CN117033807A - Recommendation analysis method and system based on big data and social network - Google Patents
Recommendation analysis method and system based on big data and social network Download PDFInfo
- Publication number
- CN117033807A CN117033807A CN202311057493.8A CN202311057493A CN117033807A CN 117033807 A CN117033807 A CN 117033807A CN 202311057493 A CN202311057493 A CN 202311057493A CN 117033807 A CN117033807 A CN 117033807A
- Authority
- CN
- China
- Prior art keywords
- preference
- target user
- target
- users
- user
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Pending
Links
- 238000004458 analytical method Methods 0.000 title claims abstract description 33
- 238000001914 filtration Methods 0.000 claims abstract description 45
- 238000004422 calculation algorithm Methods 0.000 claims abstract description 40
- 239000013598 vector Substances 0.000 claims description 151
- 239000011159 matrix material Substances 0.000 claims description 42
- 238000004088 simulation Methods 0.000 claims description 23
- 230000006399 behavior Effects 0.000 abstract description 72
- 238000000034 method Methods 0.000 abstract description 17
- 241001494479 Pecora Species 0.000 abstract description 11
- 230000009471 action Effects 0.000 abstract description 4
- 230000000694 effects Effects 0.000 abstract description 4
- 230000003542 behavioural effect Effects 0.000 description 5
- 238000004364 calculation method Methods 0.000 description 5
- 230000008569 process Effects 0.000 description 4
- 102100029469 WD repeat and HMG-box DNA-binding protein 1 Human genes 0.000 description 1
- 101710097421 WD repeat and HMG-box DNA-binding protein 1 Proteins 0.000 description 1
- 238000013459 approach Methods 0.000 description 1
- 230000009286 beneficial effect Effects 0.000 description 1
- 238000011161 development Methods 0.000 description 1
- 238000010586 diagram Methods 0.000 description 1
- 230000003993 interaction Effects 0.000 description 1
- 238000012986 modification Methods 0.000 description 1
- 230000004048 modification Effects 0.000 description 1
- 238000005457 optimization Methods 0.000 description 1
- 230000000750 progressive effect Effects 0.000 description 1
- 238000012216 screening Methods 0.000 description 1
- 239000002699 waste material Substances 0.000 description 1
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
- G06F16/9536—Search customisation based on social or collaborative filtering
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06Q—INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES; SYSTEMS OR METHODS SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES, NOT OTHERWISE PROVIDED FOR
- G06Q50/00—Systems or methods specially adapted for specific business sectors, e.g. utilities or tourism
- G06Q50/01—Social networking
-
- Y—GENERAL TAGGING OF NEW TECHNOLOGICAL DEVELOPMENTS; GENERAL TAGGING OF CROSS-SECTIONAL TECHNOLOGIES SPANNING OVER SEVERAL SECTIONS OF THE IPC; TECHNICAL SUBJECTS COVERED BY FORMER USPC CROSS-REFERENCE ART COLLECTIONS [XRACs] AND DIGESTS
- Y02—TECHNOLOGIES OR APPLICATIONS FOR MITIGATION OR ADAPTATION AGAINST CLIMATE CHANGE
- Y02D—CLIMATE CHANGE MITIGATION TECHNOLOGIES IN INFORMATION AND COMMUNICATION TECHNOLOGIES [ICT], I.E. INFORMATION AND COMMUNICATION TECHNOLOGIES AIMING AT THE REDUCTION OF THEIR OWN ENERGY USE
- Y02D10/00—Energy efficient computing, e.g. low power processors, power management or thermal management
Abstract
Compared with the prior art, the recommendation analysis method and system based on big data and the social network are improved on the basis of the existing collaborative filtering algorithm, and whether the user can find the user with higher similarity is judged through the action sparseness. If the behavior sparsity is high, the behavior of the representative user is high, and similar users can be found at the moment, and then the recommendation is performed by using the existing collaborative filtering algorithm. If the sparseness of the behaviors is low, the behaviors of the representative users are less, and similar users are difficult to find, then the social association users with social relations with the target users are acquired, and recommendation is performed through actual social association between the users, so that a more accurate recommendation effect is achieved. The method solves the problem of gray sheep in the existing collaborative filtering algorithm, is expected to provide more accurate and personalized recommendation service, and improves the experience of social network users.
Description
Technical Field
The invention relates to the technical field of big data information recommendation, in particular to a recommendation analysis method and a recommendation analysis system based on big data and a social network.
Background
Today, intelligent recommendations are increasingly important in the relevant software. In today's networks, the amount of information is so large that it is difficult for users to find content from them that meets personal interests. Thus, intelligent recommendation algorithms become indispensable in the network. The method can be used for screening and recommending the content related to the user interests by analyzing the interests and behaviors of the user, so that the user experience and efficiency are improved.
Collaborative filtering algorithms are a common intelligent recommendation algorithm. It recommends content that may be of interest to the target user by analyzing similarities between users, using the behavior and preferences of other similar users. User-based collaborative filtering and item-based collaborative filtering are two common collaborative filtering methods.
Although collaborative filtering algorithms perform well in many cases, when the similarity between users is low, the problem of gray sheep, which means that when the similarity between the target user and other users is low, collaborative filtering is difficult to give accurate recommendations. In this case, the recommendation result may be less accurate and may not meet the personalized needs of the user. Therefore, an intelligent recommendation algorithm capable of solving the problem of gray sheep is needed.
Disclosure of Invention
In view of the foregoing, it is necessary to provide a recommendation analysis method and system based on big data and social network, so as to solve the problem of gray sheep in the collaborative filtering algorithm in the prior art.
The invention provides a recommendation analysis method based on big data and a social network, which comprises the following steps:
acquiring a preset preference matrix, wherein the preset preference matrix comprises preference vectors of a plurality of preset users, each element position in the preference vectors represents an object to be recommended, and each element value represents the preference degree of the user to the object to be recommended;
acquiring a preference vector of a target user, and acquiring the behavior sparsity of the target user according to the sparsity of the preference vector of the target user;
if the behavior sparsity of the target user is larger than a preset threshold, obtaining a target recommended object of the target user based on a collaborative filtering algorithm according to the preference vector of the target user and the preset preference matrix;
if the behavior sparsity of the target user is smaller than a preset threshold, acquiring a plurality of social related users of the target user and preference vectors of the social related users, and acquiring a target recommended object of the target user according to the preference vectors of the social related users.
Further, the element values of the preference vector include 1, -1 and 0, wherein 1 indicates that the user has made a behavior indicating that the user prefers the object to be recommended, -1 indicates that the user has made a behavior indicating that the user does not prefer the object to be recommended, and 0 indicates that the user has not made a behavior indicating whether the user prefers the object to be recommended.
Further, the obtaining the behavior sparsity of the target user according to the sparsity of the preference vector of the target user includes:
and counting the number of non-zero elements in the preference vector of the target user, and taking the number of the non-zero elements as the behavior sparsity of the target user.
Further, the obtaining, based on a collaborative filtering algorithm, the target recommended object of the target user according to the preference vector of the target user and the preset preference matrix includes:
calculating the similarity of the target user and the preset user based on the similarity of the preference vector of the target user and the preference vector in the preset preference matrix;
obtaining similar users of the target user from the preset users according to the similarity;
and obtaining the target recommended object of the target user according to the recommended object of the similar user.
Further, the obtaining the target recommended object of the target user according to the preference vectors of the social related users includes:
obtaining a social connection coefficient of each social connection user according to the social connection degrees of the target user and the social connection users;
taking the social association coefficient as the weight of each element in the preference vector, combining preference vectors of a plurality of social association users to obtain a preference prediction vector of the target user;
and obtaining a target recommended object of the target user according to the preference prediction vector of the target user.
Further, the preference prediction vector of the target user is obtained by the following formula:
wherein (1)>Preference prediction vector representing the target user, < >> Respectively indicate->A preference vector for each of the socially-associated users,respectively represent the target user and +>Social association coefficients of individual social associated users.
Further, the obtaining the target recommended object of the target user according to the preference prediction vector of the target user includes:
normalizing the preference prediction vector of the target user to obtain a preference simulation vector of the target user, wherein the element values of the preference simulation vector comprise 1, -1 and 0, wherein 1 represents that the target user is simulated to make a behavior representing a preference of an object to be recommended, -1 represents that the target user is simulated to make a behavior representing a non-preference of the object to be recommended, and 0 represents that the target user is simulated to not make a behavior representing whether the object to be recommended is favored;
and obtaining a target recommended object of the target user based on a collaborative filtering algorithm according to the preference simulation vector of the target user and the preset preference matrix.
Further, normalizing the preference prediction vector of the target user to obtain a preference simulation vector of the target user, including:
acquiring a target element value, wherein the target element value is an element value to be normalized in a preference prediction vector of the target user;
acquiring a preset upper limit threshold and a preset lower limit threshold, wherein the preset upper limit threshold and the preset lower limit threshold are opposite numbers;
if the target element value is larger than the preset upper threshold, the element value corresponding to the target element value in the preference simulation vector is 1;
if the target element value is smaller than the preset lower limit threshold value, the element value corresponding to the target element value in the preference simulation vector is-1;
and if the target element value is between the preset upper limit threshold value and the preset lower limit threshold value, the element value corresponding to the target element value in the preference simulation vector is 0.
The invention also provides a recommendation analysis system based on big data and a social network, which comprises:
the big data acquisition module is used for acquiring a preset preference matrix, wherein the preset preference matrix comprises preference vectors of a plurality of preset users, each element position in the preference vectors represents an object to be recommended, and each element value represents the preference degree of the user to the object to be recommended;
the user behavior analysis module is used for acquiring a preference vector of a target user and obtaining the behavior sparsity of the target user according to the sparsity of the preference vector of the target user;
the first recommendation module is used for obtaining a target recommendation object of the target user based on a collaborative filtering algorithm according to the preference vector of the target user and the preset preference matrix when the behavior sparsity of the target user is larger than a preset threshold;
the second recommendation module is used for acquiring a plurality of social related users of the target user and preference vectors of the social related users when the behavior sparseness of the target user is smaller than a preset threshold value, and acquiring target recommended objects of the target user according to the preference vectors of the social related users.
The beneficial effects of the invention are as follows:
the invention provides a recommendation analysis method and a recommendation analysis system based on big data and a social network, wherein a preset preference matrix is firstly obtained, the preset preference matrix comprises preference vectors of a plurality of preset users, each element position in the preference vectors represents an object to be recommended, each element value represents preference degree of the user to be recommended, then the preference vectors of the target users are obtained, according to sparsity of the preference vectors of the target users, behavior sparsity of the target users is obtained, if the behavior sparsity of the target users is larger than a preset threshold, target recommended objects of the target users are obtained based on a collaborative filtering algorithm according to the preference vectors of the target users and the preset preference matrix, if the behavior sparsity of the target users is smaller than the preset threshold, a plurality of social associated users of the target users and the preference vectors of the social associated users are obtained, and according to the preference vectors of the social associated users, the target recommended objects of the target users are obtained. Compared with the prior art, the invention improves the existing collaborative filtering algorithm, and judges whether the user can find the user with higher similarity or not through the action sparseness. If the behavior sparsity is high, the behavior of the representative user is high, and similar users can be found at the moment, and then the recommendation is performed by using the existing collaborative filtering algorithm. If the sparseness of the behaviors is low, the behaviors of the representative users are less, and similar users are difficult to find, then the social association users with social relations with the target users are acquired, and recommendation is performed through actual social association between the users, so that a more accurate recommendation effect is achieved. The method solves the problem of gray sheep in the existing collaborative filtering algorithm, is expected to provide more accurate and personalized recommendation service, and improves the experience of social network users.
Drawings
FIG. 1 is a flowchart of a method for an embodiment of a recommendation analysis method based on big data and social networks provided by the present invention;
FIG. 2 is a flowchart of a method according to an embodiment of step S104 in FIG. 1;
fig. 3 is a system architecture diagram of an embodiment of a recommendation analysis system based on big data and social networks provided by the present invention.
Detailed Description
The following description of the embodiments of the present invention will be made clearly and completely with reference to the accompanying drawings, in which it is apparent that the embodiments described are only some embodiments of the present invention, but not all embodiments. All other embodiments, which can be made by those skilled in the art based on the embodiments of the invention without making any inventive effort, are intended to be within the scope of the invention.
Before describing particular embodiments, the related terms referred to herein will be explained first:
collaborative filtering algorithm: this is one of the most common recommended analysis methods. It finds other users that are similar to the target user's interests based on user behavior data, such as the user's purchase records, scores, browsing history, etc., and then makes recommendations using these users ' preferences. Collaborative filtering can be categorized into user-based and item-based approaches.
Gray sheep problem: the gray sheep problem refers to a phenomenon in a recommendation system, namely when the similarity between users is low, the traditional collaborative filtering algorithm is difficult to accurately recommend the content of interest to the users, so that the recommendation result is inaccurate or does not meet the personalized requirements of the users.
In collaborative filtering algorithms, recommendations are made based on the similarity between users' behavioral data and the users. When the similarity between users is high, neighbor users with similar interests and preferences can be well found and related content is recommended to the target user based on their behavior. However, when the similarity between users is low, it is often difficult to find neighbor users that are similar to the interests of the target user, so that the recommendation results are not accurate or personalized.
The problem of sheep is generally caused by the following conditions:
1. data sparsity: when the user behavior data is relatively limited, it is difficult to find a neighbor user that has sufficiently similar behavior to the target user.
2. Cold start problem: for newly joined users or newly online items, there is a lack of sufficient behavioral data, making it difficult to find neighboring users or items similar thereto.
3. Small amount of behavioral data: when the user behavior data are small, the similarity between users cannot be accurately calculated, and the accuracy of the recommendation result is affected.
Referring to fig. 1, in one embodiment of the present invention, a recommendation analysis method based on big data and a social network is disclosed, the method comprising:
s101, acquiring a preset preference matrix, wherein the preset preference matrix comprises preference vectors of a plurality of preset users, each element position in the preference vectors represents an object to be recommended, and each element value represents the preference degree of the users to the object to be recommended;
s102, obtaining a preference vector of a target user, and obtaining the behavior sparsity of the target user according to the sparsity of the preference vector of the target user;
s103, if the behavior sparsity of the target user is larger than a preset threshold, obtaining a target recommended object of the target user based on a collaborative filtering algorithm according to the preference vector of the target user and the preset preference matrix;
s104, if the behavior sparsity of the target user is smaller than a preset threshold, acquiring a plurality of social related users of the target user and preference vectors of the social related users, and acquiring a target recommended object of the target user according to the preference vectors of the social related users.
Compared with the prior art, the invention improves the existing collaborative filtering algorithm, and judges whether the user can find the user with higher similarity or not through the action sparseness. If the behavior sparsity is high, the behavior of the representative user is high, and similar users can be found at the moment, and then the recommendation is performed by using the existing collaborative filtering algorithm. If the sparseness of the behaviors is low, the behaviors of the representative users are less, and similar users are difficult to find, then the social association users with social relations with the target users are acquired, and recommendation is performed through actual social association between the users, so that a more accurate recommendation effect is achieved. The method solves the problem of gray sheep in the existing collaborative filtering algorithm, is expected to provide more accurate and personalized recommendation service, and improves the experience of social network users.
It should be noted that, the social related user refers to a user in social relationship with the target user, for example, a network friend of the target user, a corresponding user in an address book uploaded by the user, a user in a focus list or a fan list of the target user, and the like. The social association users, such as users corresponding to friend relationships of students, bloggers or fan-shaped users, represent the same preferences as the target users to a certain extent, such as popular clothing styles among friends of the students, products recommended to the fan-shaped users by the bloggers, and the like, so that the social association users in the social network can compensate the gray sheep problem in the existing collaborative filtering algorithm to a certain extent, and the users can still recommend reasonable objects for the users when the referent behavior data are less or similar users are difficult to match from the existing database.
In a preferred embodiment, the step S101 of obtaining the preset preference matrix may be obtained through existing big data, for example, collecting and sorting behavior data of the user, such as purchase records, scores, browsing history, and the like. These data are used to construct a user-item matrix (i.e. preference matrix, in this embodiment, the item refers to goods, which is one of the objects to be recommended, and according to practical situations, the object to be recommended may be video, picture, advertisement, etc. other information), where row coordinates of the user-item matrix represent a user, column coordinates represent an item, and each cell represents a user's behavior on the item, i.e. in the user-item matrix, each row is a preference vector corresponding to a user.
The system for implementing the method can analyze the big data to obtain the users with typical behavior characteristics as preset users, and establishes a preset preference matrix according to the behavior data of the users to be stored in the server for continuous optimization. The target user and the social associated user of the target user may or may not exist among a plurality of preset users. Details regarding how to obtain data and build a preference matrix are well known in the art and are not described in detail herein.
Further, in a preferred embodiment, the element values of the preference vector include 1, -1 and 0, wherein 1 indicates that the user has made a behavior (e.g., praise, favorites, orders, etc.) indicating that the object to be recommended is favored, -1 indicates that the user has made a behavior (e.g., click-through, mask, bad comment, etc.) indicating that the object to be recommended is not favored, and 0 indicates that the user has not made a behavior (e.g., only a short browsing time) indicating that the object to be recommended is favored.
It will be appreciated that in practice, the preference of the user for the object to be recommended may also be represented by a specific numerical value. The mode in this embodiment makes the vector have only two numbers of 1 and 0, which is more convenient for the computer to operate.
Further, in a preferred embodiment, step S102 is performed to obtain a preference vector of the target user, and obtain the behavior sparsity of the target user according to the sparsity of the preference vector of the target user, specifically:
and counting the number of non-zero elements in the preference vector of the target user, and taking the number of the non-zero elements as the behavior sparsity of the target user.
It can be understood that in the vector encoding manner described above, the more the number of non-zero elements, the more the behavior of the representative user is, the more accurate the recommendation result based on collaborative filtering algorithm analysis can be, the fewer the number of non-zero elements, the less the behavior data that the representative user can refer to, and the lower the probability of obtaining the accurate recommendation result through collaborative filtering algorithm analysis, the more likely the user will be bothered when the similar object is forcibly recommended to the user based on a small amount of data, and the user experience will be affected. In practice, according to different coding modes of preference vectors, the sparsity of the user behavior can be represented by other features such as the modulus of the vectors.
On the other hand, in the embodiment, before specific collaborative filtering is performed, whether the subsequent collaborative filtering calculation is worth being performed is determined according to the behavior sparseness of the user, and when the behavior sparseness indicates that the behavior is less, the analysis can be directly performed through social correlation users, so that useless collaborative filtering calculation is avoided, the waste of calculation resources is caused, the corresponding speed is improved, and the user experience is improved.
Further, in a preferred embodiment, in the step S103, the obtaining, based on a collaborative filtering algorithm, the target recommended object of the target user according to the preference vector of the target user and the preset preference matrix specifically includes:
calculating the similarity of the target user and the preset user based on the similarity of the preference vector of the target user and the preference vector in the preset preference matrix;
obtaining similar users of the target user from the preset users according to the similarity;
and obtaining the target recommended object of the target user according to the recommended object of the similar user.
In the above process, the similarity calculation may be performed by calculating cosine similarity between two preference vectors, pearson correlation coefficient, and the like. The similarity calculation may compare the interests and preferences of the users based on their behavioral data.
In the selection of similar users, a group of users most similar to the interest of the target user may be selected as similar users according to the calculated similarity. A threshold is typically set and only similar users whose similarity exceeds the threshold are selected.
After selecting similar users, item recommendations may be made based on similar behavioral conditions. The target user may be recommended by recommending items that are liked by similar users and that have not been contacted by the target user. In addition, the weight can be given to the articles according to the scores of the similar users on the articles and the similarity with the target users, so that the recommendation accuracy is improved.
Other details in the above process are well known to those skilled in the art and will not be described further herein.
Further, as shown in fig. 2, in a preferred embodiment, in step S104, the obtaining, according to preference vectors of a plurality of social related users, the target recommended object of the target user specifically includes:
s201, obtaining a social connection coefficient of each social connection user according to social connection degrees of the target user and a plurality of social connection users;
s202, combining preference vectors of a plurality of social association users by taking the social association coefficients as weights of each element in the preference vectors to obtain a preference prediction vector of the target user;
s203, obtaining a target recommended object of the target user according to the preference prediction vector of the target user.
In step S201 in the above process, the social association coefficient may be obtained according to the interaction degree (such as chat frequency, number of chat records, number of praise comments, etc.) between two users, or the geographical location distance, etc., and the specific decision mode may be flexibly formulated according to the actual situation.
Specifically, in a preferred embodiment, in the step S202, the preference prediction vector of the target user is obtained by the following formula:
wherein (1)>Preference prediction vector representing the target user, < >> Respectively indicate->A preference vector for each of the socially-associated users,respectively represent the target user and +>Social association coefficients of individual social associated users.
The above formula first multiplies each preference vector by a corresponding weight coefficient (i.e., social-related coefficient), then adds all the products, and finally divides the sum by the sum of all the weight coefficients. This ensures that the combined preference prediction vector is the result of a weighted average of the individual preference vectors, and that the element values in the preference prediction vector resulting from the above procedure may not be the previously specified 1, 0 and-1, but still retain the characteristics of the original preference vector.
After obtaining the preference prediction vector, step S203 may be performed to obtain the target recommended object of the target user according to the preference prediction vector of the target user.
In a preferred embodiment, the result recommendation may be directly performed according to the preference prediction vector, so as to obtain a target recommended object, for example, the object to be recommended with the largest element value in the preference prediction vector is selected as the target recommended object.
Further, in another preferred embodiment, the step S203 further includes:
normalizing the preference prediction vector of the target user to obtain a preference simulation vector of the target user, wherein the element values of the preference simulation vector comprise 1, -1 and 0, wherein 1 represents that the target user is simulated to make a behavior representing a preference of an object to be recommended, -1 represents that the target user is simulated to make a behavior representing a non-preference of the object to be recommended, and 0 represents that the target user is simulated to not make a behavior representing whether the object to be recommended is favored;
and obtaining a target recommended object of the target user based on a collaborative filtering algorithm according to the preference simulation vector of the target user and the preset preference matrix.
The preference prediction vector is normalized in the process to obtain the preference simulation vector with the same form as that in the preset preference matrix, and the preference simulation vector is used for simulating the preference vector which can be generated by the user under the condition that the behavior data are sufficient, so that recommendation analysis can be performed on the basis of the existing collaborative filtering algorithm according to the preset preference matrix at the moment to obtain a more scientific recommendation result.
The mode mentioned in this embodiment may be implemented by directly making improvements in the existing intelligent recommendation system that adopts a collaborative filtering algorithm, that is, a module for increasing the sparseness of computing behaviors and a module for computing preference simulation vectors may be used, which shortens the development period of software products and can implement larger functional improvements.
Specifically, in a preferred embodiment, the normalizing the preference prediction vector of the target user to obtain the preference modeling vector of the target user specifically includes:
acquiring a target element value, wherein the target element value is an element value to be normalized in a preference prediction vector of the target user;
acquiring a preset upper limit threshold and a preset lower limit threshold, wherein the preset upper limit threshold and the preset lower limit threshold are opposite numbers;
if the target element value is larger than the preset upper threshold, the element value corresponding to the target element value in the preference simulation vector is 1;
if the target element value is smaller than the preset lower limit threshold value, the element value corresponding to the target element value in the preference simulation vector is-1;
and if the target element value is between the preset upper limit threshold value and the preset lower limit threshold value, the element value corresponding to the target element value in the preference simulation vector is 0.
Referring to fig. 3, the present invention further provides a recommendation analysis system 300 based on big data and social network, including:
the big data obtaining module 310 is configured to obtain a preset preference matrix, where the preset preference matrix includes preference vectors of a plurality of preset users, each element position in the preference vector represents an object to be recommended, and each element value represents a preference degree of the user for the object to be recommended;
the user behavior analysis module 320 is configured to obtain a preference vector of a target user, and obtain a behavior sparsity of the target user according to the sparsity of the preference vector of the target user;
the first recommendation module 330 is configured to obtain, based on a collaborative filtering algorithm, a target recommendation object of the target user according to the preference vector of the target user and the preset preference matrix when the behavior sparsity of the target user is greater than a preset threshold;
the second recommendation module 340 is configured to obtain a plurality of social related users of the target user and preference vectors of the social related users when the behavior sparseness of the target user is less than a preset threshold, and obtain a target recommended object of the target user according to the preference vectors of the social related users.
What needs to be explained here is: the corresponding system 300 provided in the foregoing embodiments may implement the technical solutions described in the foregoing method embodiments, and the specific implementation principles of the foregoing modules or units may be referred to the corresponding content in the foregoing method embodiments, which is not repeated herein.
The invention provides a recommendation analysis method and a recommendation analysis system based on big data and a social network, wherein a preset preference matrix is firstly obtained, the preset preference matrix comprises preference vectors of a plurality of preset users, each element position in the preference vectors represents an object to be recommended, each element value represents preference degree of the user to be recommended, then the preference vectors of the target users are obtained, according to sparsity of the preference vectors of the target users, behavior sparsity of the target users is obtained, if the behavior sparsity of the target users is larger than a preset threshold, target recommended objects of the target users are obtained based on a collaborative filtering algorithm according to the preference vectors of the target users and the preset preference matrix, if the behavior sparsity of the target users is smaller than the preset threshold, a plurality of social associated users of the target users and the preference vectors of the social associated users are obtained, and according to the preference vectors of the social associated users, the target recommended objects of the target users are obtained. Compared with the prior art, the invention improves the existing collaborative filtering algorithm, and judges whether the user can find the user with higher similarity or not through the action sparseness. If the behavior sparsity is high, the behavior of the representative user is high, and similar users can be found at the moment, and then the recommendation is performed by using the existing collaborative filtering algorithm. If the sparseness of the behaviors is low, the behaviors of the representative users are less, and similar users are difficult to find, then the social association users with social relations with the target users are acquired, and recommendation is performed through actual social association between the users, so that a more accurate recommendation effect is achieved. The method solves the problem of gray sheep in the existing collaborative filtering algorithm, is expected to provide more accurate and personalized recommendation service, and improves the experience of social network users.
It should be noted that, in the present specification, each embodiment is described in a progressive manner, and each embodiment is mainly described as different from other embodiments, and identical and similar parts between the embodiments are all enough to be referred to each other.
The previous description of the disclosed embodiments is provided to enable any person skilled in the art to make or use the present invention. Various modifications to these embodiments will be readily apparent to those skilled in the art, and the generic principles defined herein may be applied to other embodiments without departing from the spirit or scope of the invention. Thus, the present invention is not intended to be limited to the embodiments shown herein but is to be accorded the widest scope consistent with the principles and novel features disclosed herein.
Claims (9)
1. The recommendation analysis method based on the big data and the social network is characterized by comprising the following steps of:
acquiring a preset preference matrix, wherein the preset preference matrix comprises preference vectors of a plurality of preset users, each element position in the preference vectors represents an object to be recommended, and each element value represents the preference degree of the user to the object to be recommended;
acquiring a preference vector of a target user, and acquiring the behavior sparsity of the target user according to the sparsity of the preference vector of the target user;
if the behavior sparsity of the target user is larger than a preset threshold, obtaining a target recommended object of the target user based on a collaborative filtering algorithm according to the preference vector of the target user and the preset preference matrix;
if the behavior sparsity of the target user is smaller than a preset threshold, acquiring a plurality of social related users of the target user and preference vectors of the social related users, and acquiring a target recommended object of the target user according to the preference vectors of the social related users.
2. The recommendation analysis method based on big data and social networks according to claim 1, wherein the element values of the preference vector include 1, -1 and 0, wherein 1 indicates that the user has made a behavior indicating that the object to be recommended is favored, -1 indicates that the user has made a behavior indicating that the object to be recommended is disfavored, and 0 indicates that the user has not made a behavior indicating whether the object to be recommended is favored.
3. The recommendation analysis method based on big data and social networks according to claim 2, wherein the obtaining the behavior sparsity of the target user according to the sparsity of the preference vector of the target user includes:
and counting the number of non-zero elements in the preference vector of the target user, and taking the number of the non-zero elements as the behavior sparsity of the target user.
4. The recommendation analysis method based on big data and social network according to claim 1, wherein the obtaining the target recommendation object of the target user based on a collaborative filtering algorithm according to the preference vector of the target user and the preset preference matrix comprises:
calculating the similarity of the target user and the preset user based on the similarity of the preference vector of the target user and the preference vector in the preset preference matrix;
obtaining similar users of the target user from the preset users according to the similarity;
and obtaining the target recommended object of the target user according to the recommended object of the similar user.
5. The recommendation analysis method based on big data and social networks according to claim 2, wherein the obtaining the target recommended object of the target user according to preference vectors of a plurality of the social associated users comprises:
obtaining a social connection coefficient of each social connection user according to the social connection degrees of the target user and the social connection users;
taking the social association coefficient as the weight of each element in the preference vector, combining preference vectors of a plurality of social association users to obtain a preference prediction vector of the target user;
and obtaining a target recommended object of the target user according to the preference prediction vector of the target user.
6. The recommendation analysis method based on big data and social networks according to claim 5, wherein the preference prediction vector of the target user is obtained by:
wherein (1)>Preference prediction vector representing the target user, < >> Respectively indicate->A preference vector for each of the socially-associated users,respectively represent the target user and +>Social association coefficients of individual social associated users.
7. The recommendation analysis method based on big data and social network according to claim 5, wherein the obtaining the target recommended object of the target user according to the preference prediction vector of the target user comprises:
normalizing the preference prediction vector of the target user to obtain a preference simulation vector of the target user, wherein the element values of the preference simulation vector comprise 1, -1 and 0, wherein 1 represents that the target user is simulated to make a behavior representing a preference of an object to be recommended, -1 represents that the target user is simulated to make a behavior representing a non-preference of the object to be recommended, and 0 represents that the target user is simulated to not make a behavior representing whether the object to be recommended is favored;
and obtaining a target recommended object of the target user based on a collaborative filtering algorithm according to the preference simulation vector of the target user and the preset preference matrix.
8. The recommendation analysis method based on big data and social networks according to claim 7, wherein normalizing the preference prediction vector of the target user to obtain a preference simulation vector of the target user comprises:
acquiring a target element value, wherein the target element value is an element value to be normalized in a preference prediction vector of the target user;
acquiring a preset upper limit threshold and a preset lower limit threshold, wherein the preset upper limit threshold and the preset lower limit threshold are opposite numbers;
if the target element value is larger than the preset upper threshold, the element value corresponding to the target element value in the preference simulation vector is 1;
if the target element value is smaller than the preset lower limit threshold value, the element value corresponding to the target element value in the preference simulation vector is-1;
and if the target element value is between the preset upper limit threshold value and the preset lower limit threshold value, the element value corresponding to the target element value in the preference simulation vector is 0.
9. A recommendation analysis system based on big data and social networks, comprising:
the big data acquisition module is used for acquiring a preset preference matrix, wherein the preset preference matrix comprises preference vectors of a plurality of preset users, each element position in the preference vectors represents an object to be recommended, and each element value represents the preference degree of the user to the object to be recommended;
the user behavior analysis module is used for acquiring a preference vector of a target user and obtaining the behavior sparsity of the target user according to the sparsity of the preference vector of the target user;
the first recommendation module is used for obtaining a target recommendation object of the target user based on a collaborative filtering algorithm according to the preference vector of the target user and the preset preference matrix when the behavior sparsity of the target user is larger than a preset threshold;
the second recommendation module is used for acquiring a plurality of social related users of the target user and preference vectors of the social related users when the behavior sparseness of the target user is smaller than a preset threshold value, and acquiring target recommended objects of the target user according to the preference vectors of the social related users.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202311057493.8A CN117033807A (en) | 2023-08-22 | 2023-08-22 | Recommendation analysis method and system based on big data and social network |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202311057493.8A CN117033807A (en) | 2023-08-22 | 2023-08-22 | Recommendation analysis method and system based on big data and social network |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN117033807A true CN117033807A (en) | 2023-11-10 |
Family
ID=88642896
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202311057493.8A Pending CN117033807A (en) | 2023-08-22 | 2023-08-22 | Recommendation analysis method and system based on big data and social network |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN117033807A (en) |
-
2023
- 2023-08-22 CN CN202311057493.8A patent/CN117033807A/en active Pending
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN110321422B (en) | Method for training model on line, pushing method, device and equipment | |
| CN108921221B (en) | User feature generation method, device, equipment and storage medium | |
| CN110503531B (en) | Dynamic social scene recommendation method based on time sequence perception | |
| CN111797321B (en) | Personalized knowledge recommendation method and system for different scenes | |
| CN107862022B (en) | Culture resource recommendation system | |
| CN110717098A (en) | Meta-path-based context-aware user modeling method and sequence recommendation method | |
| EP4083857A1 (en) | Information prediction model training method and apparatus, information prediction method and apparatus, storage medium, and device | |
| CN111177559B (en) | Text travel service recommendation method and device, electronic equipment and storage medium | |
| CN115917535A (en) | Recommendation model training method, recommendation device and computer readable medium | |
| CN111611488A (en) | Information recommendation method and device based on artificial intelligence and electronic equipment | |
| CN111159563A (en) | Method, device and equipment for determining user interest point information and storage medium | |
| CN112149352B (en) | Prediction method for marketing activity clicking by combining GBDT automatic characteristic engineering | |
| CN113239182A (en) | Article recommendation method and device, computer equipment and storage medium | |
| CN111429161B (en) | Feature extraction method, feature extraction device, storage medium and electronic equipment | |
| CN113536105A (en) | Recommendation model training method and device | |
| CN115712780A (en) | Information pushing method and device based on cloud computing and big data | |
| CN113656699B (en) | User feature vector determining method, related equipment and medium | |
| CN113688306A (en) | Recommendation strategy generation method and device based on reinforcement learning | |
| CN114358807A (en) | User portrayal method and system based on predictable user characteristic attributes | |
| CN111177564B (en) | Product recommendation method and device | |
| CN113836388A (en) | Information recommendation method and device, server and storage medium | |
| CN115965439A (en) | Data recall method, data recall device, electronic equipment and storage medium | |
| CN115829683A (en) | Power integration commodity recommendation method and system based on inverse reward learning optimization | |
| CN115168700A (en) | Information flow recommendation method, system and medium based on pre-training algorithm | |
| CN117033807A (en) | Recommendation analysis method and system based on big data and social network |
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 |