CN110209899A - Individualized learning resource recommendation system - Google Patents
Individualized learning resource recommendation system Download PDFInfo
- Publication number
- CN110209899A CN110209899A CN201910435651.6A CN201910435651A CN110209899A CN 110209899 A CN110209899 A CN 110209899A CN 201910435651 A CN201910435651 A CN 201910435651A CN 110209899 A CN110209899 A CN 110209899A
- Authority
- CN
- China
- Prior art keywords
- user
- recommendation
- training
- information
- course
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Pending
Links
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F16/00—Information retrieval; Database structures therefor; File system structures therefor
- G06F16/90—Details of database functions independent of the retrieved data types
- G06F16/903—Querying
- G06F16/9035—Filtering based on additional data, e.g. user or group profiles
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F16/00—Information retrieval; Database structures therefor; File system structures therefor
- G06F16/90—Details of database functions independent of the retrieved data types
- G06F16/907—Retrieval characterised by using metadata, e.g. metadata not derived from the content or metadata generated manually
Landscapes
- Engineering & Computer Science (AREA)
- Databases & Information Systems (AREA)
- Theoretical Computer Science (AREA)
- Data Mining & Analysis (AREA)
- Physics & Mathematics (AREA)
- General Engineering & Computer Science (AREA)
- General Physics & Mathematics (AREA)
- Computational Linguistics (AREA)
- Library & Information Science (AREA)
- Information Retrieval, Db Structures And Fs Structures Therefor (AREA)
Abstract
The invention discloses a kind of individualized learning resource recommendation systems, comprising: log database；Course recommending module, for the preference using trained course recommended models prediction user to each course；Log recommending module is researched, for the preference using trained advanced study and training log recommended models prediction user to each advanced study and training log；Information technology skill recommending module, for the preference using trained information technology skill recommended models prediction user to each information technology skill；Activity recommendation module is researched, for researching movable preference to each using trained advanced study and training activity recommendation model prediction user；Resource recommendation module, for the preference using trained resource recommendation model prediction user to each resource；Recommendation results database.The present invention can realize individualized learning function in education cloud platform.
Description
Technical field
The present invention relates to education cloud platform field, in particular to a kind of individualized learning resource recommendation systems.
Background technique
The sequence of original platform Learning Service list, it is basic using defaulting or being successively ranked up according to issuing time, do not have
Consider that user in the factors such as the various actions habit of platform and study preference, leads to the exhibition when recommending course or activity to user
It is now essentially identical to the course of user or movable list, it is mechanical and without personalized difference.Individual character of the original platform to multiple modules
Change and recommend also to provide selection substantially in the form of identical the default list, can not be to use according to the function and content of each module
Family selection best list is shown.
Summary of the invention
In view of this, the present invention is intended to provide a kind of individualized learning resource recommendation system, to be realized in education cloud platform
Individualized learning function.
Specifically, the present invention provides a kind of individualized learning resource recommendation system, comprising: log database, for depositing
Store up user's course selection information, user researches log access information, user information technique and skill browses information, user's advanced study and training activity
Participation information and user resources access information；
Course recommending module, for reading user's course selection information from the log database, and by user's class
Journey selection information is handled to obtain user's course recommendation training data, user's course recommendation training data is input to pre-
If course recommended models be trained, obtain trained course recommended models, the trained course utilized to recommend mould
Type predicts user to the preference of each course, using the high course of preference as recommendation course；
Log recommending module is researched, researches log access information for reading user from the log database, and by institute
It states user's advanced study and training log access information to be handled to obtain user's advanced study and training log recommendation training data, the user is researched into log
Recommend training data to be input to preset advanced study and training log recommended models to be trained, obtains trained advanced study and training log and recommend mould
Type, using the trained advanced study and training log recommended models prediction user to the preference of each advanced study and training log, by preference
High advanced study and training log is as recommendation advanced study and training log；
Information technology skill recommending module, for reading user information technique and skill browsing letter from the log database
Breath, and user information technique and skill browsing information is handled to obtain user information technique and skill recommendation training data,
User information technique and skill recommendation training data is input to preset information technology skill recommended models to be trained, is obtained
To trained information technology skill recommended models, user couple is predicted using the trained information technology skill recommended models
The preference of each information technology skill, using the high information technology skill of preference as recommendation information technique and skill；
Activity recommendation module is researched, for reading user's advanced study and training activity participation information from the log database, and by institute
It states user's advanced study and training activity participation information to be handled to obtain user's advanced study and training activity recommendation training data, by user's advanced study and training activity
Recommend training data to be input to preset advanced study and training activity recommendation model to be trained, obtains trained advanced study and training activity recommendation mould
Type, using the trained advanced study and training activity recommendation model prediction user to each movable preference of advanced study and training, by preference
High advanced study and training activity is as recommendation advanced study and training activity；
Resource recommendation module, for reading user resources access information from the log database, and by institute's user resources
Access information is handled to obtain user resources access recommendation training data, and the user resources are accessed and recommend training data defeated
Enter to preset resource recommendation model and be trained, obtain trained resource recommendation model, utilizes the trained resource
Recommended models predict user to the preference of each resource, using the high resource of preference as recommendation resource；
Recommendation results database, for storing the recommendation course, recommendation advanced study and training log, recommendation information technique and skill, pushing away
It recommends information technology skill and recommends resource.
Further, the course recommending module includes:
First data center, for reading the corresponding first user course selection letter of the first user from the log database
Breath and the corresponding second user course selection information of second user, and the first user course selection information is handled
It obtains first user's course and recommends training data, handled the second user course selection information to obtain second user class
Journey recommends training data；
Course recommendation unit recommends mould for the first user course recommendation training data to be input to preset course
Type is trained, and obtains trained first course recommended models, is predicted using the trained first course recommended models
First user recommends course for the highest preceding n subject of preference as first to the preference of each course；Being also used to will
The second user course recommendation training data is input to preset course recommended models and is trained, and obtains trained second
Course recommended models, using the trained second course recommended models prediction second user to the preference of each course,
Recommend course for the highest preceding n subject of preference as second；It is also used to calculate the first user and recommends course for first
Recommendation, and calculate second user for second recommend course recommendation；It is also used to recommend course for described first
Recommendation and second recommendation course recommendation by first data center be sent to the recommendation results database into
Row storage；
First pass control unit, the data for being transmitted according to first data center control the course and recommend
Unit is acted.
Further, it includes the id information of the first user, the first user choosing that the first user course, which recommends training data,
The id information and the first user of selecting course select the first user the preference information of course；And second user course
Recommending training data includes the id information of second user, and the id information and second user that second user selects course are to second
The preference information of user's selection course；
The preference information p is calculated by following first formula:
P=1+4 × (Min (α, d/c))/α
Wherein d indicates that the first user selects the average clear total duration of course or second user selection course, and c indicates that first uses
Family selects the browsing time of course or second user selection course, and α indicates preset conversion coefficient；
The recommendation r is calculated by following second formula:
R=80+20 × (p_cur-p_min)/(p_max-p_min)
Wherein, p_cur indicates that the first user of prediction selects the first user the preference of course, p_max first
The maximum value of preference in all courses that user recommends, p_min are preference in all courses of the first user recommendation
Minimum value；Alternatively, p_cur indicates preference of the second user of prediction to second user selection course, p_max second
The maximum value of preference in all courses that user recommends, p_min are preference in all courses of second user recommendation
Minimum value.
Further, the advanced study and training log recommended models include:
Log visit is researched for reading corresponding first user of the first user from the log database by second data center
It asks information and the corresponding second user advanced study and training log access information of second user, and first user is researched into log access
Information, which is handled to obtain the first user, to be researched log and recommends training data, by the second user research log access information into
Row processing obtains second user advanced study and training log and recommends training data；First user researches log access information and visits including the first user
The m advanced study and training log asked；Second user advanced study and training log access information includes the m advanced study and training log that second user accessed；
Content similarity unit, the m advanced study and training log for accessing respectively the first user or second user, obtain with
N1/m most like log of each log combines n1/m log to obtain n1 advanced study and training log recommendation results；
Collaborative filtering unit researches the hot topic of each log in log for predicting the first user or second user to m
Degree takes the highest preceding n2 advanced study and training log of popular degree as advanced study and training log recommendation results；N1 is the integer greater than m, and n2 is less than m
Integer；Popular degree h is calculated by following third formula:
H=w1 × mt/ (max (mt, now-t))+w2 × (min (ml, nl))/ml+w3 × ns/5+w4 × (min (mr,
nr))/mr+w5×(min(mc,nc))/mc
Wherein, w1-w5 is weight coefficient, and t is log issuing time, and now is current time, and mt is issuing time coefficient,
Nl is like time, and ml is like time coefficient, and ns is averagely to comment stellar magnitude, and nr is to read number, and mr is to read number system number, and nc is
Number is commented on, mc is comment number system number；
Research log recommendation results acquiring unit, the n for obtaining to content similarity model and collaborative filtering model
=n1+n2 recommendation results are ranked up, and obtain sorted n recommendation results；It is also used to calculate the first user for advanced study and training
The recommendation of log recommendation results, and second user is calculated for the recommendation of advanced study and training log recommendation results；It is also used to institute
It states the first user and second user and is sent to for researching the recommendation of log recommendation results by first data center
The recommendation results database is stored；
First user or second user pass through following 4th formula meter for researching the recommendation r of log recommendation results
It obtains:
R=80+20 × (n-pi)/(n-1)
Wherein, it is the first user or second user that n, which is the first user advanced study and training log number pi that perhaps second user is recommended,
The advanced study and training of recommendation aim at the ranking position by popular degree from high to low in recommendation list day；
Second procedure control unit, the data for being transmitted according to second data center, it is similar to control the content
Degree unit, collaborative filtering unit and advanced study and training log recommendation results acquiring unit are acted.
Further, the information technology skill recommending module includes:
Third data center, for reading the corresponding first user information technology skill of the first user from the log database
Skilful browsing information and the corresponding second user information technology skill of second user browse information, and by first user information
Technique and skill browsing information is handled to obtain the first user information technique and skill recommendation training data, and the second user is believed
Breath technique and skill browsing information is handled to obtain second user information technology skill recommendation training data；
Information technology skill recommendation unit, for the first user information technique and skill recommendation training data to be input to
Preset first information technique and skill recommended models are trained, and obtain trained first information technique and skill recommended models,
Predict the first user to the preference journey of each information technology skill using the trained first information technique and skill recommended models
Degree, using the highest preceding n information technology skill of preference as the first recommendation information technique and skill；It is also used to described second
User information technique and skill recommendation training data is input to preset second information technology skill recommended models and is trained, and obtains
Trained second information technology skill recommended models are predicted using the trained second information technology skill recommended models
Second user is to the preference of each information technology skill, using the highest preceding n information technology skill of preference as second
Recommendation information technique and skill；It is also used to calculate the first user for the recommendation of the first recommendation information technique and skill, and calculates
Recommendation of the second user for the second recommendation information technique and skill；It is also used to pushing away the first recommendation information technique and skill
The recommendation of degree of recommending and the second recommendation information technique and skill is sent to the recommendation results number by first data center
It is stored according to library；Wherein, first information technique and skill recommended models are the collaborative filtering model based on hybrid similarity；Second
Information technology skill recommended models are the popular degree model based on user characteristics；
Third flow control unit, the data for being transmitted according to the third data center, controls the information technology
Skill recommendation unit is operated.
Further, the advanced study and training activity recommendation module includes:
4th data center, for reading user's advanced study and training activity participation information from the log database, and by the use
Family advanced study and training activity participation information is handled to obtain user's advanced study and training activity recommendation training data；
Activity recommendation unit is researched, is lived for user advanced study and training activity recommendation training data to be input to preset advanced study and training
Dynamic recommended models are trained, and obtain trained advanced study and training activity recommendation model, utilize the trained advanced study and training activity recommendation
Model prediction user is described to grind using the high advanced study and training activity of preference as recommendation results to each movable preference of advanced study and training
Repairing activity recommendation unit is the popular degree model based on feature；
4th flow control unit, the data for being transmitted according to the 4th data center, controls the advanced study and training activity
Recommendation unit is operated.
Further, the resource recommendation module includes:
5th data center, for reading the corresponding first user resources access letter of the first user from the log database
Breath and the corresponding second user resource access information of second user, and the first user resources access information is handled
It obtains the first user resources and recommends training data, the second user resource access information is handled to obtain second user money
Recommend training data in source；
Resource recommendation unit, for first user resources recommendation training data to be input to preset resource recommendation mould
Type is trained, and obtains trained first resource recommended models, is predicted using the trained first resource recommended models
First user recommends resource for the highest preceding n resource of preference as first to the preference of each resource；Being also used to will
The second user resource recommendation training data is input to preset resource recommendation model and is trained, and obtains trained second
Resource recommendation model, using the trained Secondary resource recommended models prediction second user to the preference of each resource,
Recommend resource for the highest preceding n resource of preference as second；It is also used to calculate the first user and recommends resource for first
Recommendation, and calculate second user for second recommend resource recommendation；It is also used to recommend resource for described first
Recommendation and second recommendation resource recommendation by first data center be sent to the recommendation results database into
Row storage；
5th flow control unit, the data for being transmitted according to the 5th data center, controls the resource recommendation
Unit is acted.
Individualized learning resource recommendation system of the invention, by combine platform interior to the course of user, advanced study and training log,
Information technology skill, advanced study and training activity, the personalized service demand for researching resource five big systems module, with reference in application machine learning
Collaborative filtering individualized learning Course Recommendation System is realized based on the algorithm of the machine learning such as commending contents.
Detailed description of the invention
It is incorporated into specification and the attached drawing for constituting part of specification shows the embodiment of the present invention, and with
Principle for explaining the present invention together is described.In the drawings, similar appended drawing reference is for indicating similar element.Under
Attached drawing in the description of face is some embodiments of the present invention, rather than whole embodiments.Those of ordinary skill in the art are come
It says, without creative efforts, other drawings may be obtained according to these drawings without any creative labor.
Fig. 1 is a kind of structural block diagram of individualized learning resource recommendation system provided in an embodiment of the present invention.
Specific embodiment
Exemplary embodiments of the present invention are introduced referring now to the drawings, however, the present invention can use many different shapes
Formula is implemented, and is not limited to the embodiment described herein, and to provide these embodiments be at large and fully disclose
The present invention, and the scope of the present invention is sufficiently conveyed to person of ordinary skill in the field.Show for what is be illustrated in the accompanying drawings
Term in example property embodiment is not limitation of the invention.In the accompanying drawings, identical cells/elements use identical attached
Icon note.
Unless otherwise indicated, term (including scientific and technical terminology) used herein has person of ordinary skill in the field
It is common to understand meaning.Further it will be understood that with the term that usually used dictionary limits, should be understood as and its
The context of related fields has consistent meaning, and is not construed as Utopian or too formal meaning.
A kind of individualized learning resource recommendation system provided in an embodiment of the present invention, comprising: log database, for storing
User's course selection information, user research log access information, user information technique and skill browsing information, user's advanced study and training activity ginseng
With information and user resources access information；Wherein, user's course selection information, user research log access information, Yong Huxin
Breath technique and skill browsing information, user's advanced study and training activity participation information and user resources access information can store in different tables
In lattice.
Course recommending module, for reading user's course selection information from the log database, and by user's class
Journey selection information is handled to obtain user's course recommendation training data, user's course recommendation training data is input to pre-
If course recommended models be trained, obtain trained course recommended models, the trained course utilized to recommend mould
Type predicts user to the preference of each course, using the high course of preference as recommendation course；
Log recommending module is researched, researches log access information for reading user from the log database, and by institute
It states user's advanced study and training log access information to be handled to obtain user's advanced study and training log recommendation training data, the user is researched into log
Recommend training data to be input to preset advanced study and training log recommended models to be trained, obtains trained advanced study and training log and recommend mould
Type, using the trained advanced study and training log recommended models prediction user to the preference of each advanced study and training log, by preference
High advanced study and training log is as recommendation advanced study and training log；
Information technology skill recommending module, for reading user information technique and skill browsing letter from the log database
Breath, and user information technique and skill browsing information is handled to obtain user information technique and skill recommendation training data,
User information technique and skill recommendation training data is input to preset information technology skill recommended models to be trained, is obtained
To trained information technology skill recommended models, user couple is predicted using the trained information technology skill recommended models
The preference of each information technology skill, using the high information technology skill of preference as recommendation information technique and skill；
Activity recommendation module is researched, for reading user's advanced study and training activity participation information from the log database, and by institute
It states user's advanced study and training activity participation information to be handled to obtain user's advanced study and training activity recommendation training data, by user's advanced study and training activity
Recommend training data to be input to preset advanced study and training activity recommendation model to be trained, obtains trained advanced study and training activity recommendation mould
Type, using the trained advanced study and training activity recommendation model prediction user to each movable preference of advanced study and training, by preference
High advanced study and training activity is as recommendation advanced study and training activity；
Resource recommendation module, for reading user resources access information from the log database, and by institute's user resources
Access information is handled to obtain user resources access recommendation training data, and the user resources are accessed and recommend training data defeated
Enter to preset resource recommendation model and be trained, obtain trained resource recommendation model, utilizes the trained resource
Recommended models predict user to the preference of each resource, using the high resource of preference as recommendation resource；
Recommendation results database, for storing the recommendation course, recommendation advanced study and training log, recommendation information technique and skill, pushing away
It recommends information technology skill and recommends resource.
A kind of detailed construction of individualized learning resource recommendation system provided in an embodiment of the present invention may refer to shown in Fig. 1
Structural block diagram.Specifically, above-mentioned five modules can respective independent operating, the operation logic of each module is similar.
Each module includes three submodules again, is data center respectively (to distinguish modules, the corresponding number of each module
According to center respectively with the first data center, second data center's (not shown), third data center (not shown),
Four data center's (not shown)s and the 5th data center name (not shown)), Row control is (to distinguish each mould
Block, the corresponding Row control of each module is respectively with first pass control unit, second procedure control unit (not shown),
Three flow control unit (not shown)s, the 4th flow control unit (not shown) and the 5th flow control unit life
Name (not shown)) and corresponding recommended models (unit).When module is run, each data center reads from log database
The recommended models input data and data to be predicted of current block are taken and construct, recommended models, can be to treatment after training
Prediction result is predicted and returned to measured data, and prediction result is finally stored in recommendation results database by each data center.Always
Body process is controlled by each process control module.Each module specific explanations are as follows:
(1) course recommending module
Since model and training data are identical, so elective course and open class are all recommended by course recommending module.Ensemble stream
The operation logic of journey is controlled by first pass control unit, and the process of operation is described more fully below.
(1.1) construct course recommended models training data: the first data center reads data, and (i.e. user's course selection is believed
Breath), and the data set of (User ID, course ID, preference) format is processed data into, as recommended models, (i.e. course is pushed away
Recommend unit) input data.User can calculate the preference p of course by formula 1.
Formula 1:p=1+4 × (Min (α, d/c))/α
Wherein d indicates average clear total duration, and c indicates that browsing time α indicates conversion coefficient.On this data set, it can incite somebody to action
The quartile 600 for averagely browsing duration is used as conversion coefficient α.
(1.2) training course recommended models: course recommendation unit can choose SVD model, and SVD model is user and article
Both sides information MAP is to a hidden semantic space of new joint, while the interactive information between user-article is modeled as this
Inner product in space.The major parameter of SVD model is vector length, can be set to 32 here.
(1.3) prediction ordinary user's elective course scoring: defining ordinary user here is the first user, to there is historical behavior
The user of data can predict its interest；Such as the user occurred simultaneously in the different tables of log database.Prediction is common
When user's elective course scores, the course ID list of acquisition user first (contains an at user option system in course ID list
Column course ID), it is respectively then each user, each door in elective course list is taken as an elective course using SVD model prediction user
The preference of class finally selects the highest preceding n subject of preference as recommendation course.
(1.4) training elective course recommends cold start-up model: what elective course was cold-started model selection is popular degree model.Model
Input data be (User ID, course ID, preference) format data set, i.e., the training constructed in the step (1.1)
Data.Model can calculate average preference's degree (number of users of scoring is participated at total preference) of all course ID, by class
Average preference's degree of journey ID is as the prediction result to current course preference.
(1.5) prediction cold start-up user's elective course scoring: the user of definition cold start-up here is second user, refers to no history
The user of behavioral data, its unpredictable interest.When prediction cold start-up user's elective course scoring, the course ID of user is obtained first
List, is then respectively each user, recommends cold start-up model prediction user to every in elective course list using elective course
The preference of one elective course finally selects the highest preceding n subject of preference as recommendation course.
(1.6) elective course recommendation results are saved: the recommendation results of step (1.3) and step (1.5) being combined, the choosing of formation
Class recommendation results data are repaired, for each user, needs for recommendation results to be normalized, score is converted to 80~100
Section, generate the data shaped like (User ID, article ID, recommendation).The formula that user preference degree is converted into recommendation r is such as public
Shown in formula 2:
Formula 2:r=80+20 × (p_cur-p_min)/(p_max-p_min)
Wherein, for the user of p_cur prediction to the preference of article, p_max is preference in the article of user's recommendation
Maximum value, p_min is the minimum value of preference in article that user recommends, and is stored in recommendation results database.
(1.7) the open class scoring of prediction ordinary user (the first user): first before the open class scoring of prediction ordinary user
First Candidate Recommendation list can be constructed for each user.The Candidate Recommendation list for defining user is to possess identical subject with the user
All courses that all users of term accessed.It then is respectively each user, using SVD model prediction user to candidate
The preference of every a branch of instruction in school in recommendation list finally selects the highest preceding n subject of preference as recommendation course.
(1.8) the open class of training recommends cold start-up model: open class cold start-up model selection is based on user characteristics
Popular degree model.The input data of model is the data set of (User ID, course ID, preference) format and (User ID is learned
Section, term, age bracket, gender) format data set.Defining user characteristics first is (subject-term-age bracket-gender).If
Total subject quantity is n1, and total term quantity is n2, and total age segment number is n3, and total gender number is n4, then the total quantity of feature is
n1×n2×n3×n4.Here we will be divided into 5 age brackets (0~26,26~31,31~41,41~51,51 age
~), gender number takes 2, and subject and the quantity of term can be according to data set adjust automaticallies.Each user only possesses a user spy
Sign.Model can count average preference degree of every a branch of instruction in school under each user characteristics.It is carried out to (User ID, course ID)
When prediction, the corresponding user characteristics of the User ID can be obtained first, then by course ID average preference's journey under the user characteristics
Degree is as user to the prediction result of course preference.
(1.9) prediction cold start-up user (second user) opens class scoring: when the open class of prediction cold start-up user scores, meeting
Then it is respectively each user using the Candidate Recommendation list of step (1.7), recommends cold start-up model prediction using open class
User finally selects the highest preceding n subject of preference to the preference of each open class in Candidate Recommendation list
As recommendation course.
(1.10) open class recommendation results are saved: the recommendation results of step (1.7) and step (1.9) being combined, and are used
Recommendation results are normalized conversion formula in step (1.6), are converted into recommendation, form final opening class and recommend knot
Fruit data are stored in recommendation results database.
(2) researches log recommending module
(2.1) constructs content similarity model training data: it reads data (i.e. user researches log access information), and
Process data into the training dataset shaped like (advanced study and training log ID, item id, content).
(2.2) training content similarity model: when model training, training can be gone one small interior for each item id
Hold similarity model.It is similar that content similarity model can calculate given two contents with the advanced study and training log ID under item id
Degree.The training process of content similarity model are as follows: step 1: each log is segmented, and after removing stop words, generates word frequency
Vector；Step 2: converting tf-idf vector for word frequency vector, and every log is converted to tf-idf vector；Step 3: passing through
The cosine similarity of log tf-idf vector two-by-two is calculated, can get the content similarity matrix of log.
(2.3) constructs collaborative filtering model training data: reading data (i.e. user researches log access information), and will
Training dataset of the data processing forming such as (User ID researches log ID, item id).
(2.4) training collaborative filtering model: collaborative filtering model used herein is the collaborative filtering mould of item-base
Type.The calculating of similarity uses jaccard similarity.It is identical, one small collaboration of training can be gone for each item id
Filter model.When item id determines, model may estimate that user is to the preference of advanced study and training log at current project ID.
(2.5) generates ordinary user's recommendation results: defining ordinary user here is the use in two tables while occurred
Family.First user has historical behavior data, and second user does not have historical behavior data；First user can predict its hobby, the
Its unpredictable interest of two users.As follows, the recommendation results to each user can be generated:
Step 1, n1 recommendation results are obtained using content similarity model: obtains m that user's recent visit is crossed first
Log is researched, then m advanced study and training log is obtained and the most like n1/m of current log using content similarity model respectively
N1 recommendation results can be obtained in these logs combination by log
Step 2, n2 recommendation results are obtained using collaborative filtering model: obtained first with the institute being accessed in item id
There is advanced study and training log as Candidate Recommendation set, then goes prediction user to each in this candidate collection using collaborative filtering model
The preference of a article finally takes the highest preceding n2 advanced study and training log of preference as recommendation results
Step 3, the recommendation results of combined content similarity model and collaborative filtering model: this step will be to two models
The n=n1+n2 recommendation results provided are ranked up, and finally obtain sorted n recommendation results.The foundation of sequence is article
The formula of popular degree, article hot topic degree h is as shown in formula 3
Formula 3:h=w1 × mt/ (max (mt, now-t))+w2 × (min (ml, nl))/ml+w3 × ns/5+w4 × (min
(mr,nr))/mr+w5×(min(mc,nc))/mc
Wherein, w1~w5 is weight coefficient, and t is log issuing time, and now is current time, and mt is issuing time coefficient,
Nl is like time, and ml is like time coefficient, and ns is averagely to comment stellar magnitude, and nr is to read number, and mr is to read number system number, and nc is
Number is commented on, mc is comment number system number.The article hot topic degree of appearance such as (log ID, item id, popular degree) can be constructed.W1~w5,
Mt, ml, mr, mc can be configured in configuration file.
(2.6) training cold start-up recommended models: cold start-up recommended models use the model based on article hot topic degree.
Training data is the article hot topic degree in step (2.5).The principle of model is respectively each item id, is calculated in item id
The popular highest preceding n log of degree.
(2.7) generates cold start-up user's recommendation results: the user of definition cold start-up here is the use only occurred in a table
Family.The highest preceding n log conduct of hot topic degree in the item id can be returned to and push away according to the item id of user by being cold-started recommended models
Recommend result.
(2.8) saves recommendation results: the recommendation results of step (2.5) and step (2.7) being combined, in recommendation results only
It contains according to the sorted log of article hot topic degree, therefore is returned recommendation results using the conversion formula in step formula 4
One changes, and is converted into recommendation r.
Formula 4:r=80+20 × (n-p_i)/(n-1)
Wherein, n is the article number that user recommends, and pi is that ranking position of the article in recommendation list is (popular by article
Degree is from high to low), it is stored in recommendation results database.
(3) information technology skill recommending module
(3.1) tectonic model training data: model training data include two parts, first is that read in data configuration shaped like
The training set data of (information technology skill ID, keyword), second is that reading in data configuration shaped like (User ID, information technology skill
ID training set data) indicates that user browses the information technology skill.
(3.2) the item-base association based on hybrid similarity training information technique and skill recommended models: is used herein
Same filtering model.Similarity uses the hybrid similarity of jaccard similarity and keyword similarity.Hybrid similarity calculates public
Formula is as shown in formula 5:
Formula 5:sim=1/ (1+e^ (- (l-5))) × jaccard_sim+ (1-1/ (1+e^ (- (l-5)))) ×
content_sim
Wherein l indicates that the article number that user's history browsed, jaccard_sim indicate jaccard similarity, content_
Sim indicates keyword similarity.Model can be (User ID, information technology skill ID), go prediction user to the information technology skill
Skilful preference.
(3.3) generates ordinary user's recommendation results: for each user, use information technique and skill recommended models are pre-
The user is surveyed to the preference of information technology skills all in system, and takes the maximum preceding n1 information technology skill of preference to be used as and pushes away
Recommend result.In addition, obtaining issue date away from the information technology skill in modern d days as new article, n2 new article is randomly selected,
It is inserted into recommendation results 1/3 to 2/3 position.Ultimately generate the recommendation results comprising n=n1+n2 article.
(3.4) training cold start-up recommended models: what information technology skill was cold-started model selection is based on user characteristics
Popular degree model.The input data of model is the data set of (User ID, information technology skill ID) format and (User ID is learned
Section, term, age bracket, gender) format data set.Model can count each information technology skill under each user characteristics
Accessed number.When being predicted (User ID, information technology skill ID), the corresponding use of the User ID can be obtained first
Family feature, then the accessed number of the information technology skill ID under the user characteristics is inclined to information technology skill as user
The prediction result of good degree.
(3.5) generates cold start-up user's recommendation results: when generating cold start-up user's recommendation results, for each user,
Use cold start-up recommended models obtain number is accessed under the user characteristics before n information technology skill as recommendation results.
(3.6) saves prediction result: the recommendation results of step (3.3) and step (3.5) being combined, in recommendation results only
Sorted information technology skill is contained, recommendation results are normalized, is converted into recommendation r, is stored in recommendation results number
According to library.
(4) researches activity recommendation module
(4.1) tectonic model training data: training data is broadly divided into two parts, first is that the shape for reading data and constructing
Such as training set of (movable ID, subject, term), second is that the training set shaped like (User ID, movable ID) for reading data and constructing.
The advanced study and training activity being over can be excluded when constructing training set.
(4.2) training advanced study and training activity recommendation model: advanced study and training activity recommendation model selection is the popular degree mould based on feature
Type.Defining user characteristics first is (item id-subject-term).If total subject quantity is n1, total term quantity is n2, total item
Mesh ID quantity is n3, then the total quantity of feature is n1 × n2 × n3.Each user only possesses a user characteristics.Model can count
Each advanced study and training activity under each user characteristics by participation number.Final mask can be calculated in each feature and always be participated in
The highest preceding n advanced study and training activity of number.
(4.3) generates advanced study and training activity recommendation result: for each user, it is special to obtain the corresponding user of user first
Then sign is returned by advanced study and training activity recommendation model and is made in the user characteristics by the highest preceding n1 advanced study and training activity of participation number
For recommendation results.In addition, acquisition activity date Start Date randomly selects n2 as new article away from the advanced study and training activity in modern d days
A new article is inserted into recommendation results 1/3 to 2/3 position.It ultimately generates comprising the n=n1+n2 movable recommendation of advanced study and training
As a result.
(4.4) saves recommendation results: only containing sorted advanced study and training activity in recommendation results, recommendation results are carried out
Normalization is converted into recommendation r, is stored in recommendation results database.
(5) resource recommendation module
(5.1) structure classes recommended models training set: data are read, construction is shaped like (resource ID, resource classification, resource warm
Training set data Men Du).Wherein, resource hot topic degree h can be calculated by formula 6
Formula 6:h=w1 × (min (mx, nx))/mx+w2 × (min (ml, nl))/ml+w3 × ns/5+w4 × (min
(md,nd))/md
Wherein, w1~w4 is weight coefficient, and nl is like time, and mx is download time coefficient, and nx is download time, and nl is
Number is browsed, ml is browsing number system number, and nd is to thumb up number, and md is to thumb up number system number.Ns is to comment stellar magnitude.W1~w4, mx, ml, md
It can be configured in configuration file.
(5.2) training classification recommended models: the basic principle of classification recommended models is, for each resource class, construction
One resource ID list sorted from high to low by resource hot topic degree.List can save resource ID and corresponding resource is popular
Degree.According to given resource type and resource ID, model can return corresponding resource hot topic degree.
(5.3) constructs collaborative filtering model training set: reading data, and processes data into shaped like (User ID, resource
ID, item id) training dataset.
(5.4) training collaborative filtering model: collaborative filtering model used herein is the collaborative filtering mould of item-base
Type.The calculating of similarity uses jaccard similarity.It is identical, one small collaboration of training can be gone for each item id
Filter model.When item id determines, model may estimate that user is to the preference of resource at current project ID.
(5.5) generates ordinary user's recommendation results: as follows, the recommendation results to each user can be generated:
Step 1, use classes recommended models obtain n1 recommendation results: obtaining the m money that user's recent visit is crossed first
Source, and according to accounting p1, p2 ... the pi of each resource class of m resource statistics (i indicates i-th of classification), and calculate each classification
Need the number of resources n1_i=n1 × pi recommended for user；Defining the list of user's Candidate Recommendation is with the user with the money of item id
Source, can structuring user's Candidate Recommendation list；The number of resources and Candidate Recommendation list recommended for user are needed according to each classification,
Use classes recommended models recommend the highest preceding n1_i resource of current class resource hot topic degree for user, are finally combined into n1
A recommendation results, and sort according to resource hot topic degree.
Step 2, obtain n2 recommendation results using collaborative filtering model: defining the list of user's Candidate Recommendation is and the user
With the resource of item id, can structuring user's Candidate Recommendation list, then go prediction user to this time using collaborative filtering model
The preference for selecting each resource recommendation list Nei finally takes the highest preceding n2 resource of preference as recommendation results.
Step 3, in conjunction with the recommendation results of classification recommended models and collaborative filtering model: this step will give two models
N=n1+n2 recommendation results out are combined.Eventually it is ranked up according to resource of the resource hot topic degree to all recommendations,
Generate a sorted resource recommendation result.
(5.6) training cold start-up recommended models: what resource recommendation was cold-started model selection is the hot topic based on user characteristics
Spend model.The input data of model is data set and (User ID, subject, term, the age of (User ID, resource ID) format
Section, gender) format data set.Model can count accessed number of each resource under each user characteristics.Final mould
Type, which can calculate, is always accessed the highest preceding n resource of number in each feature.
(5.7) generates cold start-up user's recommendation results: when generating cold start-up user's recommendation results, for each user,
Use cold start-up recommended models to obtain resource under the user characteristics and is accessed the resource of n before number as recommendation results.
(5.8) saves recommendation results: the recommendation results of step (5.5) and step (5.8) being combined, in recommendation results only
Sorted resource is contained, recommendation results are normalized, is converted into recommendation r, is stored in recommendation results database.
The present embodiment, in the behavioural habits of platform, learns the similarity between preference calculating user, calculating side by user
Method covers collaborative filtering, classification recommendation, content similarity based on text vector etc..The trained model of usage history data
It is predicted according to user and said module or curriculum activity information, recommends the n resource that resource temperature sorts from high to low out；
Under five big application modules, each module is data center, Row control and recommendation respectively again by three sub- module compositions
Model.Module run when, data center read from log database and construct the recommended models input data of current block with
And data to be predicted, recommended models can predict treatment measured data and return to prediction result, finally by counting after training
Prediction result is stored in recommendation results database according to center.Overall procedure is controlled by recommender system process control module.Using
Above-mentioned individualized learning and recommended technology can carry out personalized service in the Behavior preference that platform learns according to user, often
The list seen in platform of position user will no longer be stereotyped second is that thousand people, thousand face personalized customization.After this system
Platform data is subjected to Calculation of the shunted current, off-line model training reduces the pressure that original platform data calls, steady quickly to provide
The table data of user-customized recommended, facilitates research staff to use.
The present invention is described by reference to a small amount of embodiment.However, it is known in those skilled in the art, as
Defined by subsidiary Patent right requirement, in addition to the present invention other embodiments disclosed above equally fall in it is of the invention
In range.
Normally, all terms used in the claims are all solved according to them in the common meaning of technical field
It releases, unless in addition clearly being defined wherein.All references " one/described/be somebody's turn to do [device, component etc.] " are all opened ground
At least one example being construed in described device, component etc., unless otherwise expressly specified.Any method disclosed herein
Step need not all be run with disclosed accurate sequence, unless explicitly stated otherwise.
Claims (7)
1. a kind of individualized learning resource recommendation system characterized by comprising
Log database, for storing user's course selection information, user researches log access information, user information technique and skill
Browse information, user's advanced study and training activity participation information and user resources access information；
Course recommending module for reading user's course selection information from the log database, and user's course is selected
It selects information to be handled to obtain user's course recommendation training data, user's course recommendation training data is input to preset
Course recommended models are trained, and obtain trained course recommended models, pre- using the trained course recommended models
User is surveyed to the preference of each course, using the high course of preference as recommendation course；
Log recommending module is researched, researches log access information for reading user from the log database, and by the use
Family advanced study and training log access information is handled to obtain user's advanced study and training log recommendation training data, and the user is researched log and is recommended
Training data is input to preset advanced study and training log recommended models and is trained, and obtains trained advanced study and training log recommended models, benefit
With the trained advanced study and training log recommended models prediction user to the preference of each advanced study and training log, grind preference is high
Log is repaired as recommendation advanced study and training log；
Information technology skill recommending module browses information for reading user information technique and skill from the log database, and
User information technique and skill browsing information is handled to obtain user information technique and skill recommendation training data, it will be described
User information technique and skill recommendation training data is input to preset information technology skill recommended models and is trained, and is trained
Good information technology skill recommended models, using the trained information technology skill recommended models prediction user to each information
The preference of technique and skill, using the high information technology skill of preference as recommendation information technique and skill；
Activity recommendation module is researched, for reading user's advanced study and training activity participation information from the log database, and by the use
Family advanced study and training activity participation information is handled to obtain user's advanced study and training activity recommendation training data, and the user is researched activity recommendation
Training data is input to preset advanced study and training activity recommendation model and is trained, and obtains trained advanced study and training activity recommendation model, benefit
With the trained advanced study and training activity recommendation model prediction user to the movable preference of each advanced study and training, grind preference is high
The activity of repairing is as recommendation advanced study and training activity；
Resource recommendation module for reading user resources access information from the log database, and institute's user resources is accessed
Information is handled to obtain user resources access recommendation training data, the user resources is accessed, training data is recommended to be input to
Preset resource recommendation model is trained, and obtains trained resource recommendation model, utilizes the trained resource recommendation
Model prediction user is to the preference of each resource, using the high resource of preference as recommending resource；
Recommendation results database, for storing the recommendation course, recommending advanced study and training log, recommendation information technique and skill, recommendation
It ceases technique and skill and recommends resource.
2. individualized learning resource recommendation system as described in claim 1, which is characterized in that the course recommending module packet
It includes:
First data center, for from the log database read the corresponding first user course selection information of the first user with
And the corresponding second user course selection information of second user, and the first user course selection information is handled to obtain
First user's course recommends training data, is handled the second user course selection information to obtain second user course and pushes away
Recommend training data；
Course recommendation unit, for by the first user course recommendation training data be input to preset course recommended models into
Row training, obtains trained first course recommended models, utilizes the trained first course recommended models prediction first
User recommends course for the highest preceding n subject of preference as first to the preference of each course；Being also used to will be described
Second user course recommendation training data is input to preset course recommended models and is trained, and obtains trained second course
Recommended models will be inclined using the trained second course recommended models prediction second user to the preference of each course
The good highest preceding n subject of degree recommends course as second；It is also used to calculate the first user pushing away for the first recommendation course
Degree of recommending, and calculate the recommendation that second user recommends course for second；The recommendation for being also used to recommend course for described first
The recommendation of degree and the second recommendation course is sent to the recommendation results database by first data center and is deposited
Storage；
First pass control unit, the data for being transmitted according to first data center, controls the course recommendation unit
It is acted.
3. individualized learning resource recommendation system as claimed in claim 2, which is characterized in that the first user course is recommended
Training data includes the id information of the first user, and the id information of the first user selection course and the first user are to the first user
Select the preference information of course；And second user course recommend training data include second user id information, second
The id information and second user of user's selection course select second user the preference information of course；
The preference information p is calculated by following first formula:
P=1+4 × (Min (α, d/c))/α
Wherein d indicates that the first user selects the average clear total duration of course or second user selection course, and c indicates the first user choosing
The browsing time of course or second user selection course is selected, α indicates preset conversion coefficient；
The recommendation r is calculated by following second formula:
R=80+20 × (p_cur-p_min)/(p_max-p_min)
Wherein, p_cur indicates that the first user of prediction selects the first user the preference of course, and p_max is the first user
Recommend all courses in preference maximum value, p_min be the first user recommend all courses in preference most
Small value；Alternatively, p_cur indicates the second user of prediction to the preference of second user selection course, p_max is second user
Recommend all courses in preference maximum value, p_min be second user recommend all courses in preference most
Small value.
4. individualized learning resource recommendation system as claimed in claim 3, which is characterized in that the advanced study and training log recommended models
Include:
Log access letter is researched for reading corresponding first user of the first user from the log database by second data center
Breath and the corresponding second user of second user research log access information, and first user is researched log access information
It is handled to obtain the first user advanced study and training log recommendation training data, the second user is researched at log access information
Reason obtains second user advanced study and training log and recommends training data；It includes that the first user accessed that first user, which researches log access information,
M advanced study and training log；Second user advanced study and training log access information includes the m advanced study and training log that second user accessed；
Content similarity unit, the m advanced study and training log for accessing respectively the first user or second user obtain and each day
N1/m most like log of will combines n1/m log to obtain n1 advanced study and training log recommendation results；
Collaborative filtering unit is researched the popular degree of each log in log to m for predicting the first user or second user, is taken
The popular highest preceding n2 advanced study and training log of degree is as advanced study and training log recommendation results；N1 is the integer greater than m, and n2 is whole less than m
Number；Popular degree h is calculated by following third formula:
H=w1 × mt/ (max (mt, now-t))+w2 × (min (ml, nl))/ml+w3 × ns/5+w4 × (min (mr, nr))/
mr+w5×(min(mc,nc))/mc
Wherein, w1-w5 is weight coefficient, and t is log issuing time, and now is current time, and mt is issuing time coefficient, default
Value is that 1, nl is like time, and ml is like time coefficient, and ns is averagely to comment stellar magnitude, and nr is to read number, and mr is to read number system
Number, nc are comment number, and mc is comment number system number；
Research log recommendation results acquiring unit, the n=n1+ for obtaining to content similarity model and collaborative filtering model
N2 recommendation results are ranked up, and obtain sorted n recommendation results；It is also used to calculate the first user to push away advanced study and training log
The recommendation of result is recommended, and calculates second user for the recommendation of advanced study and training log recommendation results；It is also used to described first
User and second user are sent to described push away by first data center for researching the recommendation of log recommendation results
Result database is recommended to be stored；
First user or second user are calculated for researching the recommendation r of log recommendation results by following 4th formula
It arrives:
R=80+20 × (n-pi)/(n-1)
Wherein, it is that the first user or second user are recommended that n, which is the first user advanced study and training log number pi that perhaps second user is recommended,
Advanced study and training aim at the ranking position by popular degree from high to low in recommendation list day；
Second procedure control unit, the data for being transmitted according to second data center, controls the content similarity list
Member, collaborative filtering unit and advanced study and training log recommendation results acquiring unit are acted.
5. individualized learning resource recommendation system as claimed in claim 4, which is characterized in that the information technology skill is recommended
Module includes:
Third data center, it is clear for reading the corresponding first user information technique and skill of the first user from the log database
Look at information and second user corresponding second user information technology skill browsing information, and by the first user information technology
Skill browsing information is handled to obtain the first user information technique and skill recommendation training data, by the second user information skill
Art skill browsing information is handled to obtain second user information technology skill recommendation training data；
Information technology skill recommendation unit, it is default for the first user information technique and skill recommendation training data to be input to
First information technique and skill recommended models be trained, obtain trained first information technique and skill recommended models, utilize
The trained first information technique and skill recommended models predict that the first user, will to the preference of each information technology skill
The highest preceding n information technology skill of preference is as the first recommendation information technique and skill；It is also used to the second user
Information technology skill recommendation training data is input to preset second information technology skill recommended models and is trained, and is trained
The second good information technology skill recommended models utilize the trained second information technology skill recommended models prediction second
User recommends the preference of each information technology skill using the highest preceding n information technology skill of preference as second
Information technology skill；It is also used to calculate the first user for the recommendation of the first recommendation information technique and skill, and calculates second
Recommendation of the user for the second recommendation information technique and skill；It is also used to the recommendation of the first recommendation information technique and skill
And second the recommendation of recommendation information technique and skill the recommendation results database is sent to by first data center
It is stored；Wherein, first information technique and skill recommended models are the collaborative filtering model based on hybrid similarity；Second information
Technique and skill recommended models are the popular degree model based on user characteristics；
Third flow control unit, the data for being transmitted according to the third data center, controls the information technology skill
Recommendation unit is operated.
6. individualized learning resource recommendation system as claimed in claim 5, which is characterized in that the advanced study and training activity recommendation module
Include:
4th data center for reading user's advanced study and training activity participation information from the log database, and the user is ground
The activity participation information of repairing is handled to obtain user's advanced study and training activity recommendation training data；
Activity recommendation unit is researched, is pushed away for user advanced study and training activity recommendation training data to be input to preset advanced study and training activity
It recommends model to be trained, obtains trained advanced study and training activity recommendation model, utilize the trained advanced study and training activity recommendation model
User is predicted to each movable preference of advanced study and training, using the high advanced study and training activity of preference as recommendation results, the advanced study and training are lived
Dynamic recommendation unit is the popular degree model based on feature；
4th flow control unit, the data for being transmitted according to the 4th data center, controls the advanced study and training activity recommendation
Unit is operated.
7. individualized learning resource recommendation system as claimed in claim 6, which is characterized in that the resource recommendation module packet
It includes:
5th data center, for from the log database read the corresponding first user resources access information of the first user with
And the corresponding second user resource access information of second user, and the first user resources access information is handled to obtain
First user resources recommend training data, are handled the second user resource access information to obtain second user resource and push away
Recommend training data；
Resource recommendation unit, for by first user resources recommendation training data be input to preset resource recommendation model into
Row training, obtains trained first resource recommended models, utilizes the trained first resource recommended models prediction first
User recommends resource for the highest preceding n resource of preference as first to the preference of each resource；Being also used to will be described
Second user resource recommendation training data is input to preset resource recommendation model and is trained, and obtains trained Secondary resource
Recommended models will be inclined using the trained Secondary resource recommended models prediction second user to the preference of each resource
The good highest preceding n resource of degree recommends resource as second；It is also used to calculate the first user pushing away for the first recommendation resource
Degree of recommending, and calculate the recommendation that second user recommends resource for second；The recommendation for being also used to recommend resource for described first
The recommendation of degree and the second recommendation resource is sent to the recommendation results database by first data center and is deposited
Storage；
5th flow control unit, the data for being transmitted according to the 5th data center, controls the resource recommendation unit
It is acted.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201910435651.6A CN110209899A (en) | 2019-05-23 | 2019-05-23 | Individualized learning resource recommendation system |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201910435651.6A CN110209899A (en) | 2019-05-23 | 2019-05-23 | Individualized learning resource recommendation system |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN110209899A true CN110209899A (en) | 2019-09-06 |
Family
ID=67788408
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201910435651.6A Pending CN110209899A (en) | 2019-05-23 | 2019-05-23 | Individualized learning resource recommendation system |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN110209899A (en) |
Cited By (6)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN110929163A (en) * | 2019-12-09 | 2020-03-27 | 上海复深蓝软件股份有限公司 | Course recommendation method and device, computer equipment and storage medium |
| CN111428138A (en) * | 2020-03-26 | 2020-07-17 | 中国建设银行股份有限公司 | Course recommendation method, system, equipment and storage medium |
| CN112214670A (en) * | 2020-10-09 | 2021-01-12 | 平安国际智慧城市科技股份有限公司 | Online course recommendation method and device, electronic equipment and storage medium |
| CN113407848A (en) * | 2021-07-15 | 2021-09-17 | 广州汇思信息科技股份有限公司 | Online learning recommendation method and device based on big data and computer equipment |
| CN113641908A (en) * | 2021-08-17 | 2021-11-12 | 中国联合网络通信集团有限公司 | Course pushing method, device, server and computer storage medium |
| CN114652131A (en) * | 2022-03-24 | 2022-06-24 | 慕思健康睡眠股份有限公司 | Control method and device of intelligent mattress, electronic equipment and storage medium |
Citations (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN105718582A (en) * | 2016-01-25 | 2016-06-29 | 重庆邮电大学 | Personalized learning resource recommendation system and method under E-learning platform |
| CN109213863A (en) * | 2018-08-21 | 2019-01-15 | 北京航空航天大学 | A kind of adaptive recommended method and system based on learning style |
| CN109388746A (en) * | 2018-09-04 | 2019-02-26 | 四川文轩教育科技有限公司 | A kind of education resource intelligent recommendation method based on learner model |
| CN109582875A (en) * | 2018-12-17 | 2019-04-05 | 武汉泰乐奇信息科技有限公司 | A kind of personalized recommendation method and system of online medical education resource |
-
2019
- 2019-05-23 CN CN201910435651.6A patent/CN110209899A/en active Pending
Patent Citations (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN105718582A (en) * | 2016-01-25 | 2016-06-29 | 重庆邮电大学 | Personalized learning resource recommendation system and method under E-learning platform |
| CN109213863A (en) * | 2018-08-21 | 2019-01-15 | 北京航空航天大学 | A kind of adaptive recommended method and system based on learning style |
| CN109388746A (en) * | 2018-09-04 | 2019-02-26 | 四川文轩教育科技有限公司 | A kind of education resource intelligent recommendation method based on learner model |
| CN109582875A (en) * | 2018-12-17 | 2019-04-05 | 武汉泰乐奇信息科技有限公司 | A kind of personalized recommendation method and system of online medical education resource |
Non-Patent Citations (2)
| Title |
|---|
| 牛亚男: ""教学资源个性化推荐系统的设计和实现"", 《万方》 * |
| 陈云: ""学习资源混合推荐策略的研究与实现"", 《中国优秀硕士学位论文全文数据库 信息科技辑》 * |
Cited By (8)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN110929163A (en) * | 2019-12-09 | 2020-03-27 | 上海复深蓝软件股份有限公司 | Course recommendation method and device, computer equipment and storage medium |
| CN110929163B (en) * | 2019-12-09 | 2020-10-02 | 上海复深蓝软件股份有限公司 | Course recommendation method and device, computer equipment and storage medium |
| CN111428138A (en) * | 2020-03-26 | 2020-07-17 | 中国建设银行股份有限公司 | Course recommendation method, system, equipment and storage medium |
| CN112214670A (en) * | 2020-10-09 | 2021-01-12 | 平安国际智慧城市科技股份有限公司 | Online course recommendation method and device, electronic equipment and storage medium |
| CN113407848A (en) * | 2021-07-15 | 2021-09-17 | 广州汇思信息科技股份有限公司 | Online learning recommendation method and device based on big data and computer equipment |
| CN113641908A (en) * | 2021-08-17 | 2021-11-12 | 中国联合网络通信集团有限公司 | Course pushing method, device, server and computer storage medium |
| CN113641908B (en) * | 2021-08-17 | 2023-06-20 | 中国联合网络通信集团有限公司 | Course pushing method, course pushing device, server and computer storage medium |
| CN114652131A (en) * | 2022-03-24 | 2022-06-24 | 慕思健康睡眠股份有限公司 | Control method and device of intelligent mattress, electronic equipment and storage medium |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN110209899A (en) | Individualized learning resource recommendation system | |
| Yeh et al. | Rural politics in contemporary China | |
| Benson | News media as a “journalistic field”: What Bourdieu adds to new institutionalism, and vice versa | |
| Hout et al. | The persistence of classes in post-industrial societies | |
| Bach et al. | Personalized recommendation of stories for commenting in forum-based social media | |
| McKean | The unseen hand in government | |
| Quental et al. | Sustainability: characteristics and scientific roots | |
| CN101354714B (en) | Method for recommending problem based on probability latent semantic analysis | |
| Chen et al. | Combining factorization model and additive forest for collaborative followee recommendation | |
| Hooshmand Khaligh et al. | A mathematical model for vehicle routing problem under endogenous uncertainty | |
| Zeleny | Knowledge of enterprise: knowledge management or knowledge technology | |
| Kuang et al. | Product platform design for a product family based on Kansei engineering | |
| Brandt et al. | On the discriminative power of tournament solutions | |
| Ahmed et al. | Ranking ideas for diversity and quality | |
| Newman et al. | Ethnic change, concern over immigration, and approval of state government | |
| Liu et al. | A new decision support model in multi-criteria decision making with intuitionistic fuzzy sets based on risk preferences and criteria reduction | |
| Kang et al. | How do cultural classes emerge from assimilation and distinction? An extension of the Cucker-Smale flocking model | |
| Al-Ghossein et al. | Open data in the hotel industry: leveraging forthcoming events for hotel recommendation | |
| Kamwa | On stable rules for selecting committees | |
| Kotras | Opinions that matter: the hybridization of opinion and reputation measurement in social media listening software | |
| Kondratev et al. | A ranking procedure with the Shapley value | |
| Parackal et al. | Value-based prediction of election results using natural language processing: A case of the New Zealand General Election | |
| Streviniotis et al. | Preference aggregation mechanisms for a tourism-oriented bayesian recommender | |
| Blume et al. | Dynamic coordination via organizational routines | |
| Grüne-Yanoff et al. | Preference change: An introduction |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| PB01 | Publication | ||
| PB01 | Publication | ||
| SE01 | Entry into force of request for substantive examination | ||
| SE01 | Entry into force of request for substantive examination | ||
| RJ01 | Rejection of invention patent application after publication | ||
| RJ01 | Rejection of invention patent application after publication |
Application publication date: 20190906 |