibition area and the set of learning resources in the exhibition area in the knowledge graph, as well as the top three exhibits with the highest interaction weight in the venue.

7. A learning resource recommendation device based on machine learning, characterized in that: The device is used to implement the learning resource recommendation method based on machine learning according to any one of claims 1 to 6, specifically comprising: The data collection module is used to identify each visitor entering the venue during the opening hours of the venue, record the total visiting time of each visitor in the venue, the number of exhibition areas visited in the venue and the stay time in different exhibition areas, the number and duration of interaction between the visitor and the exhibits in the exhibition area, and the total number and duration of interaction between each exhibit and all visitors; A visitor preference module is used to generate an interest concentration index based on the time that visitors stay in different exhibition areas and the total time that visitors visit the venue, generate an interaction depth index based on the total number of exhibits that visitors interact with and the number of interaction behaviors and interaction time with each exhibit, generate an exploration breadth index based on the number of exhibition areas visited by visitors and the total number of exhibition areas in the venue, normalize the interest concentration index and the interaction depth index, and generate an exhibition area preference index based on the normalized interest concentration index and the normalized interaction depth index. The exhibition area corresponding to the highest exhibition area preference index is the most preferred exhibition area of ​​the visitor; The feature prediction module is used to build a knowledge graph based on the theme and specific resources of the venue, generate a visitor feature index based on the highest exhibition area preference index and exploration breadth index, build a model based on a deep learning network, use the visitor feature index as input and the visitor portrait as a label, and train the visitor portrait prediction model; The comprehensive output module is used to input the tourist characteristic index of the tourists to be recommended learning resources into the tourist portrait prediction model to generate tourist portraits, and recommend learning resources to tourists based on the tourist portraits and knowledge graphs.

8. A terminal, characterized in that: The terminal includes at least one processor and a memory connected to the processor, wherein: the memory is used to store computer programs or instructions, and the processor is used to execute the computer program or instructions, so that the terminal implements the learning resource recommendation method based on machine learning as described in any one of claims 1-6.

9. A medium, characterized in that: The medium stores a computer program, and when the computer program is running, the medium is controlled to execute the learning resource recommendation method based on machine learning according to any one of claims 1 to 6.

Images