CN110928920A - Knowledge recommendation method, system and storage medium based on improved position social contact - Google Patents
Knowledge recommendation method, system and storage medium based on improved position social contact Download PDFInfo
- Publication number
- CN110928920A CN110928920A CN201911134890.4A CN201911134890A CN110928920A CN 110928920 A CN110928920 A CN 110928920A CN 201911134890 A CN201911134890 A CN 201911134890A CN 110928920 A CN110928920 A CN 110928920A
- Authority
- CN
- China
- Prior art keywords
- knowledge
- user
- recommendation
- learning
- data
- 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
Images
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
- G06F16/95—Retrieval from the web
- G06F16/953—Querying, e.g. by the use of web search engines
- G06F16/9535—Search customisation based on user profiles and personalisation
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F16/00—Information retrieval; Database structures therefor; File system structures therefor
- G06F16/20—Information retrieval; Database structures therefor; File system structures therefor of structured data, e.g. relational data
- G06F16/24—Querying
- G06F16/245—Query processing
- G06F16/2458—Special types of queries, e.g. statistical queries, fuzzy queries or distributed queries
- G06F16/2462—Approximate or statistical queries
-
- 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
- G06F16/953—Querying, e.g. by the use of web search engines
- G06F16/9536—Search customisation based on social or collaborative filtering
Landscapes
- Engineering & Computer Science (AREA)
- Databases & Information Systems (AREA)
- Theoretical Computer Science (AREA)
- Physics & Mathematics (AREA)
- General Engineering & Computer Science (AREA)
- General Physics & Mathematics (AREA)
- Data Mining & Analysis (AREA)
- Probability & Statistics with Applications (AREA)
- Fuzzy Systems (AREA)
- Mathematical Physics (AREA)
- Software Systems (AREA)
- Computational Linguistics (AREA)
- Information Retrieval, Db Structures And Fs Structures Therefor (AREA)
Abstract
The invention discloses a knowledge recommendation method, a knowledge recommendation system and a storage medium based on improved position social contact.A knowledge point which needs to be accessed by a user in the next step is predicted by utilizing the existing access data of the user and other data information in a knowledge network, and a prediction result is recommended to the user as a learning recommendation list of the user, so that the current situation that a common commodity-oriented recommendation model and algorithm are not suitable for recommendation on a knowledge learning system can be solved, the knowledge acquisition efficiency of the user on the knowledge learning system is improved, and the learning efficiency of the user is improved; meanwhile, the defect that the conventional common recommendation system model and algorithm mainly recommend commercial articles is overcome; and a path set of the user access knowledge points is constructed, knowledge recommendation in a knowledge social network is realized through an improved position social recommendation algorithm, and the learning and understanding efficiency of the user for certain knowledge is improved. The invention can be applied to the field of knowledge recommendation systems.
Description
Technical Field
The invention relates to the field of recommendation systems, in particular to a knowledge recommendation method, a knowledge recommendation system and a storage medium based on improved social contact.
Background
In the era of information explosion, people face the problem that data cannot be obtained from the internet any more, but the people need to face massive information on the internet. How to find truly valuable information in mass data has also become a popular research direction in the information age. Currently, the information retrieval method commonly used by users is generally a search engine, such as Google, Baidu, and the like. The common search engine has the advantages of wide application range, simple information retrieval function meeting most users and the like, but under the condition that the users have specific requirements or the users have personalized and diversified difference characteristics, the search engine cannot capture the complex characteristics of the users so as to perform personalized recommendation. Therefore, in order to meet the personalized requirements of users and improve the efficiency of users in rapidly retrieving information from mass data, concepts and applications of recommendation systems are provided. On the application of content or target service in a specific field, a personalized recommendation system is constructed by using a recommendation algorithm, the potential implicit requirements of a user can be effectively explored, and the content really needed or interested by the user is recommended to the user, so that the user experience is improved, and the stickiness of the user to the system is enhanced. However, current recommendation systems are generally directed to business application fields, such as commodity recommendation, music recommendation, movie recommendation, social user recommendation, and the like.
In the current commercial recommendation, the system generally consists of three parts, namely a user model, a product model and a recommendation algorithm, which are also called model triplets. The user model is used for acquiring, representing and storing browsing behavior and purchase history data of the user; the product model is used for representing and storing the characteristic attributes of the product; the recommendation algorithm is a core part of a recommendation system, and the interest preference and the consumption habit of a user are obtained mainly by mining rules contained in user data. Different recommendation algorithms focus on different fields, and currently mainstream recommendation algorithms include content-based recommendation, collaborative filtering-based recommendation, a hybrid recommendation algorithm, social network-based recommendation, and location interest point-based recommendation which is promoted along with the development of network technology and GPS (Global Positioning System) technology.
The current common recommendation system mainly aims at the business field, and the recommended content is generally physical commodities (such as electronic products, household articles, clothes and daily articles and the like) or virtual digital and conceptual articles (such as music works, movies, tourist sites, red restaurants and the like) in an intangible way. But in addition to the needs of purchasing goods, sharing music and the like, the other mental requirement of people in daily life is to acquire knowledge. With the continuous development of network technology and the deepening of informatization level, people are also more and more accustomed to developing network-based learning through networks. In massive learning resources, users also need a recommendation system to help the users to realize efficient learning, so that a knowledge recommendation-oriented system is very necessary to construct and develop recommendation algorithm research based on knowledge recommendation, and the recommendation system is an effective supplement to the fields of recommendation algorithms and recommendation systems which mainly adopt commercial recommendation at present.
Disclosure of Invention
In order to solve at least one of the above technical problems, an object of the present invention is to provide a knowledge recommendation method, system and storage medium based on improved location socialization.
The technical scheme adopted by the invention is as follows: in one aspect, an embodiment of the present invention includes a knowledge recommendation method based on improved location socialization, including:
extracting data required by knowledge recommendation, wherein the data comprises evaluation data of knowledge by a user;
converting implicit evaluation data in the extracted evaluation data into explicit evaluation data;
calculating the similarity between users according to a similarity algorithm;
selecting a plurality of similar user data according to the similarity between the users;
constructing a knowledge point distance matrix, fusing the similar user data on the knowledge point distance matrix, and constructing a similar user and knowledge point learning path set;
combining a plurality of similar users with a knowledge point learning path set to form a candidate path set;
respectively calculating the probability of each similar user accessing the corresponding knowledge point according to the candidate path set;
and generating a corresponding knowledge point recommendation list according to the calculated probability.
Further, the data required by the knowledge recommendation also comprises user set data, knowledge set data and interaction data among users.
Further, the evaluation data comprises implicit evaluation data and explicit evaluation data, and after the implicit evaluation data is converted into the explicit evaluation data, a user learning scoring matrix is constructed according to the evaluation data obtained after the conversion and the interaction data among the users.
Further, the knowledge recommendation method further comprises the following steps:
and respectively extracting the average scores made by the users, knowledge point sets corresponding to the same scores and the scores of the users on the same learning knowledge points according to the user learning score matrix.
Further, the knowledge recommendation method further comprises the following steps:
and calculating the similarity between the users according to the extracted average scores made by the users, the knowledge point sets corresponding to the same scores and the scores of the users on the same knowledge point.
Further, the similarity calculation is performed by the following formula:
in the formula, RU，VSet of learning knowledge points, r, representing the same score for user U and user VU,iRepresents the score, r, of the user U on the learning knowledge point iV,iRepresents the score of the user V for the learning knowledge point i.
Further, the similarity algorithm may also be performed by the following formula:
in the formula, RU，VSet of learning knowledge points, r, representing the same score for user U and user VU,iRepresents the score, r, of the user U on the learning knowledge point iV,iRepresents the score of the user V for the learning knowledge point i,
represents the average score of the learning knowledge points by the user U,
represents the average score of the user V for learning knowledge points.
Further, the probability of each of the similar users reaching the corresponding knowledge point for learning is calculated by the following formula:
where ρ is the probability of the user accessing the corresponding knowledge point,
learning the eigenvalue vectors of the path sets for the user and knowledge points,
the feature weight in the feature value vector.
In another aspect, an embodiment of the present invention further includes a knowledge recommendation system based on improved location social, including:
the extraction module is used for extracting data required by knowledge recommendation, and the data user evaluates the knowledge;
the conversion module is used for converting implicit evaluation data in the extracted evaluation data into explicit evaluation data;
the similarity calculation module is used for calculating the similarity between the users according to a similarity calculation method;
the selecting module is used for selecting a plurality of similar user data according to the similarity between the users;
the construction module is used for constructing a knowledge point distance matrix, fusing the similar user data on the knowledge point distance matrix and constructing a similar user and knowledge point learning path set;
the merging module is used for merging the similar users and the knowledge point learning path set to form a candidate path set;
the probability calculation module is used for respectively calculating the probability of each similar user accessing the corresponding knowledge point according to the candidate path set;
and the generating module is used for generating a corresponding knowledge point recommendation list according to the calculated probability.
In another aspect, the embodiment of the present invention further includes a storage medium, in which processor-executable instructions are stored, and when executed by a processor, the processor-executable instructions are used for executing the knowledge recommendation method based on improving location socialization.
The invention has the beneficial effects that: the method and the device predict the knowledge points which need to be accessed by the user in the next step by using the existing access data of the user and other data information in the knowledge network, and recommend the prediction results to the user as the learning recommendation list, so that the current situation that a common commodity-oriented recommendation model and algorithm are not suitable for recommending on a knowledge learning system can be solved, the knowledge acquisition efficiency of the user on the knowledge learning system is improved, and the learning efficiency of the user is improved; meanwhile, the method also makes up the defect that the current common recommendation system model and algorithm mainly recommend commercial articles, and mainly generates recommendation contents for a knowledge system, so that a user can understand related knowledge contents which can be further learned through system recommendation under the condition that a certain knowledge content is not known; the method comprises the steps of converting behavior data of a user on knowledge into explicit scores, measuring the understanding degree of the user on current knowledge data through the scores, virtualizing the distance of knowledge points in a knowledge network into a position distance relation in an approximate position social network, constructing a path set of the user for accessing the knowledge points, and realizing knowledge recommendation in the knowledge social network through an improved position social recommendation algorithm so as to improve the efficiency of the user in learning and understanding certain knowledge.
Drawings
FIG. 1 is a block diagram of a recommendation algorithm that is currently available;
FIG. 2 is a flowchart illustrating the steps of a knowledge recommendation method based on improved location-based social networking according to an embodiment of the present invention.
Detailed Description
As shown in fig. 1, the currently common recommendation algorithm is mainly content-based recommendation, collaborative filtering recommendation, hybrid recommendation, etc., and the basic algorithm framework thereof is shown in fig. 1; in the process of generating recommended articles, selecting a commodity recommended to a user mainly by calculating the similarity degree between the articles; for the knowledge-type recommended object, the main purposes of users for searching on a knowledge system or using the knowledge system are to learn the content which is not understood and to know the knowledge which is not clear, and for the content which is already known, the users do not need to recommend an approximate object to the system; in the conventional recommendation system, the recommendation purpose generally refers to that the user is provided with an object that is considered by the recommendation system to be similar to the object currently viewed by the user or an item that is considered by the recommendation system to be most likely to be interesting to the user, and the recommendation system model and the recommendation purpose are not suitable for the knowledge-based system.
Therefore, aiming at the defects existing in the conventional recommendation system model, the embodiment of the invention mainly aims at a knowledge-Based system or a knowledge social network, and provides an improved Location-Based social network LBSN (Location-Based social network) Interest recommendation algorithm facing the knowledge network to recommend the knowledge content with specific logical connection in the knowledge social network by combining a social network and a Point-of-Interest (POI) recommendation algorithm in a Location network. By the aid of the algorithm, the current situation that common commodity-oriented recommendation models and algorithms are not suitable for recommendation on a knowledge learning system can be solved, the knowledge acquisition efficiency of a user on the knowledge learning system is improved, and the learning efficiency of the user is improved.
As shown in fig. 2, a knowledge recommendation method based on improved location socialization includes:
s1, extracting data required by knowledge recommendation, wherein the data comprises evaluation data of a user on knowledge;
s2, converting implicit evaluation data in the extracted evaluation data into explicit evaluation data;
s3, calculating the similarity between the users according to a similarity algorithm;
s4, selecting a plurality of similar user data according to the similarity between the users;
s5, constructing a knowledge point distance matrix, fusing the similar user data on the knowledge point distance matrix, and constructing a similar user and knowledge point learning path set;
s6, combining a plurality of similar users and a knowledge point learning path set to form a candidate path set;
s7, respectively calculating the probability of each similar user accessing the corresponding knowledge point according to the candidate path set;
and S8, generating a corresponding knowledge point recommendation list according to the calculated probability.
As a preferred implementation of the embodiment of the knowledge recommendation method, the data required for knowledge recommendation in step S1 further includes user set data, knowledge set data, and interaction data among users.
As a preferred implementation of the embodiment of the knowledge recommendation method, the evaluation data in step S1 includes implicit evaluation data and explicit evaluation data, and after the implicit evaluation data is converted into the explicit evaluation data, a user learning score matrix is constructed according to the evaluation data obtained after the conversion and the interaction data between the users.
Further, as a preferred implementation of the embodiment of the knowledge recommendation method, according to the user learning score matrix, the average score made by the user, the knowledge point set corresponding to the same score, and the score of the user on the same knowledge point are respectively extracted.
And further, calculating the similarity between the users according to the extracted average scores made by the users, the knowledge point sets corresponding to the same scores and the scores of the users on the same knowledge point.
Further, the similarity calculation is performed by the following formula:
in the formula, RU，VRepresenting users U and usersV set of learning knowledge points of equal score, rU,iRepresents the score, r, of the user U on the learning knowledge point iV,iRepresents the score of the user V for the learning knowledge point i.
Further, the similarity calculation may be further performed by the following formula:
in the formula, RU，VSet of learning knowledge points, r, representing the same score for user U and user VU,iRepresents the score, r, of the user U on the learning knowledge point iV,iRepresents the score of the user V for the learning knowledge point i,
represents the average score of the learning knowledge points by the user U,
represents the average score of the user V for learning knowledge points.
Further, the probability of each of the similar users reaching the corresponding knowledge point for learning is calculated by the following formula:
where ρ is the probability of the user accessing the corresponding knowledge point,
learning the eigenvalue vectors of the path sets for the user and knowledge points,
the feature weight in the feature value vector.
In this embodiment, in step S1, the knowledge recommendation is performed on the knowledge system or the knowledge-social network, and it is first necessary to extract information content necessary for the knowledge recommendation. The information content comprises user set data, knowledge set data, interaction data among users and evaluation data of knowledge by the users. The user set data mainly refers to explicit information of registered users in the system, such as native place, cultural degree, interested fields and the like of the users; the knowledge set data mainly refers to explicit information contained in the knowledge, such as the name, the summary introduction, the contained resource content and other data of the knowledge; the inter-user interaction data mainly refers to social interaction behaviors among users in the network, such as behavior data of leaving messages, asking questions, sending messages, recommending, agreeing and the like; the evaluation data of knowledge by the user mainly comprises explicit evaluation data and implicit evaluation data. The information content required by knowledge recommendation is prepared for realizing a recommendation algorithm in a knowledge recommendation system, and the recommendation performance of the recommendation algorithm can be directly influenced by the richness of data extraction.
In this embodiment, in the knowledge social network, a set relationship including 2 entities is first constructed: a set of users and a set of knowledge points. The following is assumed in the algorithm:
① knowledge social network contains m users, setting set U ═ U1, U2, …, um };
② the network contains n knowledge points, and the set K is K { K1, K2, …, kn }.
In the knowledge social network, besides the two entity relationship sets, interaction data between users and associated data between the users and knowledge points exist, the data are generally extracted, and a user and knowledge evaluation data matrix is constructed in a recommendation system. In this embodiment, when extracting the user and the knowledge evaluation data, on the basis of extracting the knowledge evaluation data, the knowledge learning context data of the user needs to be extracted to ensure that the recommendation performance of the recommendation algorithm is not affected as much as possible. And setting a set of knowledge points k which are clicked and learned by a user u to be represented as Ku, giving a unique identification code to each knowledge point in the knowledge social network, and extracting the learning behavior record of the user to form a learning scoring matrix F belonging to RM multiplied by N of the user and the knowledge points, wherein each element fij in the learning condition matrix F represents the learning condition score of the user i on the knowledge point j.
However, due to the extracted user pair knowledgeThe evaluation data mainly includes explicit evaluation data and implicit evaluation data, and the implicit evaluation data needs to be preprocessed and converted into the displayed evaluation data, that is, regarding step S2, since most users in the knowledge social network do not want to spend time evaluating learned knowledge points, it is often known that sufficient explicit evaluation data cannot be collected in the knowledge social network; one difficulty in recommending in knowledge social networks is that the problem of sparse evaluation data needs to be solved. Although scoring data is lacking in knowledge social networks, user behavior data is easy to collect and the user preference information implied by the data is richer. Before the user learning scoring matrix is constructed, data such as access times, access duration or access behavior modes of a user to knowledge points can be extracted and converted into explicit scoring data; for example, the number of times that the user clicks reflects the implicit information of the difficulty level that the user understands the knowledge point; the higher the user clicks or the longer the dwell time, the higher the difficulty of the knowledge point for the user is; and preferentially selecting the content related to the knowledge for recommendation on the recommendation result. The knowledge social network has various implicit behaviors, and for implicit data in learning behaviors, implicit evaluation data is preprocessed and converted into explicit scoring data. For example, in the present embodiment, the scoring range is set to 1 to 5, and a scoring matrix of the user Ui for the knowledge point Kj can be obtained, where Ui represents a set of all users, that is, includes the user U1、U2、U3… … Um; kj represents the set of all knowledge points, i.e. including knowledge point K1、K2、K3… … Kn. In this embodiment, implicit evaluation data in knowledge evaluation data of a user is first converted into explicit evaluation data, which specifically refers to table 1:
TABLE 1 implicit Scoring conversion
| User' s | Click on | Asking questions | Stay | Recovery | Submitting a job | … | Conversion score |
| U1 | 2 | 0 | 15 minutes | 2 | 1 | … | 4.3 |
| U2 | 1 | 1 | 2 minutes | 0 | 0 | … | 1.5 |
| … | … | … | … | … | … | … | … |
| Um-1 | 4 | 2 | 30 minutes | 0 | 0 | … | 5 |
| um | 2 | 0 | 5 minutes | 2 | 1 | … | 1 |
If the score obtained by a knowledge point on a user is 5, the user is shown to be interested in the knowledge point or the difficulty of the knowledge point is considered to be higher. If the score is 1, it indicates that the user is not interested in the knowledge point or that the knowledge point is considered to be very simple.
Forming a user learning scoring matrix according to the display evaluation data obtained by conversion, the display evaluation data in the evaluation data of knowledge of the user and the interaction data among the users, which are collected at the beginning, and specifically referring to the table 2:
TABLE 2 user learning scoring matrix
| K1 | K2 | … | Kn-1 | kn | |
| U1 | 3 | 4 | … | R(1,n-1) | R(1,n) |
| U2 | 5 | 1 | … | R(2,n-1) | R(2,n) |
| … | … | … | … | … | … |
| Um-1 | R(m-1,1) | R(m-1,2) | … | R(m-1,n-1) | R(m-1,n) |
| um | R(m,1) | R(m,2) | … | R(m,n-1) | R(m,n-1) |
Under the initial condition of the system, the user accesses the knowledge points randomly, and a large amount of knowledge point evaluation data is possibly lost, namely the system is in cold start; there are many solutions for the system cold start situation, and a random function may be used to generate the score or an average value of scores of adjacent users may be selected according to the similarity of the users to fill in the data.
In this embodiment, the average score of each user on the knowledge points, the learning knowledge point set corresponding to the same score, and the score of each user on the same learning knowledge point can be extracted through table 2, that is, through the user learning score matrix. Then, using these extracted data, the similarity between users is calculated by a similarity calculation method.
Regarding step S3, in the learning network, the learning requirements of similar users for knowledge points are also similar, and a common calculation user similarity calculation method is cosine similarity calculation, which first uses cosine similarity to perform user similarity calculation. R for defining learning knowledge point set with same score of user U and user VU，VIs represented byU,iRepresents the score, r, of the user U on the learning knowledge point iV,iRepresents the score of the user V for the learning knowledge point i,
represents the average score of the learning knowledge points by the user U,
represents the average score of the user V for learning knowledge points. The similarity between users is measured by cosine included angle between vectors, the score of the user to the project is regarded as the vector of n-dimensional project space, and the similarity Sim (U, V) between the user U and the user V can be calculated according to a cosine similarity formula or a pearson correlation coefficient calculation formula, wherein the formula is as follows:
regarding step S4, the similarity between users can be calculated through similarity calculation, and then the most similar n user data are selected according to the similarity ranking.
Regarding step S5, in the recommendation algorithm of the social knowledge network lbs n, anything is considered relevant, but things that are close are more relevant than things that are far. In the knowledge learning network, the knowledge point B which the user visits first after learning the current knowledge point a is often the content which the user considers to be most closely logically connected with the knowledge point a, so that in the learning network, a similar concept of the distance between the knowledge points of interest in the lbs n system also exists. For the distance of the knowledge points in the knowledge network, the distance from one knowledge point A to another knowledge point B can be marked as 1, the distance from B to knowledge point C is marked as 2 when B jumps to the knowledge point C, the distance values of the nodes in the knowledge point network are sequentially acquired, and a knowledge point distance matrix is constructed. Considering the content association tightness after multiple jumps, and combining the theory of influence of three degrees, the distance between different knowledge points is marked as 3 at most, that is, if the distance from one knowledge point to another knowledge point is more than 3 jumps, the distance between the two knowledge points is considered to be very far, and no strong logic connection exists. And fusing user information on the knowledge point distance matrix, and extracting all knowledge point jumps of the user i for less than 3 times on the knowledge point set K, namely forming a path set between the user and the knowledge points, namely a learning path set between the user and the knowledge points. For example, Pi represents a path set in which a knowledge point learning path of a user i in the learning network jumps less than 3 times, and a path set M of knowledge points and users of M users in the learning network at n knowledge points is { P1, P2, … Pm-1, Pm }.
Regarding step S6, because of the problem of sparsity of the user' S scores on the knowledge points in the learning network, in order to construct a more comprehensive and valuable recommended knowledge point, n sets of paths of the user and the knowledge points extracted through similarity calculation should be constructed respectively according to the method for constructing the path sets of the user and the knowledge points, that is, n sets of paths (path sets of the user and the knowledge points) should be constructed, and then the n sets of paths are merged to form a candidate path set. Assuming that the candidate path set formed by merging is M', M ═ { M1, M2, … Mn-1, Mn }.
Regarding step S7, when calculating the probability ρ of each of the similar users accessing the corresponding knowledge point, it is necessary to obtain
And
the value of (a), wherein,
learning the eigenvalue vectors of the path sets for the user and knowledge points,
the feature weight in the feature value vector is obtained; feature weight
The characteristic value vector can be obtained through training data
Can be obtained by the following method:
according to the candidate path set M' learned by the user and the knowledge point determined in step S6, the feature value vector can be calculated from the feature values of any user i and the knowledge point l through the feature values of the path set of the user and the knowledge point:
then according to the obtained acquisition
And
the value of (c) is known through a logistic regression formula, and the probability that the user i needs to go to the knowledge point l for learning is as follows:
according to the method, the probability of each user accessing the corresponding knowledge point is calculated respectively.
And step S8, sorting according to the calculated probability rho, selecting a proper number of knowledge points from large to small according to the probability to form a recommendation list, and recommending the recommendation list to a corresponding user.
In summary, the knowledge recommendation method based on improved location social contact described in this embodiment has the following advantages:
the knowledge recommendation method provided by the embodiment of the invention makes up the defect that the conventional common recommendation system model and algorithm mainly recommend commercial articles, and is mainly oriented to a knowledge system to generate recommendation contents, so that a user can understand related knowledge contents which can be further learned through system recommendation under the condition that a certain knowledge content is not known; the method converts behavior data of the user on knowledge into explicit scores, measures the understanding degree of the user on current knowledge data through the scores, virtualizes the distance of knowledge points in the knowledge network into a position distance relation in an approximate position social network, constructs a path set of the user for accessing the knowledge points, realizes knowledge recommendation in the knowledge social network through an improved position social recommendation algorithm, and can improve the efficiency of learning and understanding a certain knowledge by the user.
The embodiment also includes a knowledge recommendation system based on improving location social contact, including:
the extraction module is used for extracting data required by knowledge recommendation, and the data user evaluates the knowledge;
the conversion module is used for converting implicit evaluation data in the extracted evaluation data into explicit evaluation data;
the similarity calculation module is used for calculating the similarity between the users according to a similarity calculation method;
the selecting module is used for selecting a plurality of similar user data according to the similarity between the users;
the construction module is used for constructing a knowledge point distance matrix, fusing the similar user data on the knowledge point distance matrix and constructing a similar user and knowledge point learning path set;
the merging module is used for merging the similar users and the knowledge point learning path set to form a candidate path set;
the probability calculation module is used for respectively calculating the probability of each similar user accessing the corresponding knowledge point according to the candidate path set;
and the generating module is used for generating a corresponding knowledge point recommendation list according to the calculated probability.
The present embodiments also include a storage medium having stored therein processor-executable instructions that, when executed by a processor, are configured to perform the knowledge recommendation method.
When the knowledge recommendation system based on social contact based on improved locations in the embodiment is executed by using a computer or other terminals to run corresponding programs, the medium refers to a storage module in the computer or other terminals. When the functions of the method and the medium are realized, the same technical effects as the knowledge recommendation system based on the improved location social contact can be realized.
The above description is only a preferred embodiment of the present invention, and the present invention is not limited to the above embodiment, and any modifications, equivalent substitutions, improvements, etc. within the spirit and principle of the present invention should be included in the protection scope of the present invention as long as the technical effects of the present invention are achieved by the same means. The invention is capable of other modifications and variations in its technical solution and/or its implementation, within the scope of protection of the invention.
Claims (10)
1. A knowledge recommendation method based on improved location socialization is characterized by comprising the following steps:
extracting data required by knowledge recommendation, wherein the data comprises evaluation data of knowledge by a user;
converting implicit evaluation data in the extracted evaluation data into explicit evaluation data;
calculating the similarity between users according to a similarity algorithm;
selecting a plurality of similar user data according to the similarity between the users;
constructing a knowledge point distance matrix, fusing the similar user data on the knowledge point distance matrix, and constructing a similar user and knowledge point learning path set;
combining a plurality of similar users with a knowledge point learning path set to form a candidate path set;
respectively calculating the probability of each similar user accessing the corresponding knowledge point according to the candidate path set;
and generating a corresponding knowledge point recommendation list according to the calculated probability.
2. The knowledge recommendation method based on improved location socialization according to claim 1, characterized in that the data required for knowledge recommendation further includes user set data, knowledge set data and interaction data among users.
3. The knowledge recommendation method based on improved location social contact as claimed in claim 2, wherein the evaluation data comprises implicit evaluation data and explicit evaluation data, and after the implicit evaluation data is converted into the explicit evaluation data, a user learning scoring matrix is constructed according to the evaluation data obtained after conversion and interaction data between the users.
4. The knowledge recommendation method based on improved location socialization according to claim 3, characterized by further comprising the following steps:
and respectively extracting the average scores made by the users, knowledge point sets corresponding to the same scores and the scores of the users on the same knowledge points according to the user learning score matrix.
5. The knowledge recommendation method based on improved location socialization according to claim 4, characterized by further comprising the following steps:
and calculating the similarity between the users according to the extracted average scores made by the users, the knowledge point sets corresponding to the same scores and the scores of the users on the same knowledge point.
6. The method of claim 5, wherein the similarity calculation is performed by the following formula:
in the formula, RU，VSet of learning knowledge points, r, representing the same score for user U and user VU,iRepresents the score, r, of the user U on the learning knowledge point iV,iRepresents the score of the user V for the learning knowledge point i.
7. The method of claim 6, wherein the similarity calculation is further performed by the following formula:
in the formula, RU，VSet of learning knowledge points, r, representing the same score for user U and user VU,iRepresents the score, r, of the user U on the learning knowledge point iV,iRepresents the score of the user V for the learning knowledge point i,
represents the average score of the learning knowledge points by the user U,
represents the average score of the user V for learning knowledge points.
8. The method of claim 1, wherein the probability of each of the similar users accessing the corresponding knowledge point is calculated by the following formula:
where ρ is the probability of the user accessing the corresponding knowledge point,
learning the eigenvalue vectors of the path sets for the user and knowledge points,
the feature weight in the feature value vector.
9. A knowledge recommendation system based on improved location socialization, comprising:
the extraction module is used for extracting data required by knowledge recommendation, and the data user evaluates the knowledge;
the conversion module is used for converting implicit evaluation data in the extracted evaluation data into explicit evaluation data;
the similarity calculation module is used for calculating the similarity between the users according to a similarity calculation method;
the selecting module is used for selecting a plurality of similar user data according to the similarity between the users;
the construction module is used for constructing a knowledge point distance matrix, fusing the similar user data on the knowledge point distance matrix and constructing a similar user and knowledge point learning path set;
the merging module is used for merging the similar users and the knowledge point learning path set to form a candidate path set;
the probability calculation module is used for respectively calculating the probability of each similar user accessing the corresponding knowledge point according to the candidate path set;
and the generating module is used for generating a corresponding knowledge point recommendation list according to the calculated probability.
10. A storage medium having stored therein processor-executable instructions, which when executed by a processor, are configured to perform the method of any one of claims 1-8.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201911134890.4A CN110928920B (en) | 2019-11-19 | 2019-11-19 | Knowledge recommendation method, system and storage medium based on improved position social contact |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201911134890.4A CN110928920B (en) | 2019-11-19 | 2019-11-19 | Knowledge recommendation method, system and storage medium based on improved position social contact |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN110928920A true CN110928920A (en) | 2020-03-27 |
| CN110928920B CN110928920B (en) | 2022-08-09 |
Family
ID=69850296
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201911134890.4A Active CN110928920B (en) | 2019-11-19 | 2019-11-19 | Knowledge recommendation method, system and storage medium based on improved position social contact |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN110928920B (en) |
Cited By (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN113505306A (en) * | 2021-06-21 | 2021-10-15 | 广东交通职业技术学院 | Interest point recommendation method, system and medium based on heterogeneous graph neural network |
Citations (5)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN102737120A (en) * | 2012-06-01 | 2012-10-17 | 西安交通大学 | Personalized network learning resource recommendation method |
| US20160224564A1 (en) * | 2013-09-29 | 2016-08-04 | Peking University Founder Group Co., Ltd. | Method and system for key knowledge point recommendation |
| CN107391503A (en) * | 2016-05-16 | 2017-11-24 | 刘洪波 | Personalized recommendation method with interest guiding function |
| CN109241405A (en) * | 2018-08-13 | 2019-01-18 | 华中师范大学 | A kind of associated education resource collaborative filtering recommending method of knowledge based and system |
| CN110288878A (en) * | 2019-07-01 | 2019-09-27 | 科大讯飞股份有限公司 | Adaptive learning method and device |
-
2019
- 2019-11-19 CN CN201911134890.4A patent/CN110928920B/en active Active
Patent Citations (5)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN102737120A (en) * | 2012-06-01 | 2012-10-17 | 西安交通大学 | Personalized network learning resource recommendation method |
| US20160224564A1 (en) * | 2013-09-29 | 2016-08-04 | Peking University Founder Group Co., Ltd. | Method and system for key knowledge point recommendation |
| CN107391503A (en) * | 2016-05-16 | 2017-11-24 | 刘洪波 | Personalized recommendation method with interest guiding function |
| CN109241405A (en) * | 2018-08-13 | 2019-01-18 | 华中师范大学 | A kind of associated education resource collaborative filtering recommending method of knowledge based and system |
| CN110288878A (en) * | 2019-07-01 | 2019-09-27 | 科大讯飞股份有限公司 | Adaptive learning method and device |
Cited By (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN113505306A (en) * | 2021-06-21 | 2021-10-15 | 广东交通职业技术学院 | Interest point recommendation method, system and medium based on heterogeneous graph neural network |
| CN113505306B (en) * | 2021-06-21 | 2022-04-22 | 广东交通职业技术学院 | Interest point recommendation method, system and medium based on heterogeneous graph neural network |
Also Published As
| Publication number | Publication date |
|---|---|
| CN110928920B (en) | 2022-08-09 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| Ren et al. | Context-aware probabilistic matrix factorization modeling for point-of-interest recommendation | |
| Guo et al. | Combining geographical and social influences with deep learning for personalized point-of-interest recommendation | |
| Mao et al. | Multiobjective e-commerce recommendations based on hypergraph ranking | |
| Jiang et al. | Social recommendation with cross-domain transferable knowledge | |
| Huang et al. | Social friend recommendation based on multiple network correlation | |
| CN111444395B (en) | Method, system and equipment for obtaining relation expression between entities and advertisement recall system | |
| Xing et al. | Points-of-interest recommendation based on convolution matrix factorization | |
| Zhao et al. | Service quality evaluation by exploring social users’ contextual information | |
| Zhu et al. | Neural attentive travel package recommendation via exploiting long-term and short-term behaviors | |
| CN112989169B (en) | Target object identification method, information recommendation method, device, equipment and medium | |
| Yu et al. | Graph neural network based model for multi-behavior session-based recommendation | |
| CN111310046A (en) | Object recommendation method and device | |
| Wang et al. | Modelling local and global dependencies for next-item recommendations | |
| Liu et al. | POI Recommendation Method Using Deep Learning in Location‐Based Social Networks | |
| CN115408618B (en) | Point-of-interest recommendation method based on social relation fusion position dynamic popularity and geographic features | |
| Xiong et al. | Affective impression: Sentiment-awareness POI suggestion via embedding in heterogeneous LBSNs | |
| Chen et al. | Learning user preference from heterogeneous information for store-type recommendation | |
| Zhao et al. | A generative model approach for geo-social group recommendation | |
| Yu et al. | News recommendation model based on encoder graph neural network and bat optimization in online social multimedia art education | |
| CN110928920B (en) | Knowledge recommendation method, system and storage medium based on improved position social contact | |
| Shilin | User Model‐Based Personalized Recommendation Algorithm for News Media Education Resources | |
| Fu et al. | Graph contextualized self-attention network for software service sequential recommendation | |
| Xu et al. | A novel travel group recommendation model based on user trust and social influence | |
| Guo et al. | Learning to Make Document Context‐Aware Recommendation with Joint Convolutional Matrix Factorization | |
| Guo et al. | Network Embedding‐Aware Point‐of‐Interest Recommendation in Location‐Based Social Networks |
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 | ||
| GR01 | Patent grant | ||
| GR01 | Patent grant |