540results about How to "Improve recommendation efficiency" patented technology

The invention discloses a content recommendation method and device based on user favorites. The method mainly includes the steps that feature word extraction is performed according to collected behavior data of all users, and extracted feature words are respectively matched with preset categories and / or entity word banks to construct a favorite topic library of the categories and / or entity words corresponding to the behavior data of the users; according to multi-dimension attribute information corresponding to the behavior data of the users, the categories and / or the entity words in the favorite topic library of the users are sequenced; before recommended content are displayed for the users, a preset recommended content library is searched for, and sequenced recommended content which are matched with the sequenced categories and / or the entity words in the favorite topic sequencing library are extracted in advance; when the users give access requests, current access network scenes of the users are extracted, and the sequenced recommended content matched with the current access scenes of the users are extracted to be displayed. According to the technical scheme, recommending efficiency can be further improved.

The invention provides a method and a device for automatically recommending an application. The method comprises the steps of: receiving an application acquisition request submitted from a client by a user, wherein the application acquisition request comprises a user identification; according to the user identification, extracting the existing user behavior information of corresponding user from a user feature library, wherein the user behavior information comprises operation information that a user aims at previously recommended application; according to the user behavior information, determining the type of the application recommended to the user; in an application data set of the type of the application, according to the operation information that the user aims at the previously recommended application, extracting an matched application; and according to the type of the application, generating a corresponding application folder, and placing the matched application in the corresponding application folder for recommending. According to the invention, the individual needs of users can be satisfied, the recommending efficiency is improved and the coverage rate is increased.

Embodiments of the invention provide a personalized video recommendation method and apparatus. The method comprises the steps of obtaining a user portrait characteristic corresponding to a user identifier; according to the user portrait characteristic, obtaining a video characteristic model corresponding to the user portrait characteristic; and according to the video characteristic model corresponding to the user portrait characteristic, obtaining a video matched with the video characteristic model corresponding to the user portrait characteristic from a video database and performing recommendation. According to the method and apparatus, a user can quickly and accurately obtain the video interested by the user, so that the video acquisition efficiency is improved, the personalized video demand of the user is met, and the experience of user video search and recommendation is greatly enhanced.

The invention provides a news recommendation system and method based on an FOLFM model. Based on a content recommendation method, a news content model is expressed abstractly through a latent class model and content characteristics, and an individual latent class preference model is built for each user. Real-time training is carried out on a real-time behavior record of a user to obtain preference, on certain latent class news, of the user, whether the news is recommended to the user is determined through calculation, and a final news recommendation list is obtained after a series of processing processes. The news recommendation system and method based on the FOLFM model deeply excavate the interest of the user, improve recommendation accuracy and satisfaction of the user, avoid a cold starting problem of the news, and guarantee performance under the condition that the recommendation effect is improved as much as possible. The experiment shows that the news recommendation system and method based on the FOLFM model not only guarantee the requirements for high accuracy and high speed, but also realize visual real-time recommendation for the user.

The invention discloses item recommendation method and system based on a user-item bipartite model. The method includes: extracting acquired user behavior information to establish a user-item bipartite model; establishing an item-to-user inverted list model on the basis of the user-item bipartite model, calculating item weights, and establishing a user-item weight list; calculating weights of item pairs corresponding to users and common user weights of the item pairs according to the established user-item weight, and establishing an item pair-common user weight list; operating to obtain an inter-item similarity list according to a preset similarity algorithm; querying the established user-item bipartite model to obtain recommended user-mapped items, querying an inter-item similarity list according to the user-mapped items to be recommended to obtain query result, and generating a recommendation list according to the query result. By the use of the method and system, calculation needed by recommendation can be reduced and recommendation efficiency can be improved.

The invention relates to a code recommendation method based on a long short-term memory (LSTM) network. For the problem that low recommendation accuracy rates, low recommendation efficiency and the like are ubiquitous in existing code recommendation technologies, the method firstly extracts source code to form an API sequence, utilizing the long short-term memory network to build a code recommendation model to learn relationships between API calls, and then carries out code recommendation. A dropout technology is used to prevent model overfitting. At the same time, using a ReLu function to instead a traditional saturation function is provided, the gradient vanishing problem is solved, a model convergence speed is accelerated, model performance is improved, and advantages of the neural network are fully exerted. The technical scheme of the invention has the characteristics of simpleness and quickness, and can better improve an accuracy rate and recommendation efficiency of code recommendation.

The present invention is applicable to the technical field of computers, and provides a recommended method, apparatus and device, and a storage medium. The method comprises: obtaining history score data of a user, a to-be-scored item, and text content of the to-be-scored item; according to the history score data of the user, the to-be-scored item, and the text content of the to-be-scored item, training the preset stack noise reduction self-encoder and the preset probability matrix decomposition model to obtain the user characteristic matrix, the item characteristic matrix, the corresponding item latent characteristics and user latent characteristics; according to the user characteristic matrix, the item characteristic matrix, the corresponding item latent characteristics and the user latent characteristics, calculating the predicted score of the to-be-scored item; and according to the predicted score, generating a recommendation list, and outputting the recommendation list to the user, so that when recommending the item to the user, the item characteristics are combined with the user characteristics, the recommended accuracy is effectively improved, and the recommended efficiency of the item is further improved.

The invention discloses a content recommendation method and device and equipment. The embodiment of the method includes the steps that feature information, corresponding to multiple preset feature tags, of a user and to-be-recommended content are acquired; based on the feature information and recommendation weights, determined with the analytic hierarchy process, of the preset feature tags, recommendation indexes of the to-be-recommended content are calculated; target recommended content which is determined from the to-be-recommended content according to the recommendation indexes is recommended to the user, wherein the recommendation weights are determined in the way that the preset feature tags are clustered to determine class tags of the preset feature tags; a hierarchical structure model comprising a criterion layer, a sub-criterion layer and a target layer is constructed, wherein elements of the sub-criterion layer are the preset feature tags, and elements of the criterion layer are the class tags; based on the hierarchical structure model, weights, corresponding to the target layer, of the elements of the sub-criterion layer are determined with the analytic hierarchy process and serve as recommendation weights of the preset feature tags. By means of the embodiment, the recommendation efficiency can be improved.

The invention provides a face identification-based personalized video recommendation method and recommendation system. The method comprises the steps of acquiring a face image of a user and extracting and storing the face features in the face image of the user; acquiring network video data and performing feature vector extracting on the network video data and then generating video feature vectors; generating a user interest model according to the video feature vectors and user behavior data corresponding to the face features; calculating the similarity degrees of the user interest model and the video feature vectors, and forming a recommendation list based on the calculated similarity degrees. The method and the system can analyze the interest and preference of a user after identifying the face of the user, can filter video with low similarity degrees and video that has been viewed already in to-be-recommended video, are high in recommendation efficiency and provide convenience for users.

The invention discloses an application recommendation method and device and server equipment. The method comprises the steps that according to a behavior data list of application acquisition terminal equipment recently browsed and downloaded by terminal equipment, the list comprises a plurality of first application identifications and nearest behavior time of first applications; according to the degree of association between the first applications and second applications in a preset application bank, multiple corresponding second applications are obtained from the preset application bank to form an alternative application assembly; according to current time, the nearest behavior time of the first applications is unified to obtain initial weighted values of the first applications; according to the degree of association of the second applications, the first application identifications corresponding to the degree of the associations and the initial weighted values of the first applications, an application recommendation list is generated, the application recommendation list is sent to the terminal equipment to allow the terminal equipment to show the application recommendation list. According to the technical scheme, individualized recommendation can be achieved, and recommendation efficiency of the applications is effectively improved.

The invention provides a rating prediction based project recommending method for time-sequence control and a system thereof. The corresponding project recommending method comprises the following steps: inputting a project to be recommended; determining a time-sequence rating model related to the project, wherein the time-sequence rating model is used for predicting rating changes of the project along with time; applying one or more recommending policies to the determined time-sequence rating model to determine a preference recommending time of the project; and recommending the project for a user at the determined preference recommending time. In different embodiments, the time-sequence rating model of the project can be selected from a pre-saved time-sequence rating model set, or is automatically generated according to historic data of the system. Furthermore, the selected time-sequence rating model can be adjusted according to user preferred information or feedback information. The project recommending system considers that the interest of users in recommended projects changes along with time so as to increase the recommending efficiency and improve user experience.

The invention belongs to the load forecasting field of a power distribution system, relating to an auto recommending method of urban power load forecasting module based on associative rules. The method comprises the steps of: establishing a historical data base; carrying out data analysis and generalization; mining the associative rules; matching the rules; and obtaining model recommending conclusion by circulating the steps. The method not only can forecast the using condition of a model in an area to be measured, but also can conclude application rules of some models; by utilizing an inference method based on cases, the efficiency of model recommending is improved; and simultaneously the load forecasting efficiency is improved by combining certain expertise.

The invention discloses a direct broadcasting room recommending method and system based on broadcaster styles, and relates to the network technical field; the method comprises the following steps: collecting characteristic parameters and user data of direct broadcasting rooms from a server in a set time period; using characteristic parameters of each direct broadcasting room as characteristic constants to build a characteristic vector of the direct broadcasting room; selecting two random direct broadcasting rooms with broadcasters of different personal information, calculating similarity between characteristic vectors of the two direct broadcasting rooms, and determining direct broadcasting rooms with similarities; recommending other direct broadcasting rooms similar to the direct broadcasting room to all users in the direct broadcasting room according to user data of each direct broadcasting room; calculating a characteristic vector evaluate index according to the visiting rate and / or return visiting rate of the recommended direct broadcasting room, using the evaluate index to screen the characteristic vector characteristic constant, and using the screened characteristic vector to determine similar direct broadcasting rooms. The method and system can precisely recommend direct broadcasting rooms with similar styles to users, thus improving recommending efficiency, and improving user experiences.

The invention discloses a network community based collaborative filtering recommendation method which mainly solves the problem that a recommendation accuracy rate is low in the prior art due to sparsity in acquisition of similarity data of users. The network community based collaborative filtering recommendation method includes: acquiring rating information for recommended items from users and generating a user relation network among the users indirectly through the rating information for the recommended items from the users; computing similarity among the users; partitioning the user relation network into multiple user communities via similarity-based community detection; selecting k users, with the largest similarity, from a local community of the user to form a neighbor user set, and predictively rating items not rated by target users according to the neighbor user set; recommending the item rated with the highest prediction value to the user. From a result of a simulation experiment, the network community based collaborative filtering recommendation method can obtain a better recommendation result as compared with a conventional collaborative filtering recommendation method, and can be used for recommending the items the users interested in to the users.

The invention discloses an information push method and device and belongs to the technical field of internet. The method comprises obtaining historical behavior data of users towards current books, wherein the historical behavior data comprises book purchasing data, book searching data and online book reading data of the users; calculating popularity scores of the current books and preferences of the users towards preset labels according to the user historical behavior data respectively; obtaining books to be recommended to the users according to the current book labels, popularity scores and the preferences of the users towards preset labels. According to the method and the device, personalized recommendation is performed through analysis of the user historical behavior data, manual editing is not required, and the web portal recommendation efficiency is improved.

The invention relates to a commodity recommendation method and device, and belongs to the field of the network technology. The method comprises the following steps of on the basis of a plurality of first user amounts, determining a commodity recommendation list, wherein each first user amount is the amount of users who purchase first-category commodities and execute an appointed behavior for each second-category commodity in a plurality of second-category commodities in a preset time period, and the commodity recommendation list comprises N second-category marks; for each second category in N second categories, on the basis of the mark of the second category, determining the marks of the plurality of commodities which belong to the second category; on the basis of target user characteristic information and the commodity characteristic information of the plurality of commodities which belong to the second category, determining a plurality of recommended purchase probabilities through an appointed logistic regression model; and on the basis of the plurality of recommended purchase probabilities, recommending a target commodity in the plurality of commodities which belong to the second category to the target user. Therefore, different commodities are recommended to different users in a targeted way, and commodity recommendation efficiency is improved.

The invention discloses a method for recommending documents based on tags and a document recommending device. The method comprises the following steps of: receiving a document tag selected by a user, searching a previously set synonym word collection and obtaining a synonym group corresponding to the document tag; according to the obtained synonym group, searching and obtaining a document list containing the synonym group from a previously stored document library; calculating the characteristic value of each document in the document list containing the synonym group, and constructing a space weight vector based on the document tag; and according to the constructed space weight vector based on the document tag, calculating the similarity of documents corresponding to the document tag with each document in the document list, selecting a preset number of documents with the highest similarity as searched results and outputting the searched results. Through the application of the method and the device, the document recommending efficiency can be increased.

The invention discloses a personalized recommendation method based on the cloud processing mode and applied in the e-business environment. The personalized recommendation method mainly solves the problem that an existing personalized recommendation method is low in recommendation efficiency and poor in recommendation precision when processing mass data. The personalized recommendation method is divided into an off-line portion and an on-line portion. According to the off-line portion, the Hadoop frame of the cloud computing technology is used for parallel processing of historical data information, an HDFS is used for storing mass data information, and four kinds of parallelized recommendation methods which are suitable for different business stages of e-business are achieved according to the MapReduce programming model. According to the on-line portion, a lightweight data base is arranged and used for storing a user behavior log, a dynamic data collection mechanism is designed and used for reading data which are processed and obtained by the off-line portion in real time, web display and information statistics service are provided, and real-time recommendation information is provided for a user. The personalized recommendation method based on the cloud processing mode and applied in the e-business environment has the remarkable advantages of processing the mass data generated by e-business application.

The invention provides a method and device for automatic recommendation application. The method includes the steps of collecting user access information, dividing categories which the user access information belongs to, searching for matched applications in an application data set of preset corresponding categories according to the user access information and the categories of the user access information, generating application files corresponding to the categories and putting the searched applications of the categories in the corresponding application files to carry out recommendation. According to the method and device, individualized demands of users can be met, and recommendation efficiency and the coverage rate are improved.

The invention relates to a recommendation sorting method based on a wide and deep gate loop joint model, and belongs to the technical field of natural language processing. According to the method, Sina Weibo data are first crawled for preprocessing to obtain a topic feature set; then generalized cross-feature conversion is used to memorize topic features, and the same is input into a linear module; then an embedding vector is learned for each classification feature, all the embedding vectors are connected with dense features, dense vectors generated by connection are input to a deep module formed by gate loop units; and finally, parameters in linear and deep loop processes are simultaneously optimized, and a recommendation sorting result is obtained through joint training on the model. According to the method, the gate loop units are used for feature generalization, the problem that conventional methods mostly do not consider sequence features of dynamic time sequences is alleviated, abetter recommendation result is achieved as a whole, and recommendation efficiency is also improved to a certain extent.

The invention provides a method for automatically completing codes on the basis of LSTM (Long Short Term Memory). The method comprises the following steps: performing source code processing, and analyzing source codes by using an abstract grammar tree; performing off-line training, and training a language model by using an LSTM model; performing on-line code completion, and completing codes according to the trained language model. The LSTM model comprises constraint character level LSTM and mark character level LSTM with former context marking character encodes. By adopting the method, automatic code completion can be achieved by inputting any character into any place in the encoding process, recommendation of any code can be also achieved, and the recommendation process accuracy can be also ensured. The technical scheme of the invention has the characteristics of being simple and rapid, and the code recommendation accuracy and the recommendation efficiency can be relatively well improved.

The embodiment of the invention discloses a news recommending method and a news recommending device based on artificial intelligence. The news recommending method comprises the following steps: acquiring a first news characteristic of news to be recommended, and a second news characteristic of browsed news; according to the first news characteristic and the second news characteristic, determining whether the news to be recommended is the same as the browsed news or not; if the news to be recommended is the same as the browsed news, refusing to recommend the news to be recommended; if the news to be recommended is not the same as the browsed news, recommending the news to be recommended. According to the embodiment of the invention, as whether the news to be recommended is the same as the browsed news or not is determined, and only different pieces of news are recommended to a user, the phenomenon that same news is recommended to the user repeatedly can be effectively avoided, and the news recommendation efficiency can be improved.

The invention relates to the technical field of artificial intelligence, and discloses an information recommendation method and device based on a knowledge graph, equipment and a storage medium, and the method comprises the steps: obtaining initial data, recognizing the relationship between entities in the initial data, and constructing the knowledge graph; when a target customer is determined, extracting initial data of the target customer from the knowledge graph, constructing a sub-graph, then training a graph convolutional neural network GCN by adopting the sub-graph and a pre-constructedproduct feature vector, inputting product information data in basic data into the trained graph convolutional neural network GCN, and carrying out binary classification processing; and obtaining a selection probability of each product, selecting the corresponding product information data as to-be-recommended information according to the selection probability, and pushing the to-be-recommended information to the target client. The invention also relates to blockchain technology. Initial data is stored in the blockchain. According to the invention, the knowledge graph is constructed to improve the efficiency of information recommendation.

The invention relates to the field of network technologies, in particular to a music recommendation method and system. The method applied to the music recommendation system includes: receiving a music recommendation request; judging whether a user corresponding to the music recommendation request is a new user or not; if yes, acquiring behavior data of the user and / or first music resource data related to the user by at least one external means, processing the behavior data of the user and / or first music resource data related to the user acquired by at least one external means, classifying the user to acquire the level of the user according to processing results, and entering a step of recommending personal music by a recommendation method corresponding to the level according to the level of the user; if the user is not new user, acquiring the corresponding level of the user; recommending personal music by a recommendation method corresponding to the level according to the corresponding level of the user.