534results about How to "Improve recommendation effect" patented technology

A recommendation system based on graph convolution technology comprises a preprocessing module, a heterogeneous graph generation module, a model training module and a recommendation result generationmodule, wherein, the preprocessing module cleans the interaction records of the user and the article and performs the data cleaning and the format standardization operation, and generates the interaction sequence for each user and outputs the interaction sequence to the heterogeneous graph generation module; The heterogeneous graph generation module constructs three heterogeneous graphs representing user preferences, dependencies among items and similarities among users according to user interaction sequence data, and outputs the generated graph structure data to the model training module. Themodel training module trains the graph convolution model based on graph structure data and generates vector representation for each user and object. The recommendation result generation module calculates the user's preference for all items according to the vector expression, and generates the final recommendation result. The invention solves the problem that the number of the neighbors of each node is not equal, and the information of the neighbors of the nodes in the heterogeneous graph is mined by the convolution operation, so that the recommendation effect is improved.

The invention discloses a multilayer quotation recommendation method based on a literature content mapping knowledge domain, and belongs to the field of information recommendation and intelligent information processing. The method comprises the following steps: firstly, obtaining the query requirement of a user, wherein the query requirement consists of the key words of the title and the digest of a thesis which needs to recommend a quotation thesis or quotation literature; then, on the basis of the literature content mapping knowledge domain, expanding and querying a retrieval word, wherein the mapping knowledge domain consists of the research object word and the research behavior word node of the literature, and edges which express various semantic relations including synonymy, synonym, an up and down position, part-whole, juxtaposition and the like; and finally, constructing the inverted index of the literature in a data set, selecting a candidate quotation, calculating the similarity between the candidate quotation and query, and adopting a gradient progressive regression tree to carry out quotation recommendation. The method carries out multilayer quotation recommendation on the basis of the literature content mapping knowledge domain, enlarges the range of the candidate quotation, accurately expresses the research object and contents of the thesis, improves efficiency for users to obtain a relevant literature and has a wide application prospect.

The invention discloses a content recommending method and system. The method comprises the steps of collecting user original behavior data through an interactive program script and conducting statistics on the collected user original behavior data to obtain quantized user quantitative behavior data; obtaining a time factor of user behaviors by simulating human life activity; scoring how much the user likes an attended content to obtain the scoring situation; inputting the scoring condition into the collaborative filtering recommendation to obtain a recommended result, and pushing the recommended result to the user.

The invention provides a training method and device of a recommendation model, and a recommendation method and device, and the training method of a recommendation model comprises the steps: obtainingthe user characteristics of at least two sample users and the attribute characteristics of at least two sample application programs; generating a positive sample which is clicked and purchased by thesample user on the exposed sample application program and a negative sample which is clicked but not purchased or unclicked by the sample user on the exposed sample application program based on the user characteristics and the attribute characteristics; and training a recommendation model on the basis of a sample set comprising at least one positive sample and at least one negative sample to obtain the recommendation model, and outputting an exposure conversion rate obtained by each sample user on the basis of the click rate and the purchase rate of each exposed sample application program by the recommendation model.

The invention discloses a recommendation method based on deep learning, belongs to the technical field of data mining, and solves the problems of an existing recommendation method that the potential factor vector of a project can not be predicted from text content information which contains the project descriptions and metadata so as to cause recommendation inaccuracy. The method comprises the following steps that: carrying out modeling on the implicit feedback characteristics of the historical behavior data of a user, and learning to obtain the implicit factor vectors of the user and the project after modeling; taking the implicit factor vector of the project as tag training to carry out modeling on the time sequence information of project text contents and deeply mine a network model; and for a new project which does not appear in the historical behavior data of the user, predicting the network model obtained through S(2) in the text content information of the project to obtain the implicit factor vector of the project, directly matching the implicit factor vector of the project with the implicit factor vector, which is obtained in S(1), of the user, and sorting matching degreesto obtain the new project recommendation list of each user. The method is used for recommending new projects.

The invention provides a method and system for recommending application resources. Included are the steps of collecting previous records of users using applications; conducting statistics on the collected previous records of the users, calculating the percentage of each type of application according to the number of different types of applications used by each user, including users, defined as typical users, into a corresponding user group to a particular type of application whose percentage is greater than a group threshold, and generating corresponding user groups, including the typical users, to the particular type of application; and recommending corresponding application resources to users in each user group according to a particular type of information corresponding to each user group. According to the invention, customized recommendations can be made when a user uses an application, and recommendation results are improved.

The invention discloses a method and a system for recommending a video resource. The method comprises that: historical records of all users in the process of watching videos are collected; statistics is performed on the collected historical records of all the users, the number ratio of various types of videos watched by all the users is calculated according to the number of various types of videos watched by all the users, and user groups which include typical users and are corresponding to specific video types are generated; data characteristics of typical users in all the specific types of user groups on other dimensions are acquired respectively, and group characteristics of all the specific types of users on other dimensions are confirmed; whether the data characteristics of other users apart from the typical users on other dimensions meet the group characteristics of certain specific types of groups is judged, and if judgment result is yes, the users are added into the specific types of user groups; and according to the corresponding specific types of information of all the user groups, the corresponding video resource is recommended to the users in the user groups. With application of the method and the system for recommending the video resource, content recommendation can be specifically performed when the users are watching videos so that recommendation effect is enhanced.

The invention discloses a personalized movie similarity calculation method based on a user interest model. The method includes: according to users' historical behaviors, namely through a personalized movie recommendation system platform, mining users' various behaviors of searching a movie resource library and users' watching and collecting behaviors; fully mining and analyzing different preference degrees of different users, upon six basic attributes including performers, directors, types, regions, times and content introductions of movies, thus acquiring a first-layer six-dimensional spatial vector representation of a user model; according to the users' behaviors above, by means of keyword extraction or semantic analysis, analyzing the different users' weights of characteristic values in the six dimensions, thus acquiring a second-layer six-dimensional spatial vector representation of the user model; using a two-layer multi-dimensional spatial vector to represent the user interest model, generating different movie similarity lists for the different users on the basis of the user interest model and basic content characteristics of movies. Therefore, recommending is more effective.

The invention provides a method and system for recommending friend information in the social network. The method comprises the following steps that firstly, after recommending request information of a user is received, a social network information set is established according to operation behavior information and attribute information of users in the social network; secondly, a friend community set including the user and other K users is found in the social network information set; thirdly, behavior orientation information of the other K users in the friend community set is analyzed and recommended to the user. According to the method and system for recommending the friend information in the social network, the information of obvious friends is analyzed in a recommended result, the preference and influence of hidden friends of the user are also taken into consideration, and compared with an existing scheme, the method and system have the advantage that the produced recommended result is further improved on the aspect of precision.

The invention discloses a personalized position recommendation method based on a knowledge graph. The method includes: acquiring job recruitment field data, acquiring the resume information of a job seeker, and normalizing the job recruitment field data; performing knowledge extraction and fusion to form structured job recruitment field knowledge; saving the acquired structured job recruitment field knowledge into a graph database to build the job recruitment field knowledge graph; building a personalized position recommendation model based on the knowledge graph; reading the resume information of the job seeker, and mapping some attributes in the resume information according to the knowledge graph; according to the knowledge graph, using the position recommendation model to filter positions according to industry categories to form a to-be-recommended position list; quantifying corresponding attributes in the resume information and position information according to the to-be-recommended position list; calculating the similarity of the resume information and the position information, screening top N positions with the highest similarity to the resume of the job seeker to generate arecommendation list, and transmitting the recommendation list to the job seeker.

The present invention relates to a personalized travel route recommendation method based on tourist trust. Firstly, photo information with a geographical label and historical tourist information in a network are gathered, a large number of photo information is subjected to preprocessing, reliable interest point information is obtained, then a body database is constructed by using body-based modeling thought, interest point probability is predicted by using the mixed model of Markov and an subject, the travel route generation algorithm based on interest point heat is generated, and finally combined with a user trust weighted tourist route, a final route is recommended to a user. According to the method, the real travel information of users in a social network is fully utilized, a personalized travel route recommendation service can be effectively provided to the user, and the method has a good reference value for transportation service departments and travel agencies.

The invention provides a method and a system for recommending a video resource. The method comprises that: historical records of all users in the process of watching videos are collected; statistics is performed on the collected historical records of all the users, number ratio or duration ratio of video wartching in each time period of each user is calculated according to time information of the watched videos, and the users are classified into the corresponding user groups of specific time periods whose number ratio or duration ratio is greater than a group threshold so that user groups which include typical users and are corresponding to each specific time period are generated; data characteristics of the typical users in the user groups of all the specific time periods on other dimensions are respectively acquired, and group characteristics of the users of all the specific time periods on other dimensions are confirmed; if the result is yes, the users are added into the user groups of the specific time periods; and according to specific time period information corresponding to all the user groups, the corresponding video resource is recommended to the users in the user groups. With application of the method and the system for recommending the video resource, content recommendation can be specifically performed when the users are watching videos so that recommendation effect is enhanced.

A user recommendation method comprises the following steps of acquiring an instruction for triggering recommended users, generating a candidate recommended user list according to the instruction, reading user social interaction quality data, calculating the matching success rate of users in the candidate recommended user list according to the social interaction quality data, and choosing at least one user with the highest matching success rate from the candidate recommended user list to be recommended. The user recommendation method is used so that the recommendation performance and the recommendation efficiency can be improved. In addition, the invention further provides a user recommendation system.

The invention discloses a commodity recommendation method based on multidimensional user consumption propensity modeling. According to the method, through analyzing multidimensional information such as a user browsing history, a consumption record and a user behavior record, the real consumption propensity of the user is speculated, and the method has a very important actual application value for the improvement of a personalized commodity recommendation effect and the conversion of the effect into an actual purchase behavior. Firstly, a multidimensional user consumption propensity model for the field of electronic commerce can be obtained, and a foundation is laid for the following personalized recommendation. Secondly, combined with a user interest classification and a cyclical consumption dimension propensity analysis, a personalized commodity recommendation effect in accordance with the consumption propensity of a client can be provided to the client. The method result can be widely applied to an electronic commerce recommendation application system.

The invention relates to a label recommendation method fused with the implicit connection relationship of users and oriented on a question and answer platform, and can solve the technical problem thatuser requirements cannot be satisfied due to the fact that the recommendation result of the traditional label recommendation method is limited. The label recommendation method comprises the steps of:constructing a question set, a label set corresponding to questions, and a user set; constructing the network of a user; processing to obtain characteristic vectors of the user; then, obtaining characteristic vectors of the questions; splicing the characteristic vectors of the user with the characteristic vectors of the questions, so that a multi-characteristic vector set fused with implicit connection of users is obtained through a layer of fully-connected network; mapping the multi-characteristic vector set into the probability by using a logistic regression model, performing descending sorting of labels based on the output probability value, and taking previous top labels; training an initial model, and obtaining the final recommendation model after training is ended; and, when the user submits the questions to a website, recommending the previous top labels to the user in the recommendation model. By means of the label recommendation method disclosed by the invention, the varietyand the accuracy of the label system can be improved; and user requirements can be satisfied.

The invention relates to a class case recommendation method based on text content. The method is divided into a pre-training part and a fine adjustment part. The pre-training part adopts a transformerencoder as a main structure, a Chinese language model is trained, Chinese language knowledge is learned from other corpora, and a high-quality language model is obtained. A triad model is used as a framework of the fine adjustment part, a preprocessed judicial document is used as training data, more knowledge about judgment is learned from the judicial field, and a better text vector representation is obtained. Compared with a traditional keyword-based class case recommendation method and a single-task neural network-based class case recommendation method, the content-based class case recommendation method provided by the invention is better in effect, and has better robustness based on a semantic training model, which indicates that the method provided by the invention is effective and practical.

A recommendation method and a recommendation apparatus based on deep reinforcement learning, and a non-transitory computer-readable recording medium are provided. In the method, entity semantic information representation vectors of products are generated based on a product knowledge graph; browsing context information representation vectors of the products are generated based on historical browsing behavior of a user with respect to products; the entity semantic information representation vectors and the browsing context information representation vectors of the respective products are merged to obtain vectors of the products; a recommendation model based on deep reinforcement learning is constructed, and the recommendation model based on the deep reinforcement learning is offline-trained using historical behavior data of the user to obtain the offline-trained recommendation model, the products in the historical behavior data of the user are represented by the vectors of the products; and products are online-recommended using the offline-trained recommendation model.

The invention discloses a collaborative filtering recommendation algorithm based on a graph convolution attention mechanism. The method comprises the steps of firstly collecting and processing data and dividing a data set, secondly constructing a GACF model, and finally training the model and recommending by predicting an association score between a user and an item. According to the graph convolution attention mechanism collaborative filtering model provided by the invention, firstly, user-project interaction information is mapped to a vector space by using a graph embedding technology, then, the embedding expression of a user-project interaction graph is learned through a graph convolution network, and then, different weights are allocated to neighbor nodes by using an attention mechanism. By aggregating the feature information of the neighbor nodes, the weight between the neighbor nodes can only depend on the feature expression between the nodes, so that the generalization ability of the model is improved, and finally, a plurality of embedded vectors learned by a graph convolution layer are weighted and aggregated to obtain the association score between the user and the project.

The invention provides a personalized recommendation method and system for friends and applications. The method comprises the steps that A, a recommendation score S is calculated; B, recommended content is displayed according to the recommendation score; C, a recommendation weight vector is adjusted according to feedback information about the recommended content from a user; D, the recommended content is updated according to the adjusted user recommendation weight vector. The personalized recommendation method and system have the advantages that the application range is wider, recommendation results are more accurate, the calculating speed is higher, calculation at idle time is achieved, a blacklist mechanism is adopted, the accuracy of information recommendation can be effectively improved, calculation cost is saved, and a good recommendation effect is achieved.

The invention discloses a socialization filtering method based on a PageRank algorithm, mainly solving the problem that a filtering method has low accuracy under the conditions that group members are numerous and social relationship is complex in the prior art. The socialization filtering method disclosed by the invention is realized by the following steps: acquiring friend relationship between a group and the group members from a webpage configuration file, creating a personal preference model of each group member; by adopting the PageBank algorithm, iteratively calculating the influence of the group members to the group, so as to obtain a preference model of the whole group; carrying out object recommendation by utilizing the model, namely initiatively providing recommended object data for a user; selecting interested and required information from the information by the user. The socialization filtering method disclosed by the invention has the advantages that the preference model of the group is analyzed, and recommendation of different field objects on a network can be realized only by modifying a keyword vector in the field of the model.

The invention relates to a multimedia data recommendation method, device and equipment and a computer readable storage medium, and belongs to the technical field of computers. The method comprises thesteps of determining a first score of target multimedia data according to attribute information and historical operation records of a target user identifier and attribute information of the target multimedia data; Determining an adjustment weight according to the negative operation frequency of the target multimedia data; According to the adjustment weight, adjusting the first score to obtain a second score of the target multimedia data; And recommending the multimedia data to the terminal corresponding to the target user identifier based on the second score. Wherein the second score synthesizes the number of times of preset negative operation of the target multimedia data; According to the method, the matching degree of the target user and the target multimedia data can be embodied, thequality of the target multimedia data can be embodied, the probability that the target user likes the target multimedia data can be accurately represented, the accuracy of scores is improved, the recommendation effect is improved, and interference to the target user is avoided.

The invention belongs to the technical field of e-commerce information processing, and discloses a cross-domain recommendation data processing method with multiple auxiliary domains and a cross-domainrecommendation data processing system. The method comprises obtaining a scoring matrix of an auxiliary domain, calculating the scoring reliability of a user, carrying out equal-proportion segmented mapping on a threshold value, and emptying scores with the number of scores lower than the threshold value in the auxiliary domain; obtaining clustering level scoring matrix of all domains by using a K-means clustering algorithm, and carrying out matrix decomposition; meanwhile, decomposing the target domain scoring matrix to learn a feature mapping function for the cold start user; evaluating predicted scoring matrix by using an average absolute error. Compared with the prior art, the method has the advantages that K-means clustering algorithm is used in the data processing process for obtaining clustering-level user project scoring matrix combined with all domains, the data sparsity of the cold start user is reduced. The problem that the recommendation effect is not ideal due to poor prediction accuracy of a traditional single auxiliary domain matrix factorization model is solved, the recommendation effect of the recommendation system is improved, and the method has higher universality.

The invention discloses an interest recommendation method and system based on a user sequence click behavior, and the method comprises the steps: firstly obtaining the historical interaction project data of a user, and forming a historical interaction project sequence of the user; constructing an interest recommendation model; carrying out session division on the historical interaction item sequence of the user by utilizing an interest recommendation model; extracting interests in each session obtained after division, and performing weighting processing on the interests in each session to obtain a session interest sequence of the user; interacting interests among different sessions to obtain a dynamic interaction model among different sessions; and inputting the session interest sequence of the user into a dynamic interaction model between different sessions, and predicting and obtaining a to-be-recommended target project sequence.