40results about How to "Realize accurate recommendation" patented technology

The embodiment of the disclosure provides a route planning method based on traffic rules. The route planning method comprises the steps: acquiring a target road network map; generating a road networktopological graph including a plurality of nodes expressing road segments according to the target road network map; acquiring preset traffic rule information corresponding to each node; judging whether every two nodes in the road network topological graph pass or not according to the preset traffic rule information; updating the road network topological graph to generate a target road network topological graph according to a judgment result that whether every two nodes pass or not; and performing route planning based on the target road network topological graph. Correspondingly, the embodimentof the invention further provides a route planning device based on the traffic rules, computer equipment and a computer readable storage medium. According to the embodiment of the invention, the traffic rules are applied to route planning, so that accurate route recommendation is realized; the route with relatively low passing cost can be recommended to the user; and the data processing efficiency can be effectively improved.

The invention discloses an information recommending method, comprising: obtaining relevant information of users wherein the relevant information comprises the attribute information and / or purchase information; quantifying the attribute information and / or purchase information; based on the quantified information, calculating the similarity among users; according to the similarity, conducting classifications; marking the users in the same classification; and recommending commodity information to users marked in the same classification. The invention also discloses an information recommending apparatus. With the invention, it is possible to mark users and to accurately recommend commodity information to the users. It is also possible to realize customized recommendations.

The invention provides a user tag extension treatment method. The method includes the steps of obtaining a user tag and candidate expanded tags, conducting mutual information calculation between the user tag and the candidate expanded tags to obtain the correlation of the candidate expanded tags relative to the user tag, screening out extension tags in the candidate expanded tags according to the correlation to obtain the extension tags correlated to the user tag, wherein an extension relation between the user tag and the correlated extension legs is formed. In addition, the invention further provides a user tag extension treatment device matched with the method and a text recommendation method and device. By means of the user tag extension treatment method and device, the user tag extension precision can be improved; by means of the text recommendation method and device, the text precise recommendation can be realized on the basis of user tag precise extension.

The invention provides an archive research and judgment method and system based on a knowledge graph technology. The archive research and judgment method comprises the following steps: uploading an archive; file entity extraction: extracting personal names, place names, organization names, colleges and universities, work, events, buildings and data entities; archive relationship extraction: extracting an association relationship between two entities, and constructing an edge of the archive knowledge graph; archive knowledge fusion: fusing the extracted archive entities from different sources to generate a unique archive entity with comprehensive attributes; archive knowledge storage: storing the extracted archive entities and relational data, wherein the data can be stored in a graph database; archive research and judgment analysis: carrying out archive intelligent retrieval. Under the support of the knowledge graph and the machine learning technology, the created archive knowledge graph is fully utilized, associated recommendation of various related archive entities and accurate recommendation of related archives are realized, archive research and judgment personnel are helped to expand reading, and the depth and breadth of archive research and judgment are expanded.

The invention discloses a product recommendation method and system combining paired optimization and matrix decomposition. The method comprises the steps of obtaining a scoring matrix of a user to a product and a social relation matrix between the users; clustering the products according to the scoring matrix of the users to the products, and dividing the products into a plurality of clusters; mapping the scoring matrix of the user to the product and the social relation matrix between the users into a weighted adjacency matrix; defining a user-product bipartite graph, and filling the weightedadjacent matrix on the user-product bipartite graph by using a random walk algorithm to obtain a probability matrix; establishing an objective function based on the paired optimization pairwise basedon the divided clusters of the products; and performing matrix decomposition on the probability matrix based on the objective function, and obtaining a decomposed matrix and a product recommendation list by taking the minimum objective function value as an objective.

The invention discloses an IoT edge equipment trust evaluation method. A time window method is employed to obtain a predicted trust value in a current cycle according to trust values of equipment in past time cycles; trust has time correlation, and thus the predicted trust value delineates a possible range of a current trust value to a certain extent; a difference value of a received actual trustvalue and the predicted trust value is compared with an error threshold, and whether the trust value of interaction can be trusted is judged; if not, the trust value is corrected through implementinga penalty system, then aggregation calculation of feedback trust is carried out after correction, and a more reliable overall trust value can be obtained; and malicious evaluation caused by maliciousbehaviors such as false feedback, malicious attacks and collusion cheating is avoided, and reliability of overall trust and security of a system are improved. The invention also discloses an IoT edgeequipment trust evaluation device, an edge proxy server and an IoT edge equipment trust evaluation system, which have the above advantages.

The embodiment of the invention discloses a service object recommendation method and device and electronic equipment. The service object recommendation method includes: acquiring anchor attribute information of a live-streaming room and all pieces of service object attribute information; determining a target service object on the basis of the anchor attribute information and all the pieces of service object attribute information; and displaying the target service object in the live-streaming room. Through the embodiment of the invention, the service object matching a video anchor can be selected from a service object library for the video anchor, the service object better meets needs of the video anchor, and precise recommendation of the service object is realized.

The invention discloses a personalized recommendation system and method based on a block chain, and belongs to personalized recommendation based on a block chain. In order to solve the technical problem of how to realize accurate recommendation of information on the premise of improving the security of user behavior data and reducing the data maintenance cost, the adopted technical scheme is as follows: the system comprises a block chain platform deployment module, a user behavior data uploading module, a user clustering module and a personalized recommendation module. The method specificallycomprises the following steps: deploying a block chain, wherein users of an e-commerce platform are noted as an organization of the block chain platform, and each block of the block chain records behavior data of the user and encrypts the behavior data of the user; writing the user behavior data into a block chain; clustering the users; and performing personalized recommendation, wherein feature indexes of the user cluster are extracted, and the advertorial and the commodity are accurately recommended to the user according to the feature indexes of the user cluster.

The invention relates to a keyword extraction method and device and a storage medium, and the method comprises the steps: receiving an original document, extracting a plurality of candidate phrases from the original document, and enabling the extracted candidate phrases to form a candidate set; obtaining the association degree between each candidate phrase in the candidate set and the original document; obtaining divergence of each candidate phrase in the candidate set; and selecting at least one candidate phrase from the candidate set as a key phrase based on the correlation degree and the divergence degree, and forming a key phrase set of the original document based on the selected at least one key phrase. Thus, the candidate phrases with the high association degree with the original document can be selected from the candidate set to serve as the key phrases, the similarity between the extracted key phrases and the original document is high, and the accuracy of extracting the key phrases is improved; the phrases with high divergence can be selected from the candidate set to serve as the key phrases, so that the difference degree between the extracted key phrases and the key phrases selected to the key phrase set is high, and the diversity of the key phrases is improved.

The invention discloses a judgment document similar dispute focus merging method and a system. The method comprises the steps of analyzing a judgment document structure; obtaining a dispute focus to be processed; classifying dispute focuses; classifying and storing dispute focuses to be processed; performing word segmentation on a dispute focus to be processed; extracting text features of a dispute focus, and converting the text of the dispute focus after word segmentation into a text feature vector with semantic information; clustering the obtained text feature vectors; obtaining a merging result of the dispute focuses to be processed based on the clustering result; according to the method and the system, judicial trial practice is taken as a foothold, and association between cases is established by combining similar or same dispute focuses, so that similar case retrieval or recommendation of dispute focus dimensions is realized.

Embodiments of the present application provide a method, apparatus, and electronic device for predicting financial behavior based on handwriting key points. The method comprises the following steps: identifying the text in the handwriting picture of the customer and the key points used for determining the structure of the text, and extracting the position information of the key points in the handwriting picture; determining handwriting characteristics used to characterize handwriting habits according to the position information of key points; according to the preset handwriting characteristicsand the corresponding relationship between financial behavior and handwriting characteristics to determine the customer 's financial behavior prediction information. Embodiments of the present application identify key points of customer signatures by key point identification techniques, by analyzing the position information of the key points, the handwriting characteristics that characterize theuser's writing habits are obtained. Then, according to the handwriting characteristics that can reflect the customer's financial behavior, the possible financial behavior of the customer is predicted,and the appropriate financial products are recommended to the users, so as to realize the accurate recommendation of the financial products, which helps to improve the transaction rate and reduce themarketing cost at the same time.

The invention discloses a method for multimedia recommendation on the basis of data grading. The method comprises the following steps of receiving a selection instruction of a user on the category, and obtaining index data of multimedia in the selected category; classifying the index data of the multimedia according to sites, and performing statistics to obtain the index data of the multimedia of each site; obtaining the grading of the multimedia in each site through calculation according to the index data of the multimedia of each site; performing unified sequencing on the multimedia of different sites according to the grading result in accordance with the grading values, and recommending the multimedia to the user according to the sequencing. The invention also discloses a system for multimedia recommendation on the basis of data grading. The method and the system for multimedia recommendation on the basis of data grading have the advantages that the grading of the multimedia is obtained through calculation through the index data of the multimedia in one site, so that the multimedia of different sites can be sequenced according to the grading; the precise recommendation of the multimedia is finally realized; further, the experience degree of the user is improved.

The invention discloses a film and television resource personalized recommendation method in a social network environment, and belongs to the technical field of data analysis and pushing. The method includes the following steps of S1, online comment obtaining and preprocessing, S2, online comment emotion value calculation, S3, film watching decision criterion and weight determination, and S4, film and television resource sorting: a comprehensive foreground value of a film xi is obtained by combining a probability language decision matrix, a value function and a weight function, wherein the larger the comprehensive foreground value is, the more worthy of recommendation is achieved, and the higher the sorting is. According to the method, the influence of irrational factors such as psychological behaviors of film viewers on decision making of the film viewers is fully considered while objective factors are considered, so that film and television resource recommendation is more practical and accurate.

The invention discloses a recommended item determination method and device, electronic equipment and a storage medium and relates to the technical field of computers. A specific embodiment of the method comprises the following steps of acquiring article information of articles browsed by each user in a preset time period, wherein the article information comprises attributes of the browsed articles and a browsing sequence corresponding to the browsed articles; according to the browsing sequence corresponding to the browsed items, the feature vector of each browsed item being calculated, and the feature vector representing the incidence relation between the browsed items; calculating similarity between each browsed article and the target article according to the feature vector of each browsed article; and according to the similarity and the attribute of the browsed article, determining a recommended article corresponding to the target article. According to the method, problems that accuracy of a mode of determining recommended articles based on categories is low, and accurate recommendation of the articles cannot be realized can be solved.

The invention provides a knowledge graph embedding-based learning resource recommendation method and system. The method comprises the steps of determining a learning resource knowledge graph; constructing learner features based on social attributes, knowledge levels, learning styles and learning concentration degrees of the learner; combining the learner features with the learning resource knowledge graph to obtain a multi-modal knowledge graph; constructing a learning resource recommendation model into which the knowledge graph is embedded, inputting the multi-modal knowledge graph, the resources preferred by the learner and the target learning resource into the learning resource recommendation model, extracting a path between the resources preferred by the learner and the target learning resource, scoring the target learning resource, judging whether the target learning resource meets the requirements of the learner or not, and recommending the learning resource meeting the requirements of the learner to the learner based on a scoring result. According to the method, the influence factors of the learning state of the learner and the incidence relation between the resources are comprehensively considered for learning resource recommendation, and the actual requirements of the learner can be met.

The invention discloses an activity information processing method, device and equipment, relates to the technical field of information processing, and aims to accurately control activity resources without configuring activity information by a service party and improve the recommendation effect of the activity information. The method comprises the steps: obtaining first activity data generated whena service party delivers activity information; based on the first activity data, selecting an activity information template suitable for being released by a service party for pushing; acquiring second activity data generated by updating the activity information template by the service party; determining a verification result of an activity information template based on the first activity data andthe second activity data; and sending a verification result of the activity information template.

The invention relates to a data recommendation method and device, a server and a storage medium, and belongs to the technical field of computers. The method comprises the steps of obtaining target data features corresponding to to-be-recommended target multimedia data, calling a feature conversion model, performing feature transformation processing on the target data features to obtain transformed predicted operation features, determining the matching degree of each user and the target multimedia data according to user features corresponding to multiple users and the predicted operation features, and recommending the target multimedia data according to the matching degree of each user and the target multimedia data. And selecting at least one target user from the plurality of users according to the matching degrees corresponding to the plurality of users, and recommending the target multimedia data to the at least one target user. According to the method, the predicted operation characteristics of the multimedia data are used for representing the operation possibly executed by the user on the multimedia data, the user matched with the multimedia data is determined, and the multimedia data are subsequently recommended to the matched user, so that accurate recommendation of the multimedia data is realized, the accuracy of data recommendation is ensured, and cold start of the multimedia data is realized.

The invention discloses a medical system labor dispatch management system and a working method thereof. The medical system labor dispatch management system comprises a labor supplier module, a labor demand entry module, a management settlement module and an offer invitation module. The offer invitation module comprises a selection pushing assembly, a confirmation feedback assembly, a contract pushing assembly, a user side and a worker offer side. The contract pushing assembly comprises a user interface and an offer interface, the user interface is connected with the user side, the offer interface is connected with the worker offer side. The user interface and the offer interface are respectively provided with a corresponding Tag marking assembly, and each Tag marking assembly comprises a timing gear returning function. The system has the advantages that when labor purchasing is carried out on line and an employment relationship of labor dispatching is reached, temporary employment relationship marking can be carried out on the corresponding labor supplier and the service purchaser through the Tag marking assembly, so that the labor supplier can be conveniently managed by a worker, and salaries and persons in charge can be conveniently issued.

The present invention provides an extension processing method for a user tag, comprising: acquiring a user tag and a candidate extended tag; calculating the mutual information between the user tag and the candidate extended tag to obtain the relative relationship between the candidate extended tag and the user tag. The degree of association; according to the degree of association, the extension tags in the candidate extension tags are filtered to obtain the extension tags associated with the user tags, and the user tags and the associated extension tags constitute an extension relationship. In addition, an extension processing device for user tags matched with the method, a text recommendation method and device are also provided. The above-mentioned user tag expansion processing method and device can improve the accuracy of user tag expansion, and the above text recommendation method and device can implement accurate text recommendation on the basis of realizing precise user tag expansion.

The present disclosure relates to a method and device for recommending an exercise program. The method includes: acquiring massage time information of each body part massaged by a target user using a massage stick; and recommending to the target user the massage time information corresponding to the first body part when the massage time information corresponding to the first body part satisfies the first condition exercise program. The present disclosure obtains the massage time information of each body part massaged by the target user with a massage stick, and recommends an exercise plan for the first body part to the target user when the massage time information corresponding to the first body part satisfies the first condition. In this way, the exercise plan can be recommended to the target user in a targeted manner according to the massage operation of the target user, and the accurate recommendation of the exercise plan can be realized.

The embodiment of the present application belongs to the field of artificial intelligence, and relates to an intelligent conference management method, which includes, when a conference establishment request is received, parsing the conference establishment request to obtain a conference label; obtaining a preset target neural network, and processing the conference label according to the target neural network Predict and sort to get the recommended meeting process; input the recommended meeting process into the preset knowledge map, and determine the target meeting feature information and target user information from the preset knowledge map according to the traversal algorithm; calculate the target meeting feature information and target user information through the preset collaborative filtering algorithm Target user information to obtain a display template; obtain target text information, fill the target text information into the display template, and obtain target display data. The application also provides an intelligent conference management device, computer equipment and storage media. In addition, this application also relates to block chain technology, and the target display data can be stored in the block chain. This application realizes the intelligent recommendation and data display of the conference process.

The invention relates to an article recommendation method and device, a meta-network processing method and device, a storage medium and equipment. The article recommendation method comprises the stepsof obtaining attribute information of an article and an object identifier of an article interaction object; performing feature extraction on the attribute information through a meta-stretching network to obtain a stretching vector; performing feature extraction on the object identifier through a meta-offset network to obtain an offset vector; performing feature intersection on the vectorized article identifier of the article and the stretching vector, and fusing the obtained intersection feature vector with the offset vector to obtain a fused feature vector; determining a recommendation probability based on the fusion feature vector, the vectorized attribute information and vectorized object information of the target object; and recommending the article to the target object when the recommendation probability meets a recommendation condition. By adopting the method, the accuracy of article recommendation can be improved, so that the effect of accurate recommendation is achieved.

The present disclosure discloses a product recommendation method and system combining pairwise optimization and matrix decomposition to obtain user rating matrices for products and social relationship matrices between users; cluster products according to user rating matrices for products, and divide products into into several clusters; map the user-product rating matrix and the social relationship matrix between users into a weighted adjacency matrix; define a user-product bipartite graph, and use a random walk algorithm on the user-product bipartite graph to weighted adjacency The matrix is ​​filled to obtain the probability matrix; based on the clusters where the products are divided, an objective function based on pairwise optimization is established; the probability matrix is ​​decomposed based on the objective function, and the objective function value is minimized to obtain the decomposed matrix and products Recommended list.

The invention relates to the field of big data and wisdom education, and discloses a courseware management system based on big data, and the system comprises an education cloud platform, a first education terminal and a second education terminal. The education cloud platform comprises a database, a courseware recommendation module, a courseware analysis module and a courseware pushing module. Andafter courseware request data sent by the first education terminal is received, the courseware recommendation module of the education cloud platform generates a first data stream according to the courseware request data. And the courseware recommendation module analyzes and obtains the courseware adaptation degree of an educational consumer according to the first data stream and the educational consumer information. And when the courseware adaptation degree is greater than the courseware learning degree, the courseware pushing module processes the courseware request data and the educational consumer information to obtain a second data stream, and sends the second data stream to a corresponding second education terminal. The courseware analysis module acquires courseware learning information of the educational consumer to obtain a courseware learning dictionary, and analyzes the courseware learning dictionary to obtain courseware analysis data.

The present invention provides a recommendation method, device, electronic device and storage medium based on action pruning. The status and rating prediction model corresponding to the recommended content predicts the rating of each content to be recommended, and recommends to target users based on the rating of each content to be recommended; the rating prediction model is obtained by reinforcement learning. In the process of reinforcement learning, the rating prediction The model obtains the regret value of each candidate score in the current sample state from the regret value set, and predicts the score based on the candidate score whose regret value is greater than the preset threshold. The regret value set stores the historical state and its corresponding regret value. The regret value is based on The advantages of each candidate score in the historical state are determined. The historical state is the sample state before the current sample state, which speeds up the convergence speed of reinforcement learning and realizes personalized and accurate recommendations for users.

The invention discloses a fresh agricultural product online recommendation method considering customer consumption behaviors and preferences. The method comprises the following steps of (1) data analysis and preprocessing; (2) constructing a fresh agricultural product utility function based on latent semantics and behavior sequence utility; (3) constructing a fresh agricultural product selection model for traversing the purchase cycle; (4) solving the constructed fresh agricultural product selection model by adopting a gradient descent method; and (5) evaluating the prediction effect by usingthe recall rate and the precision. According to the method, seasonal characteristics of fresh agricultural products and interest preference characteristic factors of customers for the agricultural products can be effectively reflected, customer selection behaviors and user product characteristics are fully considered, and experiments prove that the method has a better prediction effect compared with other methods.

A method for personalized recommendation of film and television resources in a social network environment, belonging to the technical field of data analysis and push, comprising the following steps: S1: online comment acquisition and preprocessing, S2: online comment sentiment value calculation, and S3: movie viewing decision criteria and weights OK, S4: Sorting of film and television resources: Combining the probability language decision matrix, value function, and weight function to obtain movie x i The comprehensive prospect value of , the larger the comprehensive prospect value, the more worthy of recommendation, and the higher the ranking. The present invention fully considers the influence of irrational factors such as the viewer's psychological behavior on the viewer's decision-making while considering various objective factors, so that the recommendation of film and television resources is more realistic and accurate.