CN104866557A - Customized just-in-time learning support system and method based on constructivist learning theory - Google Patents
Customized just-in-time learning support system and method based on constructivist learning theory Download PDFInfo
- Publication number
- CN104866557A CN104866557A CN201510252479.2A CN201510252479A CN104866557A CN 104866557 A CN104866557 A CN 104866557A CN 201510252479 A CN201510252479 A CN 201510252479A CN 104866557 A CN104866557 A CN 104866557A
- Authority
- CN
- China
- Prior art keywords
- knowledge
- user
- micro
- relation
- label
- 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.)
- Granted
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
- G06F16/30—Information retrieval; Database structures therefor; File system structures therefor of unstructured textual data
- G06F16/36—Creation of semantic tools, e.g. ontology or thesauri
- G06F16/367—Ontology
-
- 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
- G06F16/95—Retrieval from the web
- G06F16/958—Organisation or management of web site content, e.g. publishing, maintaining pages or automatic linking
Abstract
The present invention provides a customized just-in-time learning support system and method based on constructivist learning theory. The system comprises a user knowledge demand behavior analyzing and recording module, a label relation graph micro-knowledge base, a constructivist learning-based micro-knowledge recommendation algorithm module and the like modules. With the system according to the present invention, construction demands of knowledge are analyzed based on social networking media information of a user, and micro-knowledge for customized learning is recommended to the user, thereby quickly making up for the deficiency of knowledge for the user. The method according to the present invention, as important service content in an Internet-based social networking system, efficiently and intelligently helps a user to improve personal knowledge quality, thereby achieving good economical and social benefits.
Description
Technical field
The invention discloses user knowledge demand analysis in a kind of social networks, the System and method for that personalized instant knowledge is recommended, belong to the field that individualized knowledge is recommended.
Background technology
Nowadays, people's daily life more and more be unable to do without social networks.Single end in October, 2014 with regard to Sina's microblog data display, the microblogging moon active users reached 1.67 hundred million people.People realize Information Communication, Information Sharing, acquisition of information by social networks.Herein with this social networks of Sina's microblogging for platform is described.
Microblogging function is become stronger day by day, and a large amount of micro-blog information increased also result in the problems such as information overload easily bringing simultaneously.The different knowledge of each ken are contained in the microblog data that user receives every day, comprise a large amount of garbage of not theing least concerned with academic knowledge simultaneously, cause the irrelevant information that user needs active filtering a large amount of when " brush microblogging ", screen useful information to read, considerably increase burden during user's brush microblogging.
In addition, for the strange knowledge that some are not contacted, user retrieves relevant webpage by inputting keyword in a search engine, then therefrom selects valuable webpage to read.Therefore user often learns a new knowledge by retrieval screening process loaded down with trivial details for experience, and this process can not guarantee that the content that user finally retrieves is that oneself is interested.Therefore user's " brush microblogging " is one and has a glance at, and planless learning process, the planning of neither one system, this may cause user's repetitive learning, spends the unnecessary time.
How to help microblog users to screen effective information, filtering useless, uninterested information, and analyze the hobby of user on this basis, making personalized study to user and recommend, is a problem demanding prompt solution.
Study commending system to some extent solves problem set forth above.The part but existing study commending system also comes with some shortcomings: 1. can not the learning interest of user in real.2. content recommendation user interest level is not high.3. learning content complex forms, causes learning burden weight, reduces user learning interest.4. not for the comprehensive learning planning of custom system.
For above deficiency, the present invention discloses a kind of personalized instant learning back-up system based on constructive learning theory and method.
Summary of the invention
The present invention is intended to the deficiency making up above-mentioned study commending system, it provides the study commending system immediately recommending individualized learning resource in a kind of social networks according to user knowledge demand to user.The present invention focuses on as user builds the academic environment of property one by one, helps the instant discovery learning point of interest of user, makes up lacking in knowledge, the structure of knowledge that construction is complete.The present invention focuses on consideration two point: 1. in-situ analysis user knowledge demand, finds the knowledge disappearance that user expects, immediately recommends education resource.2. by introducing oriented label graph of a relation, the education resource realized based on construction is recommended.
Personalized instant learning back-up system based on micro-Knowledge Construction provided by the invention and method, comprise the following steps:
User knowledge context analyzer: first by obtaining the microblog label of a large amount of microblogging authenticated, statistical study different crowd is to the hobby of different microblog label, then predict crowd belonging to this user by the microblog label of evaluating objects user, and then target of prediction user is interested in which ken.Then batch obtains user and pays close attention to the up-to-date microblogging state delivered of good friend, is integrated into a microblogging state document by the time order and function delivered, and uses the TF-IDF algorithm improved automatically to extract document keyword.Finally obtain the keyword of one group of Weight, represent the knowledge point of the current concern of user, be the knowledge background of user.
Be built with to label graph of a relation: oriented label graph of a relation stores a kind of inner link existed between all knowledge points, the relation such as comprising namely on knowledge content, mutual exclusion.The structure of knowledge of whole label graph of a relation is a netted structure.The content that label graph of a relation comprises can be divided into several large ken, and the structure of knowledge of each ken is tree structure.Each large ken, by the root node down segmentation from level to level of a knowledge, finally forms a tree structure.
Build expansion knowledge base: expansion knowledge base is to meet the further demand of user to knowledge, when the micro-knowledge of user to system recommendation is interested, the expansion knowledge that system will be recommended to user for this micro-knowledge, this may be a website relevant with this micro-knowledge, or one section of article.
Find the disappearance knowledge that user expects: the knowledge that user acquires while brush microblogging is scattered incomplete, there is the disappearance of knowledge.Oriented label graph of a relation provides a complete knowledge structure system, is equivalent to for user has formulated a comprehensive study plan.The knowledge point of current for user concern is marked in label graph of a relation, but therefrom finds that closely-related with user's current concern knowledge point is the not contacted knowledge of user, recommend user learning, thus the object helping user to reach knowledge leakage detection to fill a vacancy.
Realize recommending based on the education resource of construction: disappearance knowledge user expected, as knowledge to be recommended, recommends user learning.The process of user learning disappearance knowledge is exactly a process building knowledge.User often learns a disappearance knowledge, and the structure of knowledge of user will be more perfect.
Accompanying drawing explanation
Fig. 1 learns recommend method process flow diagram;
Fig. 2 user knowledge context analyzer process flow diagram
Belonging to Fig. 3 user, crowd predicts process flow diagram
Fig. 4 user gender prediction process flow diagram
Fig. 5 obtains targeted customer's microblog label process flow diagram
Fig. 6 age of user prediction process flow diagram
Fig. 7 user's occupation prediction process flow diagram
Fig. 8 user's current concern knowledge analysis process flow diagram
Fig. 9 label graph of a relation
The tree-like branch of Figure 10 label graph of a relation
Figure 11 creates micro-knowledge base process flow diagram
Figure 12 finds the knowledge process figure lacked
Figure 13 system architecture diagram
Figure 14 working-flow figure
Embodiment
Come with reference to the accompanying drawings to be described in detail each step in the present invention.
As shown in Figure 1, the invention provides a kind of personalized instant learning support method based on constructive learning theory, comprise the steps:
User knowledge context analyzer: analyze user knowledge background by the knowledge of crowd belonging to prediction user and the current concern of discovery user.As shown in Figure 2, user knowledge context analyzer comprises the steps:
Crowd's prediction belonging to user: use Bayes' theorem by crowd belonging to microblog label prediction user.As shown in Figure 3, belonging to user, crowd's prediction comprises the following steps:
User gender prediction: predict this user's sex by microblog label.As shown in Figure 4, user gender prediction comprises the following steps:
Obtain sex and the label of a large amount of microblogging authenticated: the API provided by microblogging open platform obtains the information (comprising user's sex, age, occupation, microblog label, the microblogging state delivered) of a large amount of microblogging authenticated.
By sex statistics label: the microblog users label of all acquisitions press masculinity and femininity division, obtain microblog label storehouse that the male sex likes and the microblog label storehouse that women likes.
Expandtabs storehouse: if some keywords occur frequently in the microblogging state of user, then can think that this keyword is the microblog label of this user.Add up the microblogging state that all male user were delivered, by microblogging denoising, participle, statistics word frequency, adds keyword higher for word frequency to microblog label storehouse that the male sex likes as the microblog label that the male sex likes.Add up the microblogging state that all female user were delivered, add keyword higher for word frequency to microblogging mark storehouse that women likes as the microblog label that women likes with same method.
Calculate the probability that each label occurs: the probability that in the microblog label storehouse liked of the microblog label storehouse liked of statistics of male and women, each label occurs respectively.Suppose microblog label storehouse always total Tm the label that the male sex likes, the microblog label storehouse that women likes is total Tw label always, wherein label X occurs Sm time in the microblog label storehouse that the male sex likes, occur Sw time in the microblog label storehouse that women likes, the probability that then label X occurs in the microblog label storehouse that the male sex likes is Sm/Tm, and the probability occurred in the microblog label storehouse that women likes is Sw/Tw.
Obtain the microblog label of targeted customer: by the microblog label of user's self marker, the microblogging that the microblog label of user good friend mark, user deliver and forward finds the label that user likes.As shown in Figure 5, obtain targeted customer's microblog label to comprise the following steps:
Direct acquisition user microblog label: the API provided by microblogging open platform obtains the microblog label of targeted customer.
Pay close attention to good friend by user and obtain microblog label: paying close attention to good friend is that user initiatively selects, and normal conditions and targeted customer have similar interest to like.Obtain the microblog label paying close attention to good friend, the microblog label that the microblog label that the statistics frequency of occurrences is higher is liked as targeted customer.
The microblogging status discovery microblog label delivered by user: if user delivers in the state of forwarding frequently occur some keywords, so thinks that the keyword that these frequently occur is the microblog label that user likes.
Targeted customer gender prediction: use the sex that Bayes' theorem analyses and prediction targeted customer is possible.Suppose that Li is the microblog label that targeted customer likes, then according to Bayes' theorem, can calculate targeted customer when label Li occurs is the probability of the male sex (women).Suppose, masculinity and femininity likes that the probability of some labels is 50%, and formula is as follows:
Above formula analyzes based on a label probability that targeted customer is the male sex (women).The microblog label that generalized case microblog users is liked has several, considers each microblog label that targeted customer likes, final probability calculation formula adopts following formula: (establishing Pi=P (Men|Li))
By above analysis, final acquisition targeted customer is male user and the probability for female user.The relatively size of two probability, if the probability that targeted customer is male user is greater than the probability of targeted customer for women, then thinks that targeted customer is the male sex, otherwise, then think that targeted customer is women.
Age of user is predicted: be illustrated in figure 6 age of user prediction process flow diagram, concrete steps are see user gender prediction.
User's occupation prediction: be illustrated in figure 7 user's occupation prediction process flow diagram, concrete steps are see user gender prediction.
Predict the outcome from crowd belonging to user and know user knowledge interest preference.
User's current concern knowledge point is analyzed: the API provided by microblogging open platform is obtained user and pays close attention to the up-to-date microblogging state delivered of good friend, therefrom analyzes the knowledge point of the current concern of user.As shown in the figure 8, the current concern knowledge analysis of user comprises the following steps:
Obtain user and pay close attention to the up-to-date microblogging state delivered of good friend: the API provided by microblogging open platform is obtained user and pays close attention to the up-to-date microblogging state delivered of good friend, and microblogging state is incorporated in a document according to the time order and function delivered.
Microblog data denoising: remove the invalid information such as emoticon, link, the pet name, topic in microblog data.
Microblogging participle: the ICT segmenter using the Chinese Academy of Sciences to provide carries out participle to microblogging document.
Word segmentation result denoising: remove the adjective in word segmentation result, adverbial word, the invalid informations such as stop words, obtains one group of effective keyword (the corresponding correlated knowledge point of each effective keyword).
Key word analysis: use TF-IDF Algorithm Analysis to calculate the weight of each keyword.To the keyword weights W of i-th in document
icomputing formula as shown in the formula:
Wherein tf
irepresent the word frequency that in document, i-th keyword occurs, max tf
irepresent in document and repeat maximum keywords, N represents the total number of documents of corpus, n
irepresent the number of files comprising this keyword in corpus, D represents the length of entire chapter document, d
irepresent the length of the positional distance document beginning that i-th keyword occurs first.The thought of TF-IDF algorithm comprises 2 points: the first point, and the word frequency that keyword occurs is higher, and its weight is larger; Second point, keyword is more common, and its weight is less.If a keyword seldom occurs in other documents, and frequently occurs in the document, then represent the feature that the document be good at reflecting in this keyword.On the basis of original TF-IDF algorithm, in conjunction with the feature of microblogging state, the present invention considers that corresponding time order and function delivered in each keyword, the keyword that the time of delivering is shorter, its weight is larger.The length delivering the time shows as the distance length that this keyword starts to document in a document.
By user interest preference adjustment keyword weight: by the user knowledge interest preference adjustment knowledge point weight known in advance.Each knowledge point judges whether it is the knowledge point that belonging to this user, crowd likes, if so, then strengthens this knowledge point weight, if not, then weakens this knowledge point weight.Such as, " basketball " is the keyword that male user is liked more, if targeted customer is the male sex, then strengthens the weight of this keyword with corresponding probability, if targeted customer is women, then weakens the weight of keyword with corresponding probability.The final weight calculation formula of keyword is as follows:
Pi represents that targeted customer place crowd selects the probability of this keyword, and Pj represents that different crowd selects the probability of this keyword.On the right of equation, last represents that targeted customer selects comparing of the average probability of this keyword to the select probability of this keyword with crowd.
Finally obtained the keyword of one group of Weight by above step, the corresponding knowledge point of each keyword, these knowledge points are the current knowledge point paid close attention to most of user.
Label graph of a relation and micro-knowledge base: label graph of a relation and micro-knowledge base need manual creation, label graph of a relation realizes the structure of knowledgeization and stores and provide knowledge relation, and micro-knowledge base is for storing micro-knowledge entry of manual creation.Label graph of a relation and micro-knowledge base comprise the following steps:
Be built with to label graph of a relation: exist between knowledge and comprise the relation with mutual exclusion, this knowledge relation and corresponding knowledge point are stored in label graph of a relation.As shown in Figure 9, whole label graph of a relation contains all knowledge points, and comprising and mutex relation of existing between each knowledge point.
Knowledge classification: utilize reptile to capture a large amount of keywords from network, keyword is divided into a few class by affiliated ken, merges synonymous keyword.As shown in Figure 9, five, outer ring dashed circle in figure, each circle represents a ken.
Knowledge delamination: as shown in Figure 10, represents the keyword of each ken with the form of tree structure.The scope of one's knowledge scope that different keywords covers is different.The keyword of root node (ground floor) covers the scope of one's knowledge of whole ken; This ken is subdivided into several knowledge blocks by the keyword of the child node (second layer) inherited by root node; Each knowledge blocks is divided into a series of more detailed knowledge point (third layer) by the child node (keyword) inherited from different knowledge blocks; Down segment by layer successively, until knowledge point can not be segmented, obtain the crucial phrase that can represent the tree structure of a certain ken that has level direction thus again.As shown in Figure 9, two thick dashed line circles represent knowledge delamination, and as can be seen from the figure outer field knowledge is from the inheritance of knowledge of internal layer.
The structure of knowledgeization stores: each node represents a knowledge point, and what exist between the corresponding knowledge point of the subordinate between node and coordination comprises and mutex relation.Realize the structured storage of knowledge.As shown in Figure 9, in figure, the small circle of each solid line represents a knowledge point, and the wire list between small circle shows the contact existed between knowledge point.
Label graph of a relation upgrades automatically: the knowledge point in label graph of a relation and the structure of knowledge in use can constantly retrofits automatically.User in use finds that knowledge point expression is inaccurate can carry out edit-modify, if the knowledge point user finding that there is omission also can create an entry according to given form (content of title+300-500 word) oneself submit to system.
Create micro-knowledge base: each the knowledge point correspondence in label graph of a relation creates a micro-knowledge, forms micro-knowledge base.As shown in figure 11, create micro-knowledge base to comprise the following steps:
The micro-knowledge of search engine retrieving: using the title of micro-knowledge as keyword inputted search engine, obtains the knowledge entry relevant to this micro-knowledge from Baidupedia, wikipedia, Chinese encyclopaedia net.
Integrated searching result: integrated by the result for retrieval of acquisition, merge the content repeated, each knowledge point is integrated into the brief content about 300 words.
Store micro-knowledge: each micro-knowledge take search key as title, the content of integration is knowledge body, is stored in micro-knowledge base.
Find the knowledge of disappearance: help by the knowledge point of user's current interest the knowledge that user finds to lack, and recommend corresponding knowledge point for user, make up the deficiency of knowledge.The representation of knowledge user knowledge point that can not initiatively find of disappearance, but this part knowledge to be the knowledge point paid close attention at present with user closely-related, be very likely the knowledge point needed for user.As shown in figure 12, find that the knowledge of disappearance comprises the following steps:
User knowledge background is marked: user context knowledge (the current knowledge point paid close attention to most) marked in label graph of a relation in label graph of a relation.
Find the knowledge of disappearance: the relation existed between the knowledge point provided by label graph of a relation, find other knowledge points in close relations with the knowledge point of the current concern of user, be the disappearance knowledge that user expects.
Build and recommend: the present invention helps user on the basis of original knowledge background, finds the disappearance knowledge expected, by constantly recommending the knowledge learning disappearance, the knowledge hierarchy that construction is new.Construction and recommendation comprise the following steps:
Determine knowledge to be recommended: the knowledge point that each user pays close attention to can find that one lacks knowledge accordingly, disappearance knowledge top3 user expected is as knowledge to be recommended.
Determine expansion knowledge to be recommended: according to the study feedback of the last login system of user, if user represents interested to certain knowledge point of recommending, then using the expansion knowledge of this knowledge point as expansion knowledge to be recommended.
Recommend knowledge: by knowledge to be recommended and expansion knowledge recommendation to be recommended to user learning.
Construct knowledge: constantly finding by helping user to lack knowledge, recommending user learning, progressively helping user to make up knowledge deficiency, finally realize Knowledge Construction.
The invention also discloses a kind of personalized instant learning back-up system based on constructive learning theory, comprising:
Being system architecture diagram as shown in figure 13, illustrating the composition of each module of native system, is working-flow figure as shown in figure 14, illustrates whole flow process during system works.
Native system is by interface module, and proposed algorithm module, database design module three part forms.
Interface module comprises log-in interface, knowledge learning interface, study feedback interface.
Log-in interface: user relies on the existing account of social platform to log in native system, and system is by the current user logged in of account name unique identification.
Knowledge learning interface: show three interested micro-knowledge of users' possibility, recommend user learning.
Study feedback interface: user can feed back after having learnt the knowledge of recommendation, evaluate the knowledge of recommending, user's system when next login system recommends to expand knowledge accordingly according to the user user that feeds back in the past.
Proposed algorithm module comprises user knowledge context analyzer, finds disappearance knowledge, recommends and construction three part.
User knowledge context analyzer: system obtains social networks mandate, extracts user social contact information, and crowd belonging to prediction user analyzes the knowledge point that user pays close attention to instantly, thus accurately understands the knowledge background of user.
Find disappearance knowledge: using the knowledge of the current concern of user as the knowledge background of user, utilize label graph of a relation to find the knowledge point that user may lack.
Recommend education resource: micro-knowledge that each login system recommends three to lack to user.
Database design module comprises label graph of a relation, micro-knowledge base, expansion knowledge base three part.
Label graph of a relation: store all knowledge points in label graph of a relation, and the contact existed between knowledge point.Record the title of each knowledge point, inherit from which knowledge point, derived from which knowledge point.
Micro-knowledge base: store all knowledge points, each knowledge point stores its title and brief knowledge content.Knowledge title one_to_one corresponding in knowledge titles all in micro-knowledge base and label graph of a relation.
Expansion knowledge base: expansion knowledge base stores all expansion knowledge, when user represents interested to some knowledge points, system can recommend the expansion knowledge for this knowledge point when user logs in next time to user.Expansion knowledge form comprises detailed review, related web site, relevant forum, pertinent texts etc.
The present invention proposes a kind of personalized instant learning back-up system based on constructive learning theory and method, the interested knowledge of user is analyzed from the social information of user, and in this, as the knowledge background of user, contact between the knowledge using label graph of a relation to provide, user is helped to find the knowledge lacked, user learning is recommended with the form of micro-knowledge while user's brush microblogging, for user provides personalized knowledge services, help user to eliminate the time of retrieval knowledge, compensate for the defect of knowledge.
Actual use
Method described in this invention can as of social networks in an internet functional module, also can as the core implementation method of the study website based on text, be intended to recommend personalized learning knowledge to user, guide user to find the learning interest of oneself.
Claims (8)
1. the personalized instant learning back-up system based on constructive learning theory and method characteristic are: learn user knowledge background by analyzing user network social information, quote a label graph of a relation on this basis, knowledge user possessed marks in graph of a relation, the intrinsic knowledge relation of graph of a relation is relied on therefrom to find the knowledge that user may lack, recommend potential disappearance knowledge to make up the defect of knowledge for user by continuous to user, help the knowledge structure system that user's construction one is complete.System adopts the form of micro-knowledge and the mode of guiding to recommend knowledge for user, for the knowledge point of each disappearance of user, first system recommends micro-knowledge that a content is brief to user, user is allowed to have a preliminary understanding to this knowledge point, then according to user, the study of this knowledge point is fed back, recommend corresponding interested expansion knowledge when user's login system next time to user.
2. one kind comprises the steps: based on the personalized instant learning support method of constructive learning theory
User knowledge context analyzer: obtain user and pay close attention to one group of microblog data of good friend up-to-date issue, therefrom analyze current the paid close attention to knowledge point of user, and in this, as the knowledge background of user;
Build label graph of a relation: label graph of a relation is made up of two parts: the relation between knowledge point and knowledge point.Knowledge point: the knowledge point that whole label graph of a relation comprises can be divided into the knowledge of several large class, a large class represents a ken.The knowledge of each large class can be divided into different knowledge blocks again, and the knowledge of each block can be divided into different knowledge points again, and segmentation is always gone down, until can not segment again.The label in each knowledge point represents.Relation: whole label graph of a relation is made up of some tree structures, each tree structure represents the relation existed between all knowledge points that a ken comprises.All tree structures combine the netted label graph of a relation of formation one.Each node of tree represents a knowledge point.The segmentation of Shi Fu knowledge point, sub-knowledge point, they are relation of inclusion.In one tree, the knowledge point of same layer is mutex relation.The knowledge structure system that structure one is complete thus;
Build micro-knowledge base: each label occurred in label graph of a relation represents a micro-knowledge.Micro-knowledge base stores the knowledge point comprised in all label graphs of a relation.Each micro-knowledge package is containing the knowledge content about a title and 300 words.Each label in label graph of a relation and the title one_to_one corresponding of micro-knowledge.The advantage of micro-knowledge is that user can save the time of study, and especially when brush microblogging, user is out of patience goes the content of reading large length in detail, micro-knowledge can ensure user at short notice clearly this knowledge point be oneself interested knowledge.
Build expansion knowledge base: the knowledge content in expansion knowledge base is the expansion to micro-knowledge, utilizes the resource that search engine retrieving is relevant to micro-knowledge, comprises relevant website, relevant forum, relevant article etc., as the expansion knowledge of micro-knowledge.
Find the disappearance knowledge that user expects: be marked in label graph of a relation by interested for user knowledge entry, therefrom find and user's knowledge entry interested other knowledge entry closely-related, be the knowledge that user may lack.
Knowledge recommendation: the knowledge grasped according to user, in conjunction with knowledge structure system, recommends the knowledge of the disappearance of personal expectation for user.
3. the personalized instant learning support method of the micro-knowledge described in right 2, wherein user knowledge context analyzer step comprises:
(1) crowd's prediction belonging to user: the information capturing a large amount of microblogging authenticated, training in advance knows different crowd liking various label.Obtain targeted customer's microblog label, analyze this user and belong to any class crowd.Different crowds is different to the knowledge Interest Measure of different field.
(2) batch obtains user and pays close attention to the microblog data of the up-to-date issue of good friend: the one group of microblog data obtaining user and the up-to-date issue of concern good friend thereof, and the time order and function delivered according to state is integrated into a microblogging state document;
(3) microblog data denoising: the noise removing the current microblog data got, comprises emoticon, link, good friend's pet name, topic etc.;
(4) microblogging participle: adopt segmentation methods that the microblogging state document through denoising is carried out word segmentation processing;
(5) word segmentation result denoising: remove the noise in word segmentation result;
(6) key word analysis: use TF-IDF algorithm to calculate the weight of keyword, introduces the position of keyword appearance, targeted customer place crowd on this basis to the fancy grade of this keyword, finally provides one group of keyword with weight.
(7) user knowledge background modeling: sort to keyword by weight size, first three keyword that weighting is great, represents and is user knowledge background in the knowledge point that user pays close attention at present most.
4. the personalized instant learning support method of the micro-knowledge described in right 2, is wherein built with and comprises to label graph of a relation step:
(1) knowledge classification: utilize reptile to capture the keyword of some from network, by keyword by the classification of affiliated ken, and merges synonymous keyword.
(2) knowledge delamination: the keyword of each ken is represented with the form of tree structure.The scope of one's knowledge scope that different keywords covers is different.The keyword of root node (ground floor) covers the scope of one's knowledge of whole ken; This ken is subdivided into several knowledge blocks by the keyword of the child node (second layer) inherited by root node; Each knowledge blocks is divided into a series of more detailed knowledge point (third layer) by the child node (keyword) inherited from different knowledge blocks; Down segment by layer successively, until knowledge point can not be segmented again, obtain thus one have a level direction and the knowledge frame of relation between each knowledge point in a certain ken can be represented.
(3) structure of knowledgeization stores: each node represents a knowledge point, and what exist between the corresponding knowledge point of the subordinate between node and coordination comprises and mutex relation.Realize the structured storage of knowledge.
(4) label graph of a relation upgrades automatically: the knowledge point in label graph of a relation and the structure of knowledge in use can constantly retrofits automatically.
5. the personalized instant learning support method of the micro-knowledge described in right 2, wherein builds micro-knowledge base step and comprises:
(1) the micro-knowledge of search engine retrieving: using the title of micro-knowledge as keyword inputted search engine, obtains the knowledge entry relevant to this micro-knowledge from Baidupedia, wikipedia, Chinese encyclopaedia net.
(2) integrated searching result: integrated by the result for retrieval of acquisition, merge the content repeated, each knowledge point is integrated into the brief content about 300 words.
(3) micro-knowledge is stored: each micro-knowledge take search key as title, and the content of integration is knowledge body, is stored in micro-knowledge base.
6. the personalized instant learning support method of the micro-knowledge described in right 2, wherein builds expansion knowledge base step and comprises:
(1) retrieval and micro-knowledge related web page: with the title of micro-knowledge for search key, retrieval related web page.
(2) retrieval of content is screened: find and the maximally related content of this micro-knowledge from the webpage that retrieval obtains, comprise the article to this micro-knowledge augmented, the website, forum etc. that are the theme with this micro-knowledge.
(3) Memory Extension knowledge: with the title of micro-knowledge for expanding the title of knowledge, effective content that retrieval obtains, as expansion knowledge body, is stored into expansion knowledge base.
7. the personalized instant learning support method of the micro-knowledge described in right 2, wherein finds that the knowledge step lacked comprises:
(1) obtain user knowledge background: the microblogging state analyzing the up-to-date issue of user, grasp the knowledge point of the current concern of user, in this, as the knowledge background of user.
(2) in label graph of a relation, user knowledge background is marked: marked in label graph of a relation knowledge point current paid close attention to for user.
(3) find the knowledge of disappearance: the relation existed between the knowledge point provided by label graph of a relation, find to be the disappearance knowledge that user expects in other knowledge points that the knowledge point current paid close attention to user is in close relations.
8. the personalized instant learning support method of the micro-knowledge described in right 2, wherein knowledge recommendation step comprises:
(1) determine micro-knowledge to be recommended: using user expect disappearance knowledge first three as micro-knowledge to be recommended.
(2) expansion knowledge to be recommended is determined: according to the study feedback of user's login system in the past, analyze user's knowledge which is recommended interested, for user recommends to expand knowledge accordingly.
(3) knowledge is recommended: by micro-knowledge to be recommended and expansion knowledge recommendation to be recommended to user learning.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201510252479.2A CN104866557B (en) | 2015-05-18 | 2015-05-18 | A kind of personalized instant learning theoretical based on constructive learning supports System and method for |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201510252479.2A CN104866557B (en) | 2015-05-18 | 2015-05-18 | A kind of personalized instant learning theoretical based on constructive learning supports System and method for |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN104866557A true CN104866557A (en) | 2015-08-26 |
| CN104866557B CN104866557B (en) | 2018-03-20 |
Family
ID=53912383
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201510252479.2A Active CN104866557B (en) | 2015-05-18 | 2015-05-18 | A kind of personalized instant learning theoretical based on constructive learning supports System and method for |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN104866557B (en) |
Cited By (10)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| WO2016127922A1 (en) * | 2015-02-12 | 2016-08-18 | 马正方 | Learning apparatus in digital environment |
| CN106503172A (en) * | 2016-10-25 | 2017-03-15 | 天闻数媒科技（湖南）有限公司 | The method and apparatus that learning path recommended by knowledge based collection of illustrative plates |
| CN106651701A (en) * | 2016-12-29 | 2017-05-10 | 山东科技大学 | Learning resource constructing method and device |
| CN107741978A (en) * | 2017-10-13 | 2018-02-27 | 北京中教在线科技有限公司 | A kind of Pushing personality study resource method and its system |
| CN107808014A (en) * | 2017-11-06 | 2018-03-16 | 北京中科智营科技发展有限公司 | A kind of Knowledge Base based on natural language processing |
| WO2019033906A1 (en) * | 2017-08-15 | 2019-02-21 | 上海颐为网络科技有限公司 | Intelligent learning system and method |
| CN109874032A (en) * | 2019-03-07 | 2019-06-11 | 四川长虹电器股份有限公司 | The program special topic personalized recommendation system and method for smart television |
| CN110442670A (en) * | 2019-06-11 | 2019-11-12 | 天津交通职业学院 | A kind of consumer representation generation method based on document indexing |
| TWI679600B (en) * | 2018-02-05 | 2019-12-11 | 多利曼股份有限公司 | System and method for characteristics prediction |
| CN117573985A (en) * | 2024-01-16 | 2024-02-20 | 四川航天职业技术学院(四川航天高级技工学校) | Information pushing method and system applied to intelligent online education system |
Citations (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US20030195834A1 (en) * | 2002-04-10 | 2003-10-16 | Hillis W. Daniel | Automated online purchasing system |
| CN1670727A (en) * | 2004-08-12 | 2005-09-21 | 金德龙 | Knowledge intension based knowledge information retrieval method and system thereof |
| TW200807346A (en) * | 2006-07-17 | 2008-02-01 | Hamastar Technology Co Ltd | Knowledge framework system and method for integrating a knowledge management system with an e-learning system |
| US20120329030A1 (en) * | 2010-01-29 | 2012-12-27 | Dan Joseph Leininger | System and method of knowledge assessment |
-
2015
- 2015-05-18 CN CN201510252479.2A patent/CN104866557B/en active Active
Patent Citations (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US20030195834A1 (en) * | 2002-04-10 | 2003-10-16 | Hillis W. Daniel | Automated online purchasing system |
| CN1670727A (en) * | 2004-08-12 | 2005-09-21 | 金德龙 | Knowledge intension based knowledge information retrieval method and system thereof |
| TW200807346A (en) * | 2006-07-17 | 2008-02-01 | Hamastar Technology Co Ltd | Knowledge framework system and method for integrating a knowledge management system with an e-learning system |
| US20120329030A1 (en) * | 2010-01-29 | 2012-12-27 | Dan Joseph Leininger | System and method of knowledge assessment |
Non-Patent Citations (1)
| Title |
|---|
| 胡大伟: "基于标签协同过滤算法在微博推荐中的研究", 《中国优秀硕士学位论文全文数据库 信息科技辑》 * |
Cited By (15)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| WO2016127922A1 (en) * | 2015-02-12 | 2016-08-18 | 马正方 | Learning apparatus in digital environment |
| US10445660B2 (en) | 2015-02-12 | 2019-10-15 | Zhengfang Ma | Learning apparatus in digital environment |
| CN106503172A (en) * | 2016-10-25 | 2017-03-15 | 天闻数媒科技（湖南）有限公司 | The method and apparatus that learning path recommended by knowledge based collection of illustrative plates |
| CN106651701A (en) * | 2016-12-29 | 2017-05-10 | 山东科技大学 | Learning resource constructing method and device |
| WO2019033906A1 (en) * | 2017-08-15 | 2019-02-21 | 上海颐为网络科技有限公司 | Intelligent learning system and method |
| CN107741978A (en) * | 2017-10-13 | 2018-02-27 | 北京中教在线科技有限公司 | A kind of Pushing personality study resource method and its system |
| CN107808014A (en) * | 2017-11-06 | 2018-03-16 | 北京中科智营科技发展有限公司 | A kind of Knowledge Base based on natural language processing |
| CN107808014B (en) * | 2017-11-06 | 2020-02-21 | 北京中科智营科技发展有限公司 | Knowledge base establishing method based on natural language processing |
| TWI679600B (en) * | 2018-02-05 | 2019-12-11 | 多利曼股份有限公司 | System and method for characteristics prediction |
| CN109874032A (en) * | 2019-03-07 | 2019-06-11 | 四川长虹电器股份有限公司 | The program special topic personalized recommendation system and method for smart television |
| CN109874032B (en) * | 2019-03-07 | 2021-06-22 | 四川长虹电器股份有限公司 | Program topic personalized recommendation system and method for smart television |
| CN110442670A (en) * | 2019-06-11 | 2019-11-12 | 天津交通职业学院 | A kind of consumer representation generation method based on document indexing |
| CN110442670B (en) * | 2019-06-11 | 2023-05-26 | 天津交通职业学院 | Consumer portrait generation method based on text indexing |
| CN117573985A (en) * | 2024-01-16 | 2024-02-20 | 四川航天职业技术学院(四川航天高级技工学校) | Information pushing method and system applied to intelligent online education system |
| CN117573985B (en) * | 2024-01-16 | 2024-04-05 | 四川航天职业技术学院(四川航天高级技工学校) | Information pushing method and system applied to intelligent online education system |
Also Published As
| Publication number | Publication date |
|---|---|
| CN104866557B (en) | 2018-03-20 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN104866557A (en) | Customized just-in-time learning support system and method based on constructivist learning theory | |
| US9785888B2 (en) | Information processing apparatus, information processing method, and program for prediction model generated based on evaluation information | |
| Tatar et al. | From popularity prediction to ranking online news | |
| US9245252B2 (en) | Method and system for determining on-line influence in social media | |
| US10217058B2 (en) | Predicting interesting things and concepts in content | |
| KR101419504B1 (en) | System and method providing a suited shopping information by analyzing the propensity of an user | |
| US20170249389A1 (en) | Sentiment rating system and method | |
| CN111859160B (en) | Session sequence recommendation method and system based on graph neural network | |
| KR102155342B1 (en) | System for providing multi-parameter analysis based commercial service using influencer matching to company | |
| CN103714130A (en) | Video recommendation system and method thereof | |
| CN103473036B (en) | A kind of input method skin method for pushing and system | |
| CN112989169B (en) | Target object identification method, information recommendation method, device, equipment and medium | |
| US20180285748A1 (en) | Performance metric prediction for delivery of electronic media content items | |
| CN112288554B (en) | Commodity recommendation method and device, storage medium and electronic device | |
| CN111429161B (en) | Feature extraction method, feature extraction device, storage medium and electronic equipment | |
| Yigit et al. | Extended topology based recommendation system for unidirectional social networks | |
| Chonwiharnphan et al. | Generating realistic users using generative adversarial network with recommendation-based embedding | |
| Khan et al. | Comparative analysis on Facebook post interaction using DNN, ELM and LSTM | |
| Tareaf et al. | Personality exploration system for online social networks: Facebook brands as a use case | |
| CN113407729B (en) | Judicial-oriented personalized case recommendation method and system | |
| Ashraf et al. | Personalized news recommendation based on multi-agent framework using social media preferences | |
| CN104765752A (en) | Recommending device and method based on user model evolution | |
| CN117056619A (en) | Method and device for determining user behavior characteristics | |
| CN101655853A (en) | Device and method for building model | |
| Balali et al. | A supervised method to predict the popularity of news articles |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| C06 | Publication | ||
| PB01 | Publication | ||
| EXSB | Decision made by sipo to initiate substantive examination | ||
| SE01 | Entry into force of request for substantive examination | ||
| GR01 | Patent grant | ||
| GR01 | Patent grant | ||
| TR01 | Transfer of patent right |
Effective date of registration: 20210723 Address after: 211500 floor 6, building a, hatch Eagle building, No. 99, Tuanjie Road, Jiangbei new area, Nanjing, Jiangsu Patentee after: Jiangsu Guoji Technology Co.,Ltd. Address before: No. 1800 road 214122 Jiangsu Lihu Binhu District City of Wuxi Province Patentee before: Jiangnan University |
|
| TR01 | Transfer of patent right |