CN117540093A - User behavior analysis method and system based on big data - Google Patents
User behavior analysis method and system based on big data Download PDFInfo
- Publication number
- CN117540093A CN117540093A CN202311564214.7A CN202311564214A CN117540093A CN 117540093 A CN117540093 A CN 117540093A CN 202311564214 A CN202311564214 A CN 202311564214A CN 117540093 A CN117540093 A CN 117540093A
- Authority
- CN
- China
- Prior art keywords
- user
- video
- recommendation
- acquiring
- heat value
- 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 50
- 238000000034 method Methods 0.000 claims abstract description 35
- 238000005457 optimization Methods 0.000 claims description 18
- 238000007621 cluster analysis Methods 0.000 claims description 12
- 235000019633 pungent taste Nutrition 0.000 claims description 10
- 230000008859 change Effects 0.000 abstract description 9
- 230000008569 process Effects 0.000 abstract description 7
- 238000007796 conventional method Methods 0.000 abstract 1
- 230000006399 behavior Effects 0.000 description 94
- 230000000694 effects Effects 0.000 description 15
- 238000004422 calculation algorithm Methods 0.000 description 11
- 238000004364 calculation method Methods 0.000 description 11
- 230000007246 mechanism Effects 0.000 description 5
- 230000001960 triggered effect Effects 0.000 description 5
- 238000012545 processing Methods 0.000 description 4
- 238000000605 extraction Methods 0.000 description 3
- 230000014759 maintenance of location Effects 0.000 description 3
- 230000003542 behavioural effect Effects 0.000 description 2
- 230000009286 beneficial effect Effects 0.000 description 2
- 230000003203 everyday effect Effects 0.000 description 2
- 238000012986 modification Methods 0.000 description 2
- 230000004048 modification Effects 0.000 description 2
- 238000010606 normalization Methods 0.000 description 2
- 230000002688 persistence Effects 0.000 description 2
- 230000001105 regulatory effect Effects 0.000 description 2
- 239000013598 vector Substances 0.000 description 2
- 230000004075 alteration Effects 0.000 description 1
- 230000015572 biosynthetic process Effects 0.000 description 1
- 238000007405 data analysis Methods 0.000 description 1
- 238000013461 design Methods 0.000 description 1
- 238000011161 development Methods 0.000 description 1
- 238000010586 diagram Methods 0.000 description 1
- 238000005516 engineering process Methods 0.000 description 1
- 238000011156 evaluation Methods 0.000 description 1
- 230000003993 interaction Effects 0.000 description 1
- 238000002372 labelling Methods 0.000 description 1
- 238000012544 monitoring process Methods 0.000 description 1
- 230000001737 promoting effect Effects 0.000 description 1
- 238000003786 synthesis reaction Methods 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
- G06F16/9535—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
- G06F16/70—Information retrieval; Database structures therefor; File system structures therefor of video data
- G06F16/73—Querying
- G06F16/735—Filtering based on additional data, e.g. user or group profiles
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F16/00—Information retrieval; Database structures therefor; File system structures therefor
- G06F16/70—Information retrieval; Database structures therefor; File system structures therefor of video data
- G06F16/75—Clustering; Classification
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F16/00—Information retrieval; Database structures therefor; File system structures therefor
- G06F16/70—Information retrieval; Database structures therefor; File system structures therefor of video data
- G06F16/78—Retrieval characterised by using metadata, e.g. metadata not derived from the content or metadata generated manually
- G06F16/7867—Retrieval characterised by using metadata, e.g. metadata not derived from the content or metadata generated manually using information manually generated, e.g. tags, keywords, comments, title and artist information, manually generated time, location and usage information, user ratings
Landscapes
- Engineering & Computer Science (AREA)
- Theoretical Computer Science (AREA)
- Databases & Information Systems (AREA)
- Data Mining & Analysis (AREA)
- Physics & Mathematics (AREA)
- General Engineering & Computer Science (AREA)
- General Physics & Mathematics (AREA)
- Multimedia (AREA)
- Library & Information Science (AREA)
- Computational Linguistics (AREA)
- Information Retrieval, Db Structures And Fs Structures Therefor (AREA)
Abstract
The invention provides a user behavior analysis method and a system based on big data, which are characterized in that basic information of a user is obtained, and a first image is obtained according to the basic information; further performing first recommendation to obtain a first recommendation result; searching records according to the first recommendation result and the user; acquiring a behavior record of a user, further acquiring a viscosity value and a first heat value of the user, acquiring the total heat value according to the first heat value, comparing the first heat value or the total heat value with a preset value, and determining whether to trigger first feedback according to a comparison result; according to the method and the system, the problem that the conventional method frequently recommends some content which is not happy for the user, repeatedly recommends the same type of content, cannot update the recommended content in time to adapt to the change of the user interest and the problem that the conventional analysis process is difficult to accurately quantify and analyze the satisfaction and the favorite degree of the user on the video is solved.
Description
Technical Field
The invention relates to the technical field of Internet, in particular to a user behavior analysis method and system based on big data.
Background
With the development and popularization of internet technology, the behavior data of users on a platform is continuously increased, and the data has important application value for the design and optimization of a recommendation system. The recommendation system needs to recommend related contents according to the behaviors and interests of the user, and the satisfaction degree and viscosity of the user are improved. However, in the conventional user behavior analysis method, there are some problems, such as inability to accurately acquire interests and behavior characteristics of a user, frequent recommendation of some content which is not happy, repeated recommendation of the same type of content, inability to update the recommended content in time to adapt to changes in interests of the user and inability to effectively process calculation of a large amount of user behavior data, and difficulty in accurately quantifying and analyzing satisfaction and favorability of the user for videos in the conventional analysis process.
Disclosure of Invention
The invention provides a user behavior analysis method and a system based on big data, which realize accurate acquisition of interests and behavior characteristics of users, update recommended content in time to adapt to changes of interests of users and effectively process calculation of a large amount of user behavior data, solve the problems that the traditional method can not accurately acquire the interests and behavior characteristics of users, frequently recommend some unbiased content, repeatedly recommend the same type of content, can not update recommended content in time to adapt to changes of interests of users and can not effectively process calculation of a large amount of user behavior data, and the traditional analysis process is difficult to accurately quantify and analyze satisfaction degree of users on videos:
The invention provides a user behavior analysis method based on big data, which comprises the following steps:
s1, acquiring basic information of a user, and acquiring a first image of the user according to the basic information of the user; performing first recommendation according to a first image of a user to obtain a first recommendation result;
s2, searching records according to the first recommendation result and the user; acquiring a behavior record of a user, acquiring a viscosity value and a first heat value of the user according to the behavior record of the user, acquiring the total heat value according to the first heat value, comparing the first heat value or the total heat value with a preset value, and determining whether to trigger first feedback according to a comparison result;
and S3, adjusting recommended content according to the first feedback result and continuously iterating and optimizing.
Further, a user behavior analysis method based on big data, the S1 includes:
acquiring basic information of a user, and acquiring a first image of the user according to the basic information of the user; the basic information comprises the age, sex, occupation, position information and hobbies of the user;
performing first recommendation according to a first image of a user; the first recommendation is a default recommendation, and the default recommendation comprises popular content of a region where a platform and/or a user are located, latest content of the region where the platform and/or the user are located, content which is ranked in the top 50% of browsing content of the first portrait user with the same type of the region, and recommended content which is loved according to the interests of the user.
Further, a user behavior analysis method based on big data, the S2 includes:
s21, according to a first recommendation result and a search record of a user; acquiring a user history browsing record, and acquiring a viscosity coefficient of a user according to the user history browsing record;
s22, dividing time intervals, counting video browsing amounts of users in each time interval, and acquiring a first heat value according to the video browsing amounts of the users and viscosity coefficients;
s23, acquiring key information of the user browsing videos in the same time interval, extracting features, and performing cluster analysis on the extracted features to obtain video classification; obtaining the total heat value of the user to the classified video under the same classification;
s24, comparing the first heat value or the total heat value with a preset value, and determining whether to trigger first feedback according to a comparison result; the first feedback includes selecting non-interesting content, whether to stop recommending the same type of content, and whether to re-make a default recommendation via keywords.
Further, a user behavior analysis method based on big data, the S21 includes:
obtaining a single continuous use duration and use times of a platform in the last week of a user to obtain a user viscosity value;
The user viscosity value is:
wherein Ni is the viscosity value of a user in the last week, S l For the average duration of single continuous use of the user in the last week, S la Continuously using the average duration for all users for a single time in the last week; c the last week is the using times of the user platform, C a The number of times all users used the platform in the last week;
normalizing the viscosity value to obtain a viscosity coefficient;
wherein f is the viscosity coefficient, N min For the minimum value of the viscosity values of all users in the last week, N max The viscosity value is maximal for all users in the last week.
Further, a user behavior analysis method based on big data, the S22 includes:
dividing each day into a plurality of time intervals, and obtaining a heat value of a user for each video in the same time interval, wherein the heat value is a first heat value;
wherein the first heat value is:
wherein t is i The playing time of a certain video is the playing time of the user; t is t zi Is the total duration of the video; a1 is a praise coefficient, a1=0 when the user does not praise the video, a1=1/P when the user praise the video, and P is the total amount of the video continuously browsed by the user at this time; a2 is a comment coefficient, a2=0 when the user does not comment on the video, a2=x/P when the user comments on the video, x being the number of times the user comments on the video; p is the total amount of the video continuously browsed by the user at this time; a3 is a collection coefficient, a3=0 when the user does not collect the video, a3=1/P when the user collects the video; a4 is a forwarding coefficient, a4=0 when the user does not forward the video, a4=1/P when the user collects the video; f is the viscosity coefficient of the user in the last week.
Further, a user behavior analysis method based on big data, the S23 includes:
acquiring key information of the user browsing videos in the same time interval, extracting features, and performing cluster analysis to acquire video classification;
obtaining the total amount of hotness values of the user to the video of the category under the same category;
wherein m is the number of videos of the user in the same category browsed in the time interval in the last week, and Ri is the first heat value.
Further, a big data-based user behavior analysis method, the S24 includes:
the user uses 35% S at the platform during a certain time interval l After the continuous time period, triggering first feedback when the first heat value Ri exceeding 80% is smaller than a first threshold value and the max (Ri) is smaller than a second threshold value; wherein S is l The average duration of single continuous use for the user in the last week;
in a certain time interval, ifGreater than or equal to a third threshold; continuing to push the classified video to the user and maintaining the original recommended frequency; if->Less than the third threshold but greater than or equal to the fourth threshold, reducing the recommendation frequency of the classified video; if->Triggering the first feedback if the first feedback is smaller than a fourth threshold;
Where k is the total number of videos browsed by the user in the time interval for the last week.
Further, a big data-based user behavior analysis method is characterized in that S3 includes:
obtaining topics of current interest of a user through first feedback; randomly recommending topics of current interest of the user to the user;
obtaining topics which are not interested by a user currently through first feedback and stopping recommending the topics which are not interested by the user;
acquiring a user restoration default recommendation option through first feedback, and re-recommending according to the default recommendation;
and acquiring the latest behavior record and feedback of the user after the recommendation is adjusted, and performing iterative optimization according to the latest behavior record and feedback.
The invention provides a user behavior analysis system based on big data, which comprises:
a first recommendation module: acquiring basic information of a user, and acquiring a first image of the user according to the basic information of the user; performing first recommendation according to a first image of a user to obtain a first recommendation result;
a first feedback module: searching records according to the first recommendation result and the user; acquiring a behavior record of a user, acquiring a viscosity value and a first heat value of the user according to the behavior record of the user, acquiring the total heat value according to the first heat value, comparing the first heat value or the total heat value with a preset value, and determining whether to trigger first feedback according to a comparison result;
And (3) an iteration optimization module: and adjusting the recommended content according to the first feedback result and continuously iterating and optimizing.
The invention has the beneficial effects that: by acquiring and analyzing the basic information of the user and the behavior records of the user, the interests and the demands of the user can be known more accurately, so that recommended content which meets the demands of the user is provided. Through the first recommendation result and the user search record, the recommendation content can be updated in time, the satisfaction degree and viscosity of the user are improved, and personalized recommendation service is provided, so that the user experience is better. By acquiring the historical browsing record of the user and calculating the viscosity coefficient of the user, the attention degree and the interest persistence of the user on the current content can be reflected more accurately. Meanwhile, the interests and the demands of the user can be further and deeply known by combining the search records of the user. And counting the video browsing amount of the user in each time interval by dividing the time interval, and acquiring a first heat value according to the viscosity coefficient of the user. The method can reflect the video browsing interest of the user in a certain time interval in real time, is convenient for updating the recommended content in time, and meanwhile divides a plurality of time intervals every day to acquire the heat value of the user to each video in each time interval. This allows separate processing and analysis of the user behavior data for each time interval, thereby reducing the data throughput and computational burden. And obtaining video classification by acquiring key information of the video browsed by the user, extracting features and performing cluster analysis. And under the same classification, counting the total quantity of the hotness values of the classified videos by the user. This may help the system better understand the user's interests and preferences for a certain class of videos, providing more personalized recommended content. And comparing the first heat value or the total heat value with a preset value, and determining whether to trigger the first feedback according to a comparison result. The first feedback includes options of selecting non-interesting content, whether to stop recommending the same type of content, and whether to re-make a default recommendation, etc. through keywords. This provides a richer and flexible recommendation adjustment mechanism that can better accommodate changes in user interests and meet user needs. In summary, the user behavior analysis method based on big data provides a richer and flexible recommendation adjustment mechanism, can improve the accuracy of recommendation, enhance the user experience, improve the system performance, realize iterative optimization and other advantages, and has important application value.
Drawings
Fig. 1 is a schematic diagram of a user behavior analysis method based on big data according to the present invention.
Detailed Description
In order that the above-recited objects, features and advantages of the present invention will be more clearly understood, a more particular description of the invention will be rendered by reference to specific embodiments thereof which are illustrated in the appended drawings. It should be noted that, in the case of no conflict, the embodiments of the present application and the features in the embodiments may be combined with each other.
In the following description, numerous specific details are set forth in order to provide a thorough understanding of the present invention, and the described embodiments are merely some, rather than all, embodiments of the present invention. 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.
Unless defined otherwise, all technical and scientific terms used herein have the same meaning as commonly understood by one of ordinary skill in the art to which this invention belongs. The terminology used herein in the description of the invention is for the purpose of describing particular embodiments only and is not intended to be limiting of the invention.
The embodiment provides a user behavior analysis method based on big data, which comprises the following steps:
S1, acquiring basic information of a user, and acquiring a first image of the user according to the basic information of the user; performing first recommendation according to a first image of a user to obtain a first recommendation result;
s2, searching records according to the first recommendation result and the user; acquiring a behavior record of a user, acquiring a viscosity value and a first heat value of the user according to the behavior record of the user, acquiring the total heat value according to the first heat value, comparing the first heat value or the total heat value with a preset value, and determining whether to trigger first feedback according to a comparison result;
and S3, adjusting recommended content according to the first feedback result and continuously iterating and optimizing.
The working principle of the technical scheme is as follows: basic information of the user is acquired, and a first portrait of the user is constructed according to the basic information of the user, such as age, gender, occupation, position information, hobbies and interests. Based on the first portrait of the user, a first recommendation, typically a default recommendation, is performed, including popular content of the platform and/or the region where the user is located, latest content of the platform and/or the region where the user is located, top 50% of the browsed content of the same type of first portrait user in the region, and recommended content according to the interest of the user.
And acquiring a behavior record of the user according to the first recommendation result and the user search record. These behavior records may include browsing history of the user, search records, click behavior, etc. From these behavioral records, the user's viscosity value and first heat value may be calculated. The viscosity value reflects the degree of interest of the user in the platform content, and the first heat value reflects the degree of interest of the user in the current content.
According to the first heat value, the total heat value can be obtained, namely, the first heat value is compared with a preset value. If the first heat value or the total amount of heat values is below a preset value, the user may be considered less interested in the current content or the like, at which point the first feedback may be triggered.
The first feedback includes selecting non-interesting content, whether to stop recommending the same type of content, whether to re-make a default recommendation, etc. by keywords. Based on the first feedback result, the recommended content may be adjusted, for example by adjusting parameters of the recommendation algorithm, or adding/subtracting certain types of recommended content, etc.
And (5) continuously performing iterative optimization: and acquiring the latest behavior record and feedback of the user after the recommendation is regulated, and performing iterative optimization according to the latest behavior record and feedback so that the recommendation system better meets the requirements and interests of the user.
The technical scheme has the effects that: by acquiring and analyzing the basic information of the user and the behavior records of the user, the interests and the demands of the user can be known more accurately, so that recommended content which meets the demands of the user is provided. Through the first recommendation result and the user search record, the recommendation content can be updated in time, the satisfaction degree and viscosity of the user are improved, and personalized recommendation service is provided, so that the user experience is better. The method can effectively solve the problems of calculation, storage and the like of a large amount of user behavior data, thereby improving the performance and stability of the system. By continuously acquiring the latest behavior record and feedback of the user after the recommendation is adjusted and performing iterative optimization, the recommendation system can be more intelligent and more accords with the requirements and interests of the user. In summary, the user behavior analysis method based on big data in the embodiment can improve the recommendation accuracy, enhance the user experience, improve the system performance, realize iterative optimization and other advantages, and has important application value.
The embodiment relates to a user behavior analysis method based on big data, wherein the S1 includes:
acquiring basic information of a user, and acquiring a first image of the user according to the basic information of the user; the basic information comprises the age, sex, occupation, position information and hobbies of the user; obtaining an area where a user is located according to the user position information, wherein the area can be province, city and the like;
Performing first recommendation according to a first image of a user; the first recommendation is a default recommendation, and the default recommendation comprises popular content of a region where a platform and/or a user are located, latest content of the region where the platform and/or the user are located, content which is ranked in the first 50% of browsing content of a first portrait user of the same type of the region and recommended content which is loved according to the interests of the user; taking 10 minutes as one period, and defaulting to not more than 0.25 of similarity of each recommended content in the first three periods; the similarity is cosine similarity of the recommended content features.
The working principle of the technical scheme is as follows:
basic information of a user is acquired: including the age, sex, occupation, location information, hobbies, etc. of the user. Such information may be obtained by way of user registration, filling out questionnaires, social media data, and the like. And classifying and labeling the users according to the basic information of the users to form a first image of the users. For example, users are categorized by age, gender, occupation, etc. for subsequent recommendation and personalized analysis. And carrying out first recommendation according to the first image of the user. The default recommended content may include the following aspects:
The hot content of the platform and/or the region where the user is located: and recommending the hot content which the user may be interested in according to the hot degree of the whole platform or the hot content of the area where the user is located.
The latest content of the platform and/or the region where the user is located: and recommending the latest released content to enable the user to know the fresh things in time.
The first portrait user of the same type of region browses content ranked in the top 50 percent: according to the first image of the user, other users similar to the first image are found, and according to the browsing behaviors of the users, the content ranked at the top is recommended.
Recommended content loved according to user interest: and recommending the content related to the interests of the user according to the interests and hobbies information of the user.
Through the working principle, the first portrait can be constructed according to the basic information of the user, default recommendation is carried out, and contents which are possibly interesting to the user and accord with the characteristics of the user are provided, so that the requirements of the user are met, and the user experience is improved.
The technical scheme has the effects that: by acquiring the basic information of the user and constructing the first portrait of the user, personalized recommendation can be realized. According to the information of the interests, the positions and the like of the users, the recommended content is matched with the preferences of the users, the content which meets the requirements of the users better is provided, and the user experience is enhanced. The default recommended content includes popular content, up-to-date content, and higher ranked content of browsed content of the same type of first portrait user for the platform and/or the region in which the user is located. The recommended content has higher popularity and relativity, can attract the interests of users and improve the satisfaction of the users. By recommending according to the interests and hobbies of the user, the participation of the user on the platform can be improved. Users may prefer to browse and participate in content related to their own interests, thereby increasing user interaction and liveness, increasing user dependence on the platform and viscosity. After the user obtains the recommended content which accords with the interests of the user, the user is more likely to stay on the platform for a long time, and the activity and the user viscosity of the platform are increased. By providing popular, latest and related contents through default recommendation, a user can more quickly find the contents interested by the user, and the consumption efficiency of the contents is improved.
The user behavior analysis method based on big data in this embodiment, the S2 includes:
s21, according to a first recommendation result and a search record of a user; acquiring a user history browsing record, and acquiring a viscosity coefficient of a user according to the user history browsing record;
s22, dividing time intervals, counting video browsing amounts of users in each time interval, and acquiring a first heat value according to the video browsing amounts of the users and viscosity coefficients;
s23, acquiring key information of the user browsing videos in the same time interval, extracting features, and performing cluster analysis on the extracted features to obtain video classification; obtaining the total heat value of the user to the classified video under the same classification;
s24, comparing the first heat value or the total heat value with a preset value, and determining whether to trigger first feedback according to a comparison result; the first feedback includes selecting non-interesting content, whether to stop recommending the same type of content, and whether to re-make a default recommendation via keywords.
The working principle of the technical scheme is as follows: firstly, according to a first recommendation result and a search record of a user, acquiring a history browsing record of the user. By analyzing the historical browsing records, the viscosity coefficient of the user can be calculated, and the attention degree and the interest durability of the user to the current content are reflected. Next, by dividing the time intervals, the video browsing amount of the user in each time interval can be counted. The first heat value can be calculated by combining the viscosity coefficient of the user, and the video browsing interest of the user in the current time interval is reflected. And simultaneously, acquiring key information of the video browsed by the user in the same time interval, extracting characteristics, and obtaining video classification through cluster analysis. Under the same classification, the total heat value of the user on the classified videos can be counted, and the interests and preferences of the user on a certain type of videos are further reflected. And finally, comparing the first heat value or the total heat value with a preset value, and determining whether to trigger the first feedback according to a comparison result. The first feedback includes options of selecting non-interesting content, whether to stop recommending the same type of content, and whether to re-make a default recommendation, etc. through keywords, providing a richer and flexible recommendation adjustment mechanism. Through the steps, the user behavior analysis method based on big data can conduct deep behavior analysis according to the browsing history, the search record, the first recommendation result and other data of the user, acquire the characteristics of the viscosity coefficient, the first heat value, the total heat value and the like of the user, trigger corresponding feedback operation according to preset rules, and achieve accurate recommendation and optimization adjustment.
The technical scheme has the effects that: by acquiring the historical browsing record of the user and calculating the viscosity coefficient of the user, the attention degree and the interest persistence of the user on the current content can be reflected more accurately. Meanwhile, the interests and the demands of the user can be further and deeply known by combining the search records of the user. And counting the video browsing amount of the user in each time interval by dividing the time interval, and acquiring a first heat value according to the viscosity coefficient of the user. The method can reflect the video browsing interest of the user in a certain time interval in real time, is convenient for updating the recommended content in time, and meanwhile divides a plurality of time intervals every day to acquire the heat value of the user to each video in each time interval. This allows separate processing and analysis of the user behavior data for each time interval, thereby reducing the data throughput and computational burden. And obtaining video classification by acquiring key information of the video browsed by the user, extracting features and performing cluster analysis. And under the same classification, counting the total quantity of the hotness values of the classified videos by the user. This may help the system better understand the user's interests and preferences for a certain class of videos, providing more personalized recommended content. And comparing the first heat value or the total heat value with a preset value, and determining whether to trigger the first feedback according to a comparison result. The first feedback includes options of selecting non-interesting content, whether to stop recommending the same type of content, and whether to re-make a default recommendation, etc. through keywords. This provides a richer and flexible recommendation adjustment mechanism that can better accommodate changes in user interests and meet user needs. In summary, the above steps can more accurately grasp the interests and demands of the user, reflect the interest changes of the user in real time, provide personalized recommended content, and have a flexible recommendation adjustment mechanism. This helps to improve user satisfaction and viscosity, and improves performance and effectiveness of the recommendation system.
The user behavior analysis method based on big data in this embodiment, the S21 includes:
obtaining a single continuous use duration and use times of a platform in the last week of a user to obtain a user viscosity value;
the user viscosity value is:
wherein Ni is the viscosity value of a user in the last week, S l For the average duration of single continuous use of the user in the last week, S la Continuously using the average duration for all users for a single time in the last week; c the last week is the using times of the user platform, C a The number of times all users used the platform in the last week;
normalizing the viscosity value to obtain a viscosity coefficient;
wherein f is the viscosity coefficient, N min For the minimum of all user viscosity values in the last week, N max Is the maximum of all user viscosity values in the last week.
The working principle of the technical scheme is as follows: in the last week, the duration and the number of times of single continuous use of the user on the platform are obtained. From these data, a viscosity value for the user is calculated. The viscosity value is an index for measuring the viscosity and loyalty of the user to the platform, and the viscosity value of the user is calculated by obtaining the single continuous use time length and the use times of the platform in the last week of the user through the following formula:
Wherein Ni is the viscosity value of a user in the last week, S l For the average duration of single continuous use of the user in the last week, S la Continuously using the average duration for all users for a single time in the last week; c the last week is the using times of the user platform, C a The number of times all users used the platform in the last week;
the viscosity values are normalized to obtain viscosity coefficients, and the viscosity values of different users can be converted into the range of [0,1] through normalization, so that the subsequent processing and comparison are facilitated;
through the steps, the user behavior analysis method based on big data can calculate the user viscosity value according to the single continuous use duration and the use times of the platform in the last week of the user, and the viscosity coefficient is obtained through normalization processing. The calculation can reflect the continuous use degree and loyalty of the user to the platform, and provides important basis for subsequent behavior analysis and recommendation.
The technical scheme has the effects that: the loyalty and the use habit of the user to the platform can be reflected more accurately by calculating the single continuous use duration and the use times of the platform in the last week of the user. The method is helpful for evaluating the interest and the dependence degree of the user on the platform, and provides basis for subsequent recommendation and marketing strategies. By normalizing the viscosity values, the viscosity values of different users are converted into the range of [0,1], so that the comparability among different users is improved. This helps identify users of higher viscosity among the user population and provides them with more personalized services and offers. By calculating the viscosity value of the user in the last week, the behavior change of the user on the platform can be reflected in real time. The method is beneficial to timely finding out the change and trend of the user interest, adjusting the recommended content and marketing strategy and improving the user experience and satisfaction. By taking the viscosity value as one of the considerations of the recommendation algorithm, the recommendation effect can be optimized. The viscosity value can reflect the interest and the dependence degree of the user, and a more comprehensive user portrait is provided for the recommendation system. This helps to improve accuracy and satisfaction of recommendations, and thus, user loyalty to the platform and frequency of use. In summary, the working principle of the big data-based user behavior analysis method in the step S21 can accurately reflect the user viscosity, improve the comparability among different users, reflect the user behavior change in real time and optimize the recommendation effect. These advantages help to improve user experience and satisfaction, promote continued use of platforms by users, and foster loyalty.
The user behavior analysis method based on big data in this embodiment, the S22 includes:
dividing each day into a plurality of time intervals, and obtaining a heat value of a user for each video in the same time interval, wherein the heat value is a first heat value; wherein each three hours is a time interval;
wherein the first heat value is:
wherein t is i The playing time of a certain video is the playing time of the user; t is t zi Is the total duration of the video; a1 is a praise coefficient, a1=0 when the user does not praise the video, a1=1/P when the user praise the video, and P is the total amount of the video continuously browsed by the user at this time; a2 is a comment coefficient, a2=0 when the user does not comment on the video, a2=x/P when the user comments on the video, x being the number of times the user comments on the video; p is the total amount of the video continuously browsed by the user at this time; a3 is the collection coefficient, whenA3=0 when the user does not collect the video, a3=1/P when the user collects the video; a4 is a forwarding coefficient, a4=0 when the user does not forward the video, a4=1/P when the user collects the video; f is the viscosity coefficient of the user in the last week.
The working principle of the technical scheme is as follows: dividing time intervals: first, each day is divided into a plurality of time intervals, one time interval every three hours. This allows the user to analyze and compare the heat of each video over different time periods.
And (5) calculating a heat value: for each video in each time interval, its heat value is calculated. The hotness value is defined as a first hotness value (Ri), where i denotes the sequence number of the video. The calculation formula of the heat value is as follows:
if the playing time t of the video i Less than the minimum playing time t min The heat value is 0.
If the playing time of the video is between the minimum playing time and twice the total time (2t_zi), the heat value is t i Divided by t zi Multiplied by a coefficient.
If the playing time of the video is longer than twice the total time, the heat value is 2 times a coefficient.
Coefficient calculation: the calculation of the heat value involves four coefficients (a 1, a2, a3, a 4) and a viscosity coefficient (f).
a1 represents a praise coefficient, and if the user does not praise the video, a1 is 0; if the user has endorsed the video, a1 is 1 divided by the total number of videos the user is continuously browsing this time (P).
a2 represents a comment coefficient, and if the user does not comment on the video, a2 is 0; if the user commented on the video, a2 is the number of user comments on the video (x) divided by P.
a3 represents a collection coefficient, and if the user does not collect the video, a3 is 0; if the user has collected the video, a3 is 1 divided by P.
a4 represents a forwarding coefficient, and if the user does not forward the video, a4 is 0; if the user forwarded the video, a4 is 1 divided by P.
f represents the viscosity coefficient of the user in the last week.
And (5) heat value calculation and synthesis: and applying the four coefficients and the viscosity coefficient to a heat value calculation formula to obtain a final first heat value (Ri). If the playing time length of the video is smaller than the minimum playing time length, the heat value is 0; if the playing time length is between the minimum playing time length and twice the total time length, the heat value is t i Divided by t zi Multiplying by a coefficient; if the play time is longer than twice the total time, the heat value is 2 times a coefficient.
By the method, the popularity value of each video can be calculated according to the playing time, praise, comment, collection, forwarding and other behaviors of the user on each video and by combining the viscosity coefficient of the user. The interest and the preference degree of the user on different videos can be accurately evaluated, and personalized recommendation and customized services are provided for the platform. Meanwhile, the strategy can be timely adjusted according to the behavior mode and viscosity change of the user, and the retention and the liveness of the user are improved.
The technical scheme has the effects that: by comprehensively considering various behaviors of the user on the video, such as playing time, praise, comments, collection, forwarding and the like, and the viscosity coefficient of the user, the formula can more comprehensively reflect the interest and participation of the user on the video. This helps to improve the accuracy and diversity of the recommendation algorithm, and provides the user with recommended content that better meets his interests. By dividing each day into a plurality of time intervals, the heat value of the user to each video in each time interval can be counted in real time. The method is helpful for timely finding the change and trend of the interest of the user, and provides more real-time and accurate data support for the recommendation system. According to the change of the heat value of the user in different time intervals, the recommended content and the marketing strategy can be flexibly adjusted so as to meet the requirements and interests of the user. By calculating the praise coefficient, comment coefficient, collection coefficient and forwarding coefficient of each user, more personalized recommended content can be provided for the user according to the personalized behavior characteristics of the user. This helps to improve the accuracy and satisfaction of the recommendations, meeting the needs and preferences of different users. The formula not only considers the playing time of the user, but also considers actions such as praise, comment, collection, forwarding and the like of the user. These actions may reflect the user's different preferences and needs for video, thereby providing the user with more varied recommendations.
In summary, the working principle of the big data-based user behavior analysis method in the step S22 can improve the accuracy and individuation degree of recommendation, and meanwhile, the diversity and the real-time performance of the user behavior are considered. This helps to improve user experience and satisfaction, promoting continued use of the platform and growth of loyalty by the user.
The user behavior analysis method based on big data in this embodiment, the S23 includes:
acquiring key information of the user browsing videos in the same time interval, extracting features, and performing cluster analysis to acquire video classification;
obtaining the total amount of hotness values of the user to the video of the category under the same category;
wherein m is the number of videos of the user in the same category browsed in the time interval in the last week, and Ri is the first heat value.
The working principle of the technical scheme is as follows: acquiring user behavior data: firstly, the method needs to collect behavior data of a user on a platform, wherein the behavior data comprise key information, praise, comment, collection, forwarding and the like of the user browsing videos. Then, the method performs feature extraction on the user behavior data and performs cluster analysis. Specifically, the method classifies videos browsed by the user in the same time interval according to key information thereof, such as video theme, type, actors and the like. Then, by performing feature extraction on the hotness value of each video, a feature vector can be obtained. Finally, all feature vectors are partitioned into different video classifications by a clustering algorithm. Under the same category, the method calculates the total amount of hotness value of the video of the category. Specifically, the method calculates the total heat value Rzj of the user for the video classification according to the number m of videos in the same classification browsed by the user in the time interval and the first heat value Ri of each video. Finally, the method provides personalized recommendation for the user according to the total amount of hotness values of the user for different video classifications. For example, if a user browses many music videos in the last week, the method may recommend more music videos to the user. This may improve user satisfaction and use experience.
The technical scheme has the effects that: the video classification can be more accurately obtained by acquiring key information of the video browsed by the user, extracting the characteristics and performing cluster analysis. This helps to improve the accuracy and diversity of the recommendation algorithm, and provides the user with recommended content that better meets his interests. The total heat value of the user to the video of the category is obtained under the same category, so that the user's interest and participation in the category can be more deeply known. This helps to discover potential needs and preferences of the user, providing more accurate data support for the recommender system. By accumulating the heat values of the video by the user in the same category, the total heat value of the user in this category can be obtained. This helps to optimize the recommendation algorithm, providing more personalized recommended content for the user. By acquiring key information and extracting features of the user browsing videos in real time, the change and trend of the user interests can be found in time, and recommended content and marketing strategies can be adjusted. This helps to improve the real-time and accuracy of the recommendation, meeting the needs and preferences of the user. In summary, the working principle of the big data-based user behavior analysis method in the step S23 can improve the accuracy of video classification, the depth of user behavior analysis and the optimization of recommendation algorithm. Meanwhile, by acquiring key information and extraction characteristics of the video browsed by the user in real time, recommended content and marketing strategies can be timely adjusted, and the real-time performance and accuracy of recommendation are improved. These advantages help to improve user experience and satisfaction, promote continued use of platforms by users, and foster loyalty.
The user behavior analysis method based on big data in this embodiment, the S24 includes:
the user uses 35% S at the platform during a certain time interval l After the continuous time period, triggering first feedback when the first heat value Ri exceeding 80% is smaller than a first threshold value and the max (Ri) is smaller than a second threshold value; wherein S is l The average duration of single continuous use for the user in the last week; the platform sends feedback options to the user; wherein the first threshold is less than the second threshold; the first threshold value ranges from 0.2 to 0.4, preferably 0.3, and the second threshold value ranges from 0.4 to 0.6; preferably 0.5;
in a certain time interval, ifGreater than or equal to a third threshold; continuing to push the classified video to the user and maintaining the original recommended frequency; if->Less than the third threshold but greater than or equal to the fourth threshold, reducing the recommendation frequency of the classified video; if->Triggering the first feedback if the first feedback is smaller than a fourth threshold; the platform sends feedback options to the user; the third threshold is greater than the fourth threshold; the third threshold value is in the range of 0.5 to 0.8, preferably 0.5; the fourth threshold range is 0.1 to 0.3, preferably 0.2;
where k is the total number of videos browsed by the user in the time interval for the last week.
The working principle of the technical scheme is as follows: firstly, the method can know the interests and the preferences of the user for different video classifications by carrying out cluster analysis on the key information of the user browsing videos in the same time interval, and calculate the total heat value of each classification. The user uses 35% S at the platform during a certain time interval l After a continuous period of time, when more than 80% of the first heat value Ri is less than a first threshold value and max (Ri) is less than a second threshold value, a first feedback is triggered. This means that the user is not interested or satisfied with the currently recommended video enough, requiring the platform to provide a more consistent videoRecommendation of interests. The platform sends feedback options to the user to know the specific requirements and feedback comments, so that the personalized requirements of the user can be better met.
In the following recommendation process, when under a certain classificationWhen the video frequency is greater than or equal to a third threshold value, the platform can continue to push the classified video frequency to the user, and the original recommended frequency is kept. />
When (when)When the third threshold is less than but greater than or equal to the fourth threshold, the platform may reduce the recommendation frequency of the classified video to avoid pushing uninteresting content to the user too frequently.
When (when)When the feedback request is smaller than the fourth threshold value, the first feedback is triggered, and the platform sends feedback options to the user so as to know the specific requirements and feedback comments of the user, so that the personalized requirements of the user are better met.
Through the user behavior analysis method based on big data, the platform can better understand the user demands and behaviors and provide personalized recommendation and service, so that the user satisfaction and the platform operation effect are improved.
The technical scheme has the effects that: by analyzing the browsing behavior and interest preference of the user, the platform can provide more personalized video recommendation, so that the user can more easily find the content conforming to the interest of the user, and the user experience and satisfaction degree are improved. The dissatisfaction or discomfort of the user on the recommended content can be timely found by monitoring the using time of the user on the platform and the video heat value, and the first feedback is triggered. According to the heat value and threshold setting of video classification, the platform can intelligently adjust the recommendation frequency of different classified videos. When the video heat under a certain category is higher, the original recommended frequency is kept, so that the user can obtain interested contents in time; and when the video heat under a certain category is low, the recommendation frequency is reduced, and the content which is not interested is prevented from being excessively pushed to the user. The platform sends feedback options to the user, so that the user can participate in the adjustment of the recommended content, and the user experience and satisfaction are improved. Users will feel the platform's attention to them and personalized services, thereby preferring to use the platform and share to others. Through big data analysis, the behavior and the demand of the user can be comprehensively known, so that the target user group can be more accurately positioned. Therefore, the operation efficiency of the platform can be improved, and the user growth and service expansion are realized. In summary, the user behavior analysis method based on big data can help the platform to realize personalized recommendation, promote user participation, optimize recommendation frequency and promote platform operation effect, thereby improving user satisfaction and platform competitiveness.
The user behavior analysis method based on big data in this embodiment, the S3 includes:
obtaining topics of current interest of a user through first feedback; randomly recommending topics of current interest of the user to the user;
obtaining topics which are not interested by a user currently through first feedback and stopping recommending the topics which are not interested by the user;
acquiring a user restoration default recommendation option through first feedback, and re-recommending according to the default recommendation;
and acquiring the latest behavior record and feedback of the user after the recommendation is adjusted, and performing iterative optimization according to the latest behavior record and feedback.
The working principle of the technical scheme is as follows: when a user interacts with the platform, the platform gathers feedback information about the user, such as clicking, browsing, collecting, etc., and determines topics of current interest to the user by analyzing the behavior. According to the current interest topics of the user, the platform randomly selects contents from the corresponding topic classification to recommend the contents. The purpose of this is to keep the recommendation diverse, avoiding too single and predictive recommendations, to provide a richer and interesting experience. If the user expresses uninteresting feedback on a topic, the platform records and stops recommending related topic contents to the user, so that discontents and interference to the user are avoided. If the user wishes to restore the default recommendation option, the platform recommends according to the default recommendation algorithm. This may satisfy the need for the user to adjust for new interests or previous feedback. The platform can continuously monitor the latest behavior records and feedback of the user, including clicking, browsing, collecting and the like, and the evaluation and feedback of the recommended content by the user. By analyzing the data, the platform can continuously optimize the recommendation algorithm and the model, and the accuracy and the individuation degree of recommendation are improved.
The technical scheme has the effects that: by continuously learning and adapting to the interest change of the user, the method can provide more accurate and personalized recommended content, thereby improving the satisfaction degree of the user. By analyzing the user behavior records and feedback, the method can optimize the recommendation algorithm and the model, and improve the accuracy and the individuation degree of recommendation, thereby improving the service quality of the platform. By providing more interesting and useful recommended content, the method can increase the dependence and the use frequency of the user on the platform, thereby improving the retention rate of the user. By stopping recommending topics which are not interested by the user, the method can avoid discontent and interference to the user, so that the user loss rate is reduced. By improving the user satisfaction and the retention rate, the method can increase the advertisement click rate and the number of paid users, thereby increasing the income of the platform.
In summary, the user behavior analysis method based on big data can improve the service quality of the platform and the user satisfaction, and reduce the user loss rate, thereby realizing win-win effect.
The embodiment provides a user behavior analysis system based on big data, the system includes:
a first recommendation module: acquiring basic information of a user, and acquiring a first image of the user according to the basic information of the user; performing first recommendation according to a first image of a user to obtain a first recommendation result;
A first feedback module: searching records according to the first recommendation result and the user; acquiring a behavior record of a user, acquiring a viscosity value and a first heat value of the user according to the behavior record of the user, acquiring the total heat value according to the first heat value, comparing the first heat value or the total heat value with a preset value, and determining whether to trigger first feedback according to a comparison result;
and (3) an iteration optimization module: and adjusting the recommended content according to the first feedback result and continuously iterating and optimizing.
The working principle of the technical scheme is as follows: basic information of the user is acquired, and a first portrait of the user is constructed according to the basic information of the user, such as age, gender, occupation, position information, hobbies and interests. Based on the first portrait of the user, a first recommendation, typically a default recommendation, is performed, including popular content of the platform and/or the region where the user is located, latest content of the platform and/or the region where the user is located, top 50% of the browsed content of the same type of first portrait user in the region, and recommended content according to the interest of the user.
And acquiring a behavior record of the user according to the first recommendation result and the user search record. These behavior records may include browsing history of the user, search records, click behavior, etc. From these behavioral records, the user's viscosity value and first heat value may be calculated. The viscosity value reflects the degree of interest of the user in the platform content, and the first heat value reflects the degree of interest of the user in the current content.
According to the first heat value, the total heat value can be obtained, namely, the first heat value is compared with a preset value. If the first heat value or the total amount of heat values is below a preset value, the user may be considered less interested in the current content or the like, at which point the first feedback may be triggered.
The first feedback includes selecting non-interesting content, whether to stop recommending the same type of content, whether to re-make a default recommendation, etc. by keywords. Based on the first feedback result, the recommended content may be adjusted, for example by adjusting parameters of the recommendation algorithm, or adding/subtracting certain types of recommended content, etc.
And (5) continuously performing iterative optimization: and acquiring the latest behavior record and feedback of the user after the recommendation is regulated, and performing iterative optimization according to the latest behavior record and feedback so that the recommendation system better meets the requirements and interests of the user.
The technical scheme has the effects that: by acquiring and analyzing the basic information of the user and the behavior records of the user, the interests and the demands of the user can be known more accurately, so that recommended content which meets the demands of the user is provided. Through the first recommendation result and the user search record, the recommendation content can be updated in time, the satisfaction degree and viscosity of the user are improved, and personalized recommendation service is provided, so that the user experience is better. The method can effectively solve the problems of calculation, storage and the like of a large amount of user behavior data, thereby improving the performance and stability of the system. By continuously acquiring the latest behavior record and feedback of the user after the recommendation is adjusted and performing iterative optimization, the recommendation system can be more intelligent and more accords with the requirements and interests of the user. In summary, the user behavior analysis method based on big data in the embodiment can improve the recommendation accuracy, enhance the user experience, improve the system performance, realize iterative optimization and other advantages, and has important application value.
It will be apparent to those skilled in the art that various modifications and variations can be made to the present invention without departing from the spirit or scope of the invention. Thus, it is intended that the present invention also include such modifications and alterations insofar as they come within the scope of the appended claims or the equivalents thereof.
Claims (9)
1. A method for analyzing user behavior based on big data, the method comprising:
s1, acquiring basic information of a user, and acquiring a first image of the user according to the basic information of the user; performing first recommendation according to a first image of a user to obtain a first recommendation result;
s2, searching records according to the first recommendation result and the user; acquiring a behavior record of a user, acquiring a viscosity value and a first heat value of the user according to the behavior record of the user, acquiring the total heat value according to the first heat value, comparing the first heat value or the total heat value with a preset value, and determining whether to trigger first feedback according to a comparison result;
and S3, adjusting recommended content according to the first feedback result and continuously iterating and optimizing.
2. The big data based user behavior analysis method of claim 1, wherein S1 comprises:
Acquiring basic information of a user, and acquiring a first image of the user according to the basic information of the user; the basic information comprises the age, sex, occupation, position information and hobbies of the user;
performing first recommendation according to a first image of a user; the first recommendation is a default recommendation, and the default recommendation comprises popular content of a region where a platform and/or a user are located, latest content of the region where the platform and/or the user are located, content which is ranked in the top 50% of browsing content of the first portrait user with the same type of the region, and recommended content which is loved according to the interests of the user.
3. The big data based user behavior analysis method of claim 1, wherein S2 comprises:
s21, according to a first recommendation result and a search record of a user; acquiring a user history browsing record, and acquiring a viscosity coefficient of a user according to the user history browsing record;
s22, dividing time intervals, counting video browsing amounts of users in each time interval, and acquiring a first heat value according to the video browsing amounts of the users and viscosity coefficients;
s23, acquiring key information of the user browsing videos in the same time interval, extracting features, and performing cluster analysis on the extracted features to obtain video classification; obtaining the total heat value of the user to the classified video under the same classification;
S24, comparing the first heat value or the total heat value with a preset value, and determining whether to trigger first feedback according to a comparison result; the first feedback includes selecting non-interesting content, whether to stop recommending the same type of content, and whether to re-make a default recommendation via keywords.
4. A method of analyzing user behavior based on big data according to claim 3, wherein S21 comprises:
obtaining a single continuous use duration and use times of a platform in the last week of a user to obtain a user viscosity value;
the user viscosity value is:
wherein Ni is the viscosity value of a user in the last week, S l For the average duration of single continuous use of the user in the last week, S la Continuously using the average duration for all users for a single time in the last week; c the last week is the using times of the user platform, C a The number of times all users used the platform in the last week;
normalizing the viscosity value to obtain a viscosity coefficient;
wherein f is the viscosity coefficient, N min For the minimum value of the viscosity values of all users in the last week, N max The viscosity value is maximal for all users in the last week.
5. A method of analyzing user behavior based on big data according to claim 3, wherein S22 comprises:
Dividing each day into a plurality of time intervals, and obtaining a heat value of a user for each video in the same time interval, wherein the heat value is a first heat value;
wherein the first heat value is:
wherein t is i The playing time of a certain video is the playing time of the user; t is t zi Is the total duration of the video; a1 is a praise coefficient, a1=0 when the user does not praise the video, a1=1/P when the user praise the video, and P is the total amount of the video continuously browsed by the user at this time; a2 is a comment coefficient, a2=0 when the user does not comment on the video, a2=x/P when the user comments on the video, x being the number of times the user comments on the video; p is the total amount of the video continuously browsed by the user at this time; a3 is a collection coefficient, a3=0 when the user does not collect the video, a3=1/P when the user collects the video; a4 is a forwarding coefficient, a4=0 when the user does not forward the video, a4=1/P when the user collects the video; f is the viscosity coefficient of the user in the last week.
6. A method of analyzing user behavior based on big data according to claim 3, wherein said S23 comprises:
acquiring key information of the user browsing videos in the same time interval, extracting features, and performing cluster analysis to acquire video classification;
Obtaining the total amount of hotness values of the user to the video of the category under the same category;
wherein m is the number of videos of the user in the same category browsed in the time interval in the last week, and Ri is the first heat value.
7. The big data based user behavior analysis method of claim 6, wherein S24 comprises:
the user uses 35% S at the platform during a certain time interval l After a continuous period of time, when more than 80% of the first heat value Ri is less than the first threshold value and max (Ri) is less than the second threshold valueTriggering a first feedback; wherein S is l The average duration of single continuous use for the user in the last week;
in a certain time interval, ifGreater than or equal to a third threshold; continuing to push the classified video to the user and maintaining the original recommended frequency; if->Less than the third threshold but greater than or equal to the fourth threshold, reducing the recommendation frequency of the classified video; if->Triggering the first feedback if the first feedback is smaller than a fourth threshold;
where k is the total number of videos browsed by the user in the time interval for the last week.
8. The big data based user behavior analysis method of claim 1, wherein S3 comprises:
Obtaining topics of current interest of a user through first feedback; randomly recommending topics of current interest of the user to the user;
obtaining topics which are not interested by a user currently through first feedback and stopping recommending the topics which are not interested by the user;
acquiring a user restoration default recommendation option through first feedback, and re-recommending according to the default recommendation;
and acquiring the latest behavior record and feedback of the user after the recommendation is adjusted, and performing iterative optimization according to the latest behavior record and feedback.
9. A big data based user behavior analysis system, the system comprising:
a first recommendation module: acquiring basic information of a user, and acquiring a first image of the user according to the basic information of the user; performing first recommendation according to a first image of a user to obtain a first recommendation result;
a first feedback module: searching records according to the first recommendation result and the user; acquiring a behavior record of a user, acquiring a viscosity value and a first heat value of the user according to the behavior record of the user, acquiring the total heat value according to the first heat value, comparing the first heat value or the total heat value with a preset value, and determining whether to trigger first feedback according to a comparison result;
And (3) an iteration optimization module: and adjusting the recommended content according to the first feedback result and continuously iterating and optimizing.
Priority Applications (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN202311564214.7A CN117540093A (en) | 2023-11-21 | 2023-11-21 | User behavior analysis method and system based on big data |
Applications Claiming Priority (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN202311564214.7A CN117540093A (en) | 2023-11-21 | 2023-11-21 | User behavior analysis method and system based on big data |
Publications (1)
Publication Number | Publication Date |
---|---|
CN117540093A true CN117540093A (en) | 2024-02-09 |
Family
ID=89793435
Family Applications (1)
Application Number | Title | Priority Date | Filing Date |
---|---|---|---|
CN202311564214.7A Pending CN117540093A (en) | 2023-11-21 | 2023-11-21 | User behavior analysis method and system based on big data |
Country Status (1)
Country | Link |
---|---|
CN (1) | CN117540093A (en) |
Cited By (1)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN117972137A (en) * | 2024-04-01 | 2024-05-03 | 深圳市致尚信息技术有限公司 | Intelligent supervision and analysis method, system and storage medium for video data |
Citations (8)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
US20130046772A1 (en) * | 2011-08-16 | 2013-02-21 | Alibaba Group Holding Limited | Recommending content information based on user behavior |
CN105930425A (en) * | 2016-04-18 | 2016-09-07 | 乐视控股(北京)有限公司 | Personalized video recommendation method and apparatus |
CN106326474A (en) * | 2016-08-31 | 2017-01-11 | 南京炫佳网络科技有限公司 | Anime video personalized recommendation method |
CN113015010A (en) * | 2021-03-16 | 2021-06-22 | 杭州网易云音乐科技有限公司 | Push parameter determination method, device, equipment and computer readable storage medium |
CN114466250A (en) * | 2020-11-09 | 2022-05-10 | 江苏华软智能信息科技有限公司 | Video recommendation method |
CN115098777A (en) * | 2022-06-29 | 2022-09-23 | 浙江理工大学科技与艺术学院 | User personalized recommendation method and system based on data analysis |
CN115757952A (en) * | 2022-11-17 | 2023-03-07 | 中国平安财产保险股份有限公司 | Content information recommendation method, device, equipment and storage medium |
CN116894709A (en) * | 2023-07-14 | 2023-10-17 | 广州洋葱时尚集团有限公司 | Advertisement commodity recommendation method and system |
-
2023
- 2023-11-21 CN CN202311564214.7A patent/CN117540093A/en active Pending
Patent Citations (8)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
US20130046772A1 (en) * | 2011-08-16 | 2013-02-21 | Alibaba Group Holding Limited | Recommending content information based on user behavior |
CN105930425A (en) * | 2016-04-18 | 2016-09-07 | 乐视控股(北京)有限公司 | Personalized video recommendation method and apparatus |
CN106326474A (en) * | 2016-08-31 | 2017-01-11 | 南京炫佳网络科技有限公司 | Anime video personalized recommendation method |
CN114466250A (en) * | 2020-11-09 | 2022-05-10 | 江苏华软智能信息科技有限公司 | Video recommendation method |
CN113015010A (en) * | 2021-03-16 | 2021-06-22 | 杭州网易云音乐科技有限公司 | Push parameter determination method, device, equipment and computer readable storage medium |
CN115098777A (en) * | 2022-06-29 | 2022-09-23 | 浙江理工大学科技与艺术学院 | User personalized recommendation method and system based on data analysis |
CN115757952A (en) * | 2022-11-17 | 2023-03-07 | 中国平安财产保险股份有限公司 | Content information recommendation method, device, equipment and storage medium |
CN116894709A (en) * | 2023-07-14 | 2023-10-17 | 广州洋葱时尚集团有限公司 | Advertisement commodity recommendation method and system |
Cited By (2)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN117972137A (en) * | 2024-04-01 | 2024-05-03 | 深圳市致尚信息技术有限公司 | Intelligent supervision and analysis method, system and storage medium for video data |
CN117972137B (en) * | 2024-04-01 | 2024-06-04 | 深圳市致尚信息技术有限公司 | Intelligent supervision and analysis method, system and storage medium for video data |
Similar Documents
Publication | Publication Date | Title |
---|---|---|
US11288702B1 (en) | Exploration in a real time messaging platform | |
Unger et al. | Towards latent context-aware recommendation systems | |
US10866987B2 (en) | Evaluating performance of recommender system | |
US7921069B2 (en) | Granular data for behavioral targeting using predictive models | |
US10102307B2 (en) | Method and system for multi-phase ranking for content personalization | |
US7644427B1 (en) | Time-centric training, interference and user interface for personalized media program guides | |
US9760907B2 (en) | Granular data for behavioral targeting | |
CN110717093B (en) | Movie recommendation system and method based on Spark | |
US20120185481A1 (en) | Method and Apparatus for Executing a Recommendation | |
Elena et al. | Survey of multiarmed bandit algorithms applied to recommendation systems | |
CN107077690A (en) | The method for orienting the advertisement based on Web | |
CN117540093A (en) | User behavior analysis method and system based on big data | |
CN111026906B (en) | Recommendation system for streaming audio content listening in vehicle-mounted scene | |
CN116843388B (en) | Advertisement delivery analysis method and system | |
Yalcin et al. | Novel automatic group identification approaches for group recommendation | |
CN113239182A (en) | Article recommendation method and device, computer equipment and storage medium | |
Wang et al. | Modeling uncertainty to improve personalized recommendations via Bayesian deep learning | |
Wang et al. | Personalization on a peer-to-peer television system | |
US8429175B2 (en) | Method and apparatus for default rating estimation | |
KR20170060828A (en) | User's Contents Access History based Recommendation Method | |
Ficel et al. | A graph-based recommendation approach for highly interactive platforms | |
Elahi | Empirical evaluation of active learning strategies in collaborative filtering | |
Darvishi-Mirshekarlou et al. | Reviewing cluster based collaborative filtering approaches | |
Darvishy et al. | New attributes for neighborhood-based collaborative filtering in news recommendation | |
Kawamae | Real time recommendations from connoisseurs |
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 |