Personalized medical course recommendation method and system based on artificial intelligence

Abstract

The invention belongs to the field of medical course recommendation, relates to a data analysis technology, and is used for solving the problems that an existing medical course recommendation system cannot perform dynamic recommendation according to real-time requirements and interest changes of a user and does not have a scientific and reasonable recommendation optimization strategy. The artificial intelligence-based personalized medical course recommendation system comprises a medical course online learning platform, and the medical course online learning platform is in communication connection with a course system building module, a user registration module, a static recommendation module, a dynamic recommendation module, a recommendation monitoring module, a recommendation optimization module and a database; when a user logs in the medical course online learning platform, static recommendation is carried out according to the interest tag of the user, and dynamic recommendation is carried out according to the historical operation record of the user; according to the method, the attention labels of the users can be deeply analyzed, breadth and depth recommendation is realized, and the recommended courses better meet the actual requirements of the users.

Description

TECHNICAL FIELD

[0001] The application belongs to the field of medical course recommendation, and relates to a data analysis technique, in particular to an individualized medical course recommendation method and system based on artificial intelligence. BACKGROUND

[0002] In the digital era, the field of medical education faces many challenges and opportunities. With the development of Internet technology, online learning platforms for medical courses have emerged, providing medical users with rich learning resources.

[0003] The existing course recommendation lacks in-depth analysis and effective use of user behavior data. With the increasing richness of user operations on online learning platforms, such as searching, playing, liking, and collecting, these behaviors contain a wealth of valuable information. However, traditional recommendation systems fail to fully exploit this information to optimize the recommendation results. It is difficult to make dynamic recommendations based on real-time user needs and interest changes.

[0004] In addition, in the face of vast and complex medical course systems, traditional recommendation systems cannot analyze user needs to learn deeper knowledge or more comprehensive knowledge based on user historical operation records. Moreover, there is no scientific and reasonable recommendation optimization strategy, making it difficult to improve the quality of recommendations and affecting user learning experience and learning effectiveness.

[0005] To address the above technical problems, the present application provides a solution. SUMMARY

[0006] The present application aims to provide an individualized medical course recommendation method and system based on artificial intelligence, which can dynamically recommend according to real-time user needs and interest changes, and has a scientific and reasonable recommendation optimization strategy.

[0007] The technical problem to be solved by the present application is how to provide an individualized medical course recommendation method and system based on artificial intelligence that can dynamically recommend according to real-time user needs and interest changes, and has a scientific and reasonable recommendation optimization strategy.

[0008] The object of the present application can be achieved by the following technical solutions:

[0009] An individualized medical course recommendation system based on artificial intelligence includes an online learning platform for medical courses, which is communicatively connected with a course system building module, a user registration module, a static recommendation module, a dynamic recommendation module, a recommendation monitoring module, a recommendation optimization module, and a database.

[0010] The course system building module is configured to build a tag system of medical courses: marking medical course videos on a medical course online learning platform as video objects, obtaining text information of the video objects and extracting video tags; constructing a tag group about the video objects, the tag group including a parent tag and a plurality of child tags corresponding to the parent tag;

[0011] The user registration module is configured to register new users and manage user registration information: users input registration information and register on the medical course online learning platform, the user registration information including the user's name, age, mobile phone number and interest tags; obtaining the user's login times through a database, and generating a static recommendation signal or a dynamic recommendation signal according to the size of the login times;

[0012] The static recommendation module is configured to recommend static videos to users: if the static recommendation module receives a static recommendation signal, the user's interest tags are analyzed and video objects are recommended;

[0013] The dynamic recommendation module is configured to recommend dynamic videos to users: if the dynamic recommendation module receives a dynamic recommendation signal, the user's historical operation records are obtained through a database, the user's historical operation records are analyzed and video objects are recommended;

[0014] The recommendation monitoring module is configured to monitor the effect of video recommendation: a fixed time length of a monitoring period is generated, a recommendation effective coefficient YX of a recommended object in the monitoring period is obtained, and whether optimization of recommendation is needed is determined by comparing the recommendation effective coefficient YX;

[0015] The recommendation optimization module is configured to optimize the process of video recommendation: obtaining a time decay length SS of a recommended object, and optimizing the process of video recommendation according to the time decay length SS.

[0016] Further, the text information includes a video title and subtitle text; a character recognition tool is used to extract the video title, and a speech recognition tool is used to extract the subtitle text; an NLP library is used to extract a plurality of keywords from the text information, the keywords are standardized to obtain standard terms and marked as video tags in combination with a medical knowledge base, and a tag group about the video objects is constructed.

[0017] Further, the parent tag is a video tag extracted and standardized from the video title, and the child tag is a video tag extracted and standardized from the subtitle text.

[0018] Further, the process of judging the recommendation method comprises: when the user logs in the medical course online learning platform for learning, obtaining the login times of the user through the database, when the login times of the user is 1, generating a static recommendation signal and sending the static recommendation signal to the static recommendation module to recommend the video object of the user, when the login times of the user is greater than 1, using the dynamic recommendation module to analyze the historical operation record of the user.

[0019] Further, the specific working process of the static recommendation module comprises the following steps:

[0020] Step S1: obtaining all the interest tags of the user through the database and marking them as recommendation tags, matching the recommendation tags with the parent tags of the video objects in the database one by one, and marking the video objects corresponding to the parent tags matched successfully as to-be-recommended objects;

[0021] Step S2: obtaining the number of likes DZ, the number of collections SC and the average completion rate WB of the to-be-recommended objects; the average completion rate WB is the average value of the ratio of the viewing time of all users on the video object to the total time, and the video data row vector SJ is composed of the number of likes DZ, the number of collections SC and the average completion rate WB;

[0022] Step S3: generating a weight column vector FP, and performing dot product calculation on the video data row vector SJ and the weight column vector FP to obtain a video heat coefficient RD;

[0023] Step S4: arranging all the to-be-recommended objects in the order of video heat coefficient RD from large to small to obtain a video recommendation sequence, marking the first N to-be-recommended objects in the video recommendation sequence as recommended objects and recommending them to the user;

[0024] Step S5: after the user clicks and watches the recommended object, continue to mark the N+1th to-be-recommended object in the video recommendation sequence as a recommended object and recommend it to the user;

[0025] Further, the specific working process of the dynamic recommendation module comprises the following steps:

[0026] Step P1: obtaining the historical operation record of the user through the database; the historical operation record includes searching, playing, liking and collecting; obtaining the attention tags of the user, the attention tags include the video tags input by the user when searching and the parent tags corresponding to the video objects played, liked and collected by the user;

[0027] Step P2: match the attention label with the parent label and all child labels of the video object in the database one by one, if there are two attention labels which are the parent label and the child label of a certain video object respectively, mark the two attention labels as an attention label pair, the attention label pair includes the upper label and the lower label, the upper label is the parent label of the video object, and the lower label is the child label of the video object;

[0028] Step P3: obtain all the attention label pairs and compare them: if there are two or more attention label pairs containing the same upper label, execute step P4 to perform breadth recommendation on the upper label; if there are two attention label pairs, the lower label of the first one is the same as the upper label of the second one, mark the lower label of the second one as a recommended label, and execute step P5 to perform depth recommendation on the recommended label;

[0029] Step P4: obtain the same upper label in the attention label pair and match it with the parent label of the video object in the database one by one, and mark all the child labels corresponding to the matched parent label as recommended labels, continue to match the recommended labels with the parent labels of the video objects in the database one by one, and mark the video objects corresponding to the matched parent labels as recommended objects;

[0030] Step P5: obtain all the recommended labels in the attention label pair, match the recommended labels with the parent labels of the video objects in the database one by one, and mark the video objects corresponding to the matched parent labels as recommended objects;

[0031] Step P6: arrange all the recommended objects in the video recommendation sequence in the order of video heat coefficient RD from large to small, mark the first N recommended objects in the video recommendation sequence as recommended objects and recommend them to the user; after the user clicks and watches the recommended objects, continue to mark the N+1 recommended object in the video recommendation sequence as a recommended object and recommend it to the user.

[0032] Further, the judgment process of whether to need recommendation optimization includes: generating a fixed time length monitoring period, obtaining the total number of recommended objects in the monitoring period and marking it as the recommended number TJ, marking the number of times the user plays, likes or collects the recommended objects in the monitoring period as the operation number CJ, calculating the recommended effective coefficient YX by ratio of the operation number CJ and the recommended number TJ; compare the recommended effective coefficient YX with the preset recommended effective threshold YXmax: if the recommended effective coefficient YX is greater than or equal to the recommended effective threshold YXmax, judge that the recommendation is effective, and no processing is needed; if the recommended effective coefficient YX is less than the recommended effective threshold YXmax, judge that the recommendation is invalid, generate an optimization signal and send it to the recommendation optimization module.

[0033] Furthermore, the process of optimizing video recommendations includes: when recommending video objects to users, the time decay duration SS of the recommended object is calculated by subtracting the time from the time when the user last played, liked, or favorited the recommended object. The time decay duration SS is then compared with a preset time decay threshold SSmax. If the time decay duration SS of the recommended object is greater than or equal to the time decay threshold SSmax, the parent tag of the recommended object is removed from the recommended tags. If the time decay duration SS of the recommended object is less than the time decay threshold SSmax, the parent tag of the recommended object is retained in the recommended tags, and the recommended tags are arranged in ascending order of time decay duration SS to obtain a recommended tag sequence. When recommending videos, the recommended tags are prioritized according to the order of the recommended tags in the recommended tag sequence.

[0034] A personalized medical course recommendation method based on artificial intelligence includes the following steps:

[0035] Step 1: Mark the medical course videos on the online medical learning platform as video objects, and construct a tag group for the video objects. The tag group contains a parent tag and several child tags corresponding to the parent tag.

[0036] Step Two: Users enter their registration information and register on the online medical course learning platform. After successful registration, the user's registration information is sent to the database for storage. When users log in to the online medical course learning platform, the database is used to obtain the number of times the user logs in, and video content is recommended based on the number of logins.

[0037] Step 3: Analyze users' interest tags and recommend video content;

[0038] Step 4: Analyze the user's historical operation records and recommend video content;

[0039] Step 5: Generate a monitoring period of fixed duration, obtain the recommendation effectiveness coefficient YX of the recommended objects within the monitoring period, and determine whether recommendation optimization is needed by comparing the recommendation effectiveness coefficient YX.

[0040] Step 6: When recommending video objects to users, obtain the time decay duration SS of the recommended objects, and optimize the video recommendation process based on the time decay duration SS.

[0041] The present invention has the following beneficial effects:

[0042] 1. The curriculum system construction module lays the foundation for personalized recommendations by building detailed tag groups for medical course videos, including parent tags and child tags;

[0043] 2. The dynamic recommendation module makes full use of various user operation behavior data on the platform and conducts in-depth analysis of historical operation records. This allows for a more accurate grasp of users' learning dynamics, providing rich and valuable data support for personalized recommendations and effectively improving the intelligence and accuracy of the recommendation system.

[0044] 3. The dynamic recommendation module mines user-related tags based on their historical operation records, such as search, playback, likes, and collection behaviors. Through in-depth analysis of these tags, it can not only match parent tags but also identify the relationships between child tags, achieving both breadth and depth of recommendation. This makes recommended courses more aligned with users' ever-changing learning interests and actual needs, thereby increasing users' acceptance of recommended courses and their learning enthusiasm.

[0045] 4. The recommendation optimization module dynamically adjusts recommendation tags by considering the time decay of recommended objects. Tags of courses that have not been followed by users for a long time are removed, and tags of recently followed courses are prioritized. This makes the recommended courses more in line with the user's current interests, continuously optimizes the recommendation process, and improves recommendation effectiveness and user experience. Attached Figure Description

[0046] To more clearly illustrate the technical solutions in the embodiments of the present invention or the prior art, the drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0047] Figure 1 This is a system principle block diagram in Embodiment 1 of the present invention;

[0048] Figure 2 This is a flowchart of the method in Embodiment 2 of the present invention. Detailed Implementation

[0049] The technical solution of the present invention will be clearly and completely described below with reference to the embodiments. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0050] Example 1: As Figure 1 As shown, a personalized medical course recommendation system based on artificial intelligence includes an online medical course learning platform. The online medical course learning platform is communicatively connected to a course system construction module, a user registration module, a static recommendation module, a dynamic recommendation module, a recommendation monitoring module, a recommendation optimization module, and a database.

[0051] The curriculum system construction module is used to build a tagging system for medical courses: It tags medical course videos on online learning platforms as video objects and obtains their text information, including video titles and subtitles. The module uses text recognition tools to extract video titles and speech recognition tools to extract subtitles. It uses an NLP library to extract keywords from the text information, standardizes these keywords using a medical knowledge base to obtain standard terms, and tags them as video tags. It also constructs tag groups for each video object. Each tag group contains several video tags, namely a parent tag and several child tags corresponding to the parent tag. The parent tag is extracted and standardized from the video title, and the child tags are extracted and standardized from the subtitle text. By constructing detailed tag groups for medical course videos, including parent and child tags, the curriculum system construction module lays the foundation for personalized recommendations.

[0052] The user registration module is used to register new users and manage user registration information: When a user registers their identity via a mobile terminal, they enter registration information on the online medical course learning platform and register. The user's registration information includes the user's name, age, mobile phone number, and interest tags, which are video tags entered by the user during registration. After successful registration, the registration and login module sends the user's registration information to the database for storage.

[0053] When a user logs into the online medical course learning platform, the system retrieves the user's login count from the database. When the user's login count is 1, a static recommendation signal is generated and sent to the static recommendation module to recommend video content to the user. When the user's login count is greater than 1, the system uses the dynamic recommendation module to analyze the user's historical operation records.

[0054] The static recommendation module is used to recommend videos to users who are logging in for the first time: if the static recommendation module receives a static recommendation signal, it analyzes the user's interest tags and recommends video objects;

[0055] The specific working process of the static recommendation module includes the following steps:

[0056] Step S1: Obtain all user interest tags from the database and mark them as recommendation tags. Match the recommendation tags with the parent tags of video objects in the database one by one, and mark the video objects corresponding to the successfully matched parent tags as objects to be recommended.

[0057] Step S2: Obtain the number of likes DZ, the number of favorites SC, and the average completion rate WB of the object to be recommended; the average completion rate WB is the average of the ratio of the viewing time of all users on the video object to the total time, and the video data row vector SJ=[WB, SC, DZ] is formed by the number of likes DZ, the number of favorites SC, and the average completion rate WB.

[0058] Step S3: Generate the weighted column vector FP=[k1, k2, k3] T , where k1>k2>k3, and k1+k2+k3=1; the video heat coefficient RD is obtained by performing a dot product between the video data row vector SJ and the weight column vector FP;

[0059] Step S4: Arrange all the objects to be recommended in descending order of video popularity coefficient RD to obtain the video recommendation sequence. Mark the first N objects to be recommended in the video recommendation sequence as recommended objects and recommend them to the user.

[0060] Step S5: After the user clicks on and watches the recommended object, continue to mark the N+1th object to be recommended in the video recommendation sequence as a recommended object and recommend it to the user.

[0061] The dynamic recommendation module is used to recommend videos based on the user's historical operation records: if the dynamic recommendation module receives a dynamic recommendation signal, it analyzes the user's historical operation records and recommends video objects;

[0062] The specific working process of the dynamic recommendation module includes the following steps:

[0063] Step P1: Retrieve the user's historical operation records from the database; historical operation records include search, playback, likes, and favorites; retrieve the user's followed tags, which include the video tags entered by the user when searching and the parent tags corresponding to the video objects played, liked, and favorited;

[0064] Step P2: Match the tags of interest with the parent tags and all child tags of the video objects in the database one by one. If there are two tags of interest that are the parent tag and child tag of a certain video object, mark these two tags of interest as a tag of interest pair. The tag of interest pair includes the parent tag and the child tag. The parent tag is the parent tag of the video object and the child tag is the child tag of the video object.

[0065] Step P3: Obtain all the followed tag pairs and compare them: If there are two or more followed tag pairs that contain the same parent tag, then proceed to step P4 to perform a broad recommendation for that parent tag; if there are two followed tag pairs, and the first one's sub-tag is the same as the second one's parent tag, then mark the second one's sub-tag as a recommended tag and proceed to step P5 to perform a deep recommendation for that recommended tag.

[0066] Step P4: Obtain the same parent tags in the attention tag pairs and match them one by one with the parent tags of the video objects in the database. Mark all child tags corresponding to the successfully matched parent tags as recommended tags. Continue to match the recommended tags with the parent tags of the video objects in the database one by one, and mark the video objects corresponding to the successfully matched parent tags as objects to be recommended.

[0067] Step P5: Obtain all recommended tags from the list of tags you are interested in, match the recommended tags with the parent tags of the video objects in the database one by one, and mark the video objects corresponding to the successfully matched parent tags as objects to be recommended;

[0068] Step P6: Arrange all the videos to be recommended in descending order of their video popularity coefficient RD to obtain a video recommendation sequence. Mark the first N videos to be recommended in the video recommendation sequence as recommended videos and recommend them to the user. After the user clicks on and watches a recommended video, mark the (N+1)th video to be recommended in the video recommendation sequence as a recommended video and recommend it to the user. The dynamic recommendation module mines user-related tags based on the user's historical operation records, such as search, playback, likes, and collection behaviors. Through in-depth analysis of these tags, it can not only match parent tags but also identify the relationships between child tags, achieving both breadth and depth of recommendation. This makes the recommended courses more in line with the user's ever-changing learning interests and actual needs, improving the user's acceptance of the recommended courses and their learning enthusiasm.

[0069] The recommendation monitoring module is used to monitor the effectiveness of video recommendations: It generates a fixed-duration monitoring period, obtains the total number of recommended objects within the monitoring period and marks it as the recommendation number TJ, marks the number of times a user plays, likes, or favorites a recommended object within the monitoring period as the operation number CJ, and calculates the recommendation effectiveness coefficient YX by comparing the operation number CJ with the recommendation number TJ. The recommendation effectiveness coefficient YX is then compared with a preset recommendation effectiveness threshold YXmax: if the recommendation effectiveness coefficient YX is greater than or equal to the recommendation effectiveness threshold YXmax, the recommendation is considered effective and requires no further processing; if the recommendation effectiveness coefficient YX is less than the recommendation effectiveness threshold YXmax, the recommendation is considered invalid, an optimization signal is generated, and the optimization signal is sent to the recommendation optimization module.

[0070] The recommendation optimization module optimizes the video recommendation process: When recommending videos to users, it calculates the time decay duration (SS) of the recommended video by comparing the current time with the time the user last played, liked, or favorited the recommended video. This SS duration is then compared to a preset time decay threshold (SSmax). If the SS duration is greater than or equal to SSmax, the parent tag of the recommended video is removed from the recommended tags. If the SS duration is less than SSmax, the parent tag is retained, and the recommended tags are arranged in ascending order of SS duration to obtain a recommended tag sequence. Videos are then recommended based on the order of the recommended tags in this sequence. By considering the time decay duration of recommended videos, the module dynamically adjusts the recommended tags, removing tags for courses that have not received user attention for a long time and prioritizing tags for recently viewed courses. This ensures that recommended courses better match the user's current interests, continuously optimizing the recommendation process and improving recommendation effectiveness and user experience.

[0071] Example 2: Figure 2 As shown, a personalized medical course recommendation method based on artificial intelligence includes the following steps:

[0072] Step 1: Mark the medical course videos on the online medical learning platform as video objects, and construct a tag group for the video objects. The tag group contains a parent tag and several child tags corresponding to the parent tag.

[0073] Step Two: Users enter their registration information and register on the online medical course learning platform. After successful registration, the user's registration information is sent to the database for storage. When users log in to the online medical course learning platform, the database is used to obtain the number of times the user logs in, and video content is recommended based on the number of logins.

[0074] Step 3: Analyze users' interest tags and recommend video content;

[0075] Step 4: Analyze the user's historical operation records and recommend video content;

[0076] Step 5: Generate a monitoring period of fixed duration, obtain the recommendation effectiveness coefficient YX of the recommended objects within the monitoring period, and determine whether recommendation optimization is needed by comparing the recommendation effectiveness coefficient YX.

[0077] Step 6: When recommending video objects to users, obtain the time decay duration SS of the recommended objects, and optimize the video recommendation process based on the time decay duration SS.

[0078] A personalized medical course recommendation method and system based on artificial intelligence is disclosed. During operation, medical course videos on an online medical learning platform are labeled as video objects, and tag groups for these video objects are constructed. Users register on the online medical learning platform by entering their registration information; upon successful registration, the user's registration information is sent to a database for storage. When a user logs into the online medical learning platform to study, the system retrieves the user's login count from the database and recommends video objects based on the login count.

[0079] The system analyzes users' interest tags and historical operation records to recommend video objects; it generates fixed-duration monitoring periods and determines whether recommendation optimization is needed by comparing recommendation effectiveness coefficients; finally, when recommending video objects to users, it optimizes the video recommendation process based on time decay duration.

[0080] The above description is merely an example and illustration of the structure of the present invention. Those skilled in the art can make various modifications or additions to the specific embodiments described, or use similar methods to replace them, as long as they do not deviate from the structure of the invention or exceed the scope defined in the claims, all of which should fall within the protection scope of the present invention.

[0081] In the description of this specification, references to terms such as "an embodiment," "example," "specific example," etc., indicate that a specific feature, structure, material, or characteristic described in connection with that embodiment or example is included in at least one embodiment or example of the invention. In this specification, illustrative expressions of the above terms do not necessarily refer to the same embodiment or example. Furthermore, the specific features, structures, materials, or characteristics described may be combined in any suitable manner in one or more embodiments or examples.

[0082] The preferred embodiments of the present invention disclosed above are merely illustrative of the invention. These preferred embodiments do not exhaustively describe all details, nor do they limit the invention to any specific implementation. Clearly, many modifications and variations can be made based on the content of this specification. This specification selects and specifically describes these embodiments to better explain the principles and practical applications of the invention, thereby enabling those skilled in the art to better understand and utilize the invention. The invention is limited only by the claims and their full scope and equivalents.

Claims

1. A personalized medical course recommendation system based on artificial intelligence, characterized in that, This includes an online learning platform for medical courses, which is communicatively connected to a course system construction module, a user registration module, a static recommendation module, a dynamic recommendation module, a recommendation monitoring module, a recommendation optimization module, and a database. The curriculum system building module is used to build a tagging system for medical courses: marking medical course videos on the online learning platform as video objects, obtaining the text information of the video objects, and extracting video tags; Construct a tag group for the video object, which contains a parent tag and several child tags corresponding to the parent tag; The user registration module is used to register new users and manage user registration information: users enter registration information and register on the online medical course learning platform. The user's registration information includes the user's name, age, mobile phone number, and interest tags; the system obtains the number of times a user logs in from the database and generates static or dynamic recommendation signals based on the number of logins. The static recommendation module is used to recommend static videos to users: if the static recommendation module receives a static recommendation signal, it analyzes the user's interest tags and recommends video objects; The dynamic recommendation module is used to make dynamic video recommendations to users: if the dynamic recommendation module receives a dynamic recommendation signal, it retrieves the user's historical operation records from the database, analyzes the user's historical operation records, and recommends video objects. The recommendation monitoring module is used to monitor the effect of video recommendation: generate a fixed-duration monitoring period, obtain the recommendation effectiveness coefficient YX of the recommended objects within the monitoring period, and determine whether recommendation optimization is needed by comparing the recommendation effectiveness coefficient YX. The recommendation optimization module is used to optimize the video recommendation process: obtain the time decay duration SS of the recommended object, and optimize the video recommendation process based on the time decay duration SS.

2. The personalized medical course recommendation system based on artificial intelligence according to claim 1, characterized in that, The text information includes the video title and subtitle text; the video title is extracted using a text recognition tool, and the subtitle text is extracted using a speech recognition tool; several keywords are extracted from the text information using an NLP library, and the keywords are standardized using a medical knowledge base to obtain standard terms and be labeled as video tags, and a tag group about the video object is constructed.

3. The personalized medical course recommendation system based on artificial intelligence according to claim 2, characterized in that, The parent tags are video tags extracted and standardized from the video title, and the child tags are video tags extracted and standardized from the subtitle text.

4. The personalized medical course recommendation system based on artificial intelligence according to claim 3, characterized in that, The process of judging the recommendation method includes: when a user logs into the online medical course learning platform to study, the user's login count is obtained through the database. When the user's login count is 1, a static recommendation signal is generated and sent to the static recommendation module to recommend video objects to the user. When the user's login count is greater than 1, the dynamic recommendation module is used to analyze the user's historical operation records.

5. The personalized medical course recommendation system based on artificial intelligence according to claim 4, characterized in that, The specific working process of the static recommendation module includes the following steps: Step S1: Obtain all user interest tags from the database and mark them as recommendation tags. Match the recommendation tags with the parent tags of video objects in the database one by one, and mark the video objects corresponding to the successfully matched parent tags as objects to be recommended. Step S2: Obtain the number of likes DZ, the number of favorites SC, and the average completion rate WB of the object to be recommended; the average completion rate WB is the average of the ratio of the viewing time of all users on the video object to the total time, and the video data row vector SJ is composed of the number of likes DZ, the number of favorites SC, and the average completion rate WB. Step S3: Generate the weight column vector FP, and calculate the video popularity coefficient RD by performing a dot product between the video data row vector SJ and the weight column vector FP; Step S4: Arrange all the objects to be recommended in descending order of video popularity coefficient RD to obtain the video recommendation sequence. Mark the first N objects to be recommended in the video recommendation sequence as recommended objects and recommend them to the user. Step S5: After the user clicks on and watches the recommended object, continue to mark the N+1th object to be recommended in the video recommendation sequence as a recommended object and recommend it to the user.

6. The personalized medical course recommendation system based on artificial intelligence according to claim 5, characterized in that, The specific working process of the dynamic recommendation module includes the following steps: Step P1: Retrieve the user's historical operation records from the database; historical operation records include search, playback, likes, and favorites; retrieve the user's followed tags, which include the video tags entered by the user when searching and the parent tags corresponding to the video objects played, liked, and favorited; Step P2: Match the tags of interest with the parent tags and all child tags of the video objects in the database one by one. If there are two tags of interest that are the parent tag and child tag of a certain video object, mark these two tags of interest as a tag of interest pair. The tag of interest pair includes the parent tag and the child tag. The parent tag is the parent tag of the video object and the child tag is the child tag of the video object. Step P3: Obtain all the followed tag pairs and compare them: If there are two or more followed tag pairs that contain the same parent tag, then proceed to step P4 to perform a broad recommendation for that parent tag; if there are two followed tag pairs, and the first one's sub-tag is the same as the second one's parent tag, then mark the second one's sub-tag as a recommended tag and proceed to step P5 to perform a deep recommendation for that recommended tag. Step P4: Obtain the same parent tags in the attention tag pairs and match them one by one with the parent tags of the video objects in the database. Mark all child tags corresponding to the successfully matched parent tags as recommended tags. Continue to match the recommended tags with the parent tags of the video objects in the database one by one, and mark the video objects corresponding to the successfully matched parent tags as objects to be recommended. Step P5: Obtain all recommended tags from the list of tags you are interested in, match the recommended tags with the parent tags of the video objects in the database one by one, and mark the video objects corresponding to the successfully matched parent tags as objects to be recommended; Step P6: Arrange all the objects to be recommended in descending order of video popularity coefficient RD to obtain the video recommendation sequence. Mark the first N objects to be recommended in the video recommendation sequence as recommended objects and recommend them to the user. After the user clicks on and watches the recommended object, mark the (N+1)th object to be recommended in the video recommendation sequence as a recommended object and recommend it to the user.

7. The personalized medical course recommendation system based on artificial intelligence according to claim 6, characterized in that, The process for determining whether recommendation optimization is needed includes: generating a fixed-duration monitoring period; obtaining the total number of recommended objects within the monitoring period and marking it as the recommendation number TJ; marking the number of times users play, like, or favorite recommended objects within the monitoring period as the operation number CJ; calculating the recommendation effectiveness coefficient YX by comparing the operation number CJ with the recommendation number TJ; comparing the recommendation effectiveness coefficient YX with a preset recommendation effectiveness threshold YXmax: if the recommendation effectiveness coefficient YX is greater than or equal to the recommendation effectiveness threshold YXmax, the recommendation is considered effective and no further processing is required; if the recommendation effectiveness coefficient YX is less than the recommendation effectiveness threshold YXmax, the recommendation is considered invalid, an optimization signal is generated, and the optimization signal is sent to the recommendation optimization module.

8. The personalized medical course recommendation system based on artificial intelligence according to claim 7, characterized in that, The process of optimizing video recommendations includes: when recommending video objects to users, the time decay duration SS of the recommended object is calculated by comparing the current time with the time when the user last played, liked, or favorited the recommended object. The time decay duration SS is then compared with a preset time decay threshold SSmax. If the time decay duration SS of the recommended object is greater than or equal to the time decay threshold SSmax, the parent tag of the recommended object is removed from the recommended tags. If the time decay duration SS of the recommended object is less than the time decay threshold SSmax, the parent tag of the recommended object is retained in the recommended tags, and the recommended tags are arranged in ascending order of time decay duration SS to obtain a recommended tag sequence. When recommending videos, the recommended tags are prioritized according to the order of the recommended tags in the recommended tag sequence.

9. A personalized medical course recommendation method based on artificial intelligence, characterized in that, Includes the following steps: Step 1: Mark the medical course videos on the online medical learning platform as video objects, and construct a tag group for the video objects. The tag group contains a parent tag and several child tags corresponding to the parent tag. Step Two: Users enter their registration information and register on the online medical course learning platform. After successful registration, the user's registration information is sent to the database for storage. When users log in to the online medical course learning platform, the database is used to obtain the number of times the user logs in, and video content is recommended based on the number of logins. Step 3: Analyze users' interest tags and recommend video content; Step 4: Analyze the user's historical operation records and recommend video content; Step 5: Generate a monitoring period of fixed duration, obtain the recommendation effectiveness coefficient YX of the recommended objects within the monitoring period, and determine whether recommendation optimization is needed by comparing the recommendation effectiveness coefficient YX. Step 6: When recommending video objects to users, obtain the time decay duration SS of the recommended objects, and optimize the video recommendation process based on the time decay duration SS.

Images