25results about How to "Accurate recommendation" patented technology

The application discloses a personality matching method based on a large language model, comprising the following steps: S1, obtaining personality labels and corresponding label scores of a user based on basic information and behavior events of the user; S2, mapping the label scores of each personality label to an axis of love and a domination axis in an interpersonal circular model based on the large language model, and generating corresponding axis of love scores and domination axis scores; S3, calculating personality matching degrees between different users based on the personality labels and the corresponding axis of love scores and domination axis scores of the different users. The personality matching method based on the large language model provided by the application constructs a general personality matching degree calculation model, and personality characteristics in user behavior events are mined to improve the stability of matching results and user satisfaction; and the method can be deeply integrated into a recommendation algorithm, and finally, accurate recommendation is realized.

The embodiment of the invention provides a data processing method and system.The data processing method comprises the steps that region feature data and a user associated with a target region are determined, and candidate objects are determined based on the region feature data and user data of the user; generating object features based on the candidate objects and the regional feature data, and generating user features based on the user data and the regional feature data; and inputting the object features and the user features into an object processing model to obtain a candidate object sequence, and updating the candidate object sequence based on a region object set associated with the target region to obtain a target candidate object sequence.

The application relates to the technical field of intelligent ordering, and particularly discloses an intelligent ordering method and system based on multi-party information synchronization of a back kitchen. The application synchronously updates multi-party information of the back kitchen, synchronously optimizes a complete restaurant menu, obtains historical ordering data of multiple customers at the same table, performs dish intersection analysis on the multiple historical ordering data, selects multiple intersection dishes of the same table, performs correlation analysis, selects multiple correlation dishes of the same table, arranges and optimizes the synchronously optimized menu, generates an arranged and optimized menu, and performs ordering display. The historical ordering data of multiple customers at the same table can be subjected to dish intersection analysis, multiple intersection dishes of the same table can be selected, multiple correlation dishes of the same table can be selected, the arranged and optimized menu can be subjected to arrangement optimization and ordering display, the dish preferences of customers at the same table can be deeply mined and analyzed, and then accurate and personalized menu recommendation and optimization can be provided for the customers at the same table, so that a customer-centered ordering experience is realized.

The application relates to a renewable energy site selection method considering ecological service value and belongs to the technical field of renewable energy site selection. The application considers the ecosystem influence of energy deployment, and proposes a renewable energy site selection method which comprehensively considers energy supply potential, economic, social, technical cost and ecosystem influence, and the method comprises the following steps: collecting basic data, processing obtained geographic information data; screening ecosystem service types, and determining various indexes involved in site selection; setting a limiting factor, and removing unfeasible patches in a region to be selected; calculating traditional index values and ecosystem service value related index values; determining the index weight by using an analytic hierarchy process; and quantitatively evaluating site selection suitability to obtain a renewable energy suitability distribution map. Therefore, the economic, social, technical and ecological comprehensive benefits of renewable energy site selection are optimized.

This application relates to a digital avatar recommendation method, apparatus, computer equipment, storage medium, and computer program product. The method includes: responding to a digital avatar recommendation request initiated for a virtual branch office used for handling financial transactions, obtaining digital space scene characteristics of the virtual branch office, usage statistics for each shared digital avatar, and digital avatar description information; the shared digital avatars are digital avatars used for shared use across multiple digital space platforms associated with a user account; the multiple digital space platforms are digital space platforms other than the virtual branch office; based on the digital space scene characteristics, usage statistics, and digital avatar description information, determining a recommended digital avatar among the shared digital avatars; and recommending the user account to use the recommended digital avatar in the virtual branch office. This method can enrich the digital avatar resources of bank virtual branches.

This application discloses a cross-domain multimodal recommendation method and system based on multi-agent attribute-driven architecture. Embodiments of this application can be applied to product recommendation scenarios in an e-commerce platform. The method includes: receiving multimodal features; processing the multimodal features using a first operation to generate reconstructed original multimodal features; the reconstructed original multimodal features include attribute-related representations and attribute-independent representations; then, processing the reconstructed original multimodal features using a second operation to generate a modal cue vector; then, processing the modal cue vector and the attribute-related representation to generate a new feature vector; and finally, processing the new feature vector using a sequence encoder to generate attribute-level multimodal sequence states. By sharing basic embeddings, the parameter overhead is significantly reduced, solving the problem that the parameter size of traditional cross-domain multimodal encoders increases linearly with the number of domains.

This application discloses a refrigerant recommendation method, apparatus, electronic device, and computer-readable medium, relating to the field of logistics and transportation technology. The method includes: acquiring cold chain order data; extracting temperature zones, timeliness, and warehouse locations from the cold chain order data; querying temperature zone configurations based on warehouse location and current time to determine the corresponding temperature zone; determining a corresponding basic scheme based on the temperature zone, timeliness, and temperature zone query scheme group configuration, wherein the basic scheme includes specific refrigerant combinations, dosage quantities, and auxiliary consumables; and outputting refrigerant recommendation results based on the basic scheme upon successful verification of the basic scheme. Through hierarchical management of the configuration, structured storage and efficient matching of refrigerant recommendation rules are achieved. Multi-dimensional intelligent matching is employed, comprehensively considering multiple factors such as temperature zones, timeliness, and temperature levels, to achieve efficient and accurate refrigerant recommendation, improve cold chain order fulfillment efficiency, reduce product damage rates due to improper refrigerant dosage, and enhance product quality assurance capabilities.

The application discloses a kind of personalized intelligent recipe recommendation system based on user health information, comprising: user information management module, for obtaining and managing the health information of user;Energy calculation module, for calculating ideal body weight, body type classification, basal energy requirement, blood sugar adjusted energy requirement and meal nutrient target according to user information;Food database, for storing the nutrient composition and type attribute of food;Recipe generation module, for selecting food combination from food database according to meal type and nutrient target, and generating personalized recipe using multi-nutrient synergistic optimization intelligent algorithm.The present application is based on the principle of nutrition, on the basis of personalized health management, increase the precise energy calculation function based on ideal body weight and body type classification, realize precision and individualization in energy calculation, nutrient optimization, food selection and other multiple links.

The application discloses a cold start wine recommendation method and device based on multi-modal hypergraph smoothing and three-view contrast learning, electronic equipment and a storage medium, and belongs to the technical field of electric digital data processing. A multi-modal heterogeneous hypergraph for a cold start state of wine commodities is constructed; multi-modal heterogeneous hypergraph features are smoothed and enhanced, and collaborative knowledge of existing wine commodities is diffused to cold start wine representations through hypergraph neighborhood; a multi-view contrast learning framework including a content view, a structure view and a multi-modal fusion view is constructed; a three-way symmetric mutual supervision ternary contrast loss function is constructed, and three types of representations, namely, content, structure and multi-modal, are globally semantically aligned; an interaction probability of a customer and each new wine product is calculated, and a recommended product sequence is generated for the customer based on the interaction probability. The application realizes accurate recommendation of new wine in a cold start scenario.

The application provides a movie recommendation method and system based on a cross attention mechanism, wherein the method comprises the following steps: constructing a user-movie feature matrix according to user information and movie information; extracting attention features in the user-movie feature matrix; extracting user-movie interaction features in the user-movie feature matrix; fusing the attention features and the user-movie interaction features; and performing movie rating prediction by using the fused features, the attention features and the user-movie interaction features, so as to perform movie recommendation according to the predicted rating. The application improves the accuracy of movie rating prediction, improves the accuracy of movie viewing user recommendation results, can more fully mine user information, realizes accurate movie recommendation, and improves user use satisfaction.

Embodiments of the present application disclose a content recommendation method and device, computer equipment and a computer readable storage medium. Based on a plurality of historical interaction contents of a target object, content attribute information sequences under a plurality of attributes are obtained, each content attribute information sequence including attribute information of the plurality of historical interaction contents under the same attribute. The content attribute information sequences are subjected to attention feature mining processing to obtain an attention feature information sequence corresponding to each attribute. The attention feature information sequences are subjected to first feature fusion processing to obtain a fused attention feature information sequence. A behavior feature information sequence of the target object is generated according to a target content attribute information sequence belonging to a content identifier attribute in the content attribute information sequences. The fused attention feature information sequence and the behavior feature information sequence are subjected to second feature fusion processing to obtain object feature information of the target object. The object feature information is used to recommend content to the target object, and the content can be accurately recommended.

The application discloses an AI intelligent question analysis and interaction method and system and relates to the field of question analysis. A question sentence, an answer sentence and an answer sentence are acquired. Based on the question sentence, the answer sentence and the answer sentence, an error step and a missing step are segmented, and a keyword sequence is obtained. Through a question recommendation network, based on the keyword sequence, the relationship between the question and the error step and the missing step is detected, and a student interaction question is obtained. According to the correlation relationship between the keywords of each answer step and the keywords of the question, the importance of each answer step for solving the question can be known, and a step importance value is obtained. According to the step importance value, the question recommendation network is constructed, and the technical effect of more accurately recommending the question with easy-to-miss points to the student is achieved.

This application provides a parameter tuning method, apparatus, device, and computer storage medium for multiple objectives. The method includes: collecting historical behavior information of multiple users regarding recommended materials; determining the revenue parameter vector corresponding to each of the N sets of fusion parameter vectors in the fusion network based on the historical behavior information; determining the correlation coefficient between each set of fusion parameter vectors and the revenue parameter vector corresponding to each set of fusion parameter vectors; determining the increment vector between the first revenue parameter vector and each second revenue parameter vector in the N sets of revenue parameter vectors corresponding to the N sets of fusion parameter vectors; updating the mean vector of the fusion parameter vector corresponding to each second revenue parameter vector based on each set of increment vectors; and updating the N sets of fusion parameter vectors based on the updated mean vector and the correlation coefficient. This application improves the efficiency and speed of fusion parameter vector optimization, achieving more accurate material recommendations.

This invention discloses a method and apparatus for recommending job postings based on career prediction, relating to the field of online recruitment technology. The method includes: acquiring user profile data of job seekers; predicting the next virtual work experience based on the job seeker's resume information; vectorizing the next virtual work experience to obtain vectorized representations of multiple virtual work experiences, and adding them to the user profile data; performing vector retrieval on the vectorized representations of all job posting information to obtain vectorized representations of several job postings that match the vectorized representations of the virtual work experiences; filtering and sorting the candidate set of job postings according to screening criteria, and using the sorted job postings as the job posting recommendation results for the job seeker. This invention significantly overcomes the strong dependence on explicit user needs, resume completeness, and behavioral data, effectively solving the problems of fuzzy job seeker needs, data sparsity, and cold start.

The embodiment of the invention provides a video clip recommendation method and device, a storage medium and electronic equipment, and the method comprises the steps: determining an instant intention vector of a target object through the interaction operation of the target object on a to-be-processed video, and carrying out the similarity matching based on the instant intention vector and a plurality of candidate video clips, the plurality of candidate video clips are video clips extracted from a to-be-processed video in advance; the signal change trends of the multi-modal signals included in the candidate video clip at the multiple time points are the same, so that the candidate video clip can be matched with the instant intention vector, and the target video clip is determined. The recommendation logic not only considers the internal signal change trend of the to-be-processed video, but also combines the instant intention vector of the target object, so that accurate recommendation can be realized. Therefore, the problem that in the related technology, the precision is poor when video clip recommendation is carried out can be solved.

The invention relates to the technical field of sleep management, and discloses a sleep-aiding music playing method and system based on multi-sensor fusion, and the method comprises the steps: collecting a sleep sound signal of a user, and determining a sound feature coefficient according to the sleep sound signal; collecting a sleep acceleration signal of a user, and determining an acceleration characteristic coefficient according to the sleep acceleration signal; and collecting sleep eye movement data of the user, and determining sleep features of the user according to the sleep eye movement data, the sound feature coefficient and the acceleration feature coefficient. Multi-dimensional feature extraction is performed on the sleep signal of the user, so that more accurate sleep features of the user are extracted to realize accurate recommendation of sleep-aiding music.

The application discloses a sequence recommendation method based on an edge-enhanced global decoupling graph neural network, comprising: an input layer: all items in the training data are taken as nodes, and the interaction order between the user and the item is taken as an edge to construct a global link graph and input into the graph neural network; a decoupling learning layer: the influence probability of a preset number of factors from all neighbor items of the item and the influence probability of a preset number of factors from the target user's items in the front according to the interaction time sequence; a prediction layer: the decoupled item representation of the global level and the local level is accumulated, multiplied by the initial embedding representation of the candidate item, and the probability that the target user appears as the next interaction item v i The application has the beneficial effects of perceiving the transfer of the user's intention and more accurately recommending the product to the user.

The invention provides a marketing dialogue method and device based on a thinking chain and a user portrait and electronic equipment, and belongs to the technical field of artificial intelligence, and the method comprises the steps: based on user input and the user portrait of the current round, employing a recall and sorting mode to update a candidate marketing item set; inputting the candidate marketing item set, the user portrait, the historical dialogue state and the user state into a dialogue large model to output a reply text, the dialogue state and the user state of the current round; based on the user input and the user state of the current round, updating the user portrait to obtain the user portrait of the next round; in the next round of dialogue, the user portrait of the next round, the dialogue state of the current round and the user state serve as context information to be input, and a closed-loop feedback mechanism is formed. According to the method, the whole marketing conversation process is automatic and intelligent, the intention of the user can be dynamically understood, the strategy can be adjusted, and the accuracy and conversion efficiency of the marketing conversation are improved.

The application provides a product intelligent recommendation method, system, device and medium based on a Dify platform, belongs to the technical field of artificial intelligence information recommendation, and the method comprises the following steps: uploading structured product data and question and answer text based on the Dify, generating a product knowledge base according to the structured product data, and generating a question and answer knowledge base according to the question and answer text; receiving a question input by a user, calling a semantic analysis service of the Dify to extract a structured query object; performing retrieval and matching of the best answer to the question and answer knowledge base according to the extracted structured query object, and calling an AI large model to generate the best answer when the retrieval and matching fails; extracting keywords from the best answer, and screening the best product from the product knowledge base by using the keywords; and returning the screened best product and the matched answer text to the user. The product recommendation efficiency and accuracy are improved by combining the product knowledge base, the question and answer knowledge base, the semantic analysis service of the Dify and the AI model.

The present application relates to the field of electric digital data processing, in particular to a data analysis method for optimizing search engine recommended content, user data, information data and user interaction with information data are collected; a concern degree coefficient is obtained according to the user's stay time, interaction and recommended browsing time of information; an information data vector of each information data is constructed according to the concern degree coefficient of each user to each information data; the push weight of the user to the search material content is obtained according to the relationship between the information data vector of the information data and the search material content and the concern degree coefficient of the user to each information data; the search material content is accurately pushed to each user according to the push weight of the user to the search material content. According to different users and different interactive behaviors, the present application determines the push weight of the search material content to different users, and realizes accurate recommendation of the search material content.

The application provides a game nickname recommendation method, device, system and storage medium. A game nickname recommendation request sent by a client is received. The request carries a nickname demand parameter. According to the nickname demand parameter carried in the request, a to-be-recommended game nickname corresponding to the nickname demand parameter is obtained by searching a preset multilingual game nickname database. The multilingual game nickname database stores a plurality of game nickname data sets corresponding to a plurality of language types. Each game nickname in each game nickname data set corresponds to a category label. The to-be-recommended game nickname is returned to the client. According to the nickname recommendation request initiated by the user and containing a plurality of nickname demand parameters, and the multilingual game nickname database pre-constructed and including a plurality of language types and the nicknames labeled with corresponding category labels, accurate game nickname searching and recommendation are performed, the self-defined demand of the user in a plurality of scenarios is met, and the game experience of the user is improved.

The invention relates to the technical field of computers, and particularly provides an account exception monitoring method and device, equipment, a storage medium and a program product. The method comprises the steps of obtaining transaction order sequence data of a target account in a preset historical time period and a pre-constructed risk monitoring model; wherein the risk monitoring model comprises a depth auto-encoder and a multi-granularity feature pyramid; performing feature extraction on the transaction order sequence data through a deep auto-encoder to obtain space-time compression features; performing multi-granularity processing on the space-time compression features through a multi-granularity feature pyramid to obtain a multi-granularity anomaly monitoring result; and performing risk monitoring on the target account based on the multi-granularity anomaly monitoring result. According to the method and the device, spatial-temporal characteristics can be fused, multi-granularity anomaly recognition is supported, and the method and the device have real-time response capability, so that effective prevention and control of risk behaviors such as arbitrage are realized while products are accurately recommended.

The application relates to the technical field of digital medical treatment, and particularly discloses a cold start recommendation method and device, a storage medium and a computer device. The method comprises the following steps: determining a cold start recommendation object, and acquiring a first attribute vector of the cold start recommendation object; acquiring a second attribute vector corresponding to each of a plurality of to-be-matched objects; determining a first mapping vector of the cold start recommendation object based on an initial vector of the cold start recommendation object and the first attribute vector; determining a second mapping vector of each of the to-be-matched objects based on a target vector of each of the to-be-matched objects and the second attribute vector; determining a matching degree between the cold start recommendation object and each of the to-be-matched objects according to the first mapping vector and the second mapping vector; determining a target matching object of the cold start recommendation object based on the matching degree; and recommending the target matching object to the cold start recommendation object.

This invention relates to the field of data processing, and more specifically, to a method for profiling electricity consumption behavior based on big data multi-model analysis. The method includes: collecting historical load data to construct user electricity consumption vectors; calculating the cosine similarity of any two electricity consumption vectors as a first similarity, performing mean clustering to obtain multiple first categories; calculating the cluster centers of the first categories to obtain first and second electricity consumption vectors, and calculating the cosine similarity of the two electricity consumption vectors to obtain a second similarity, performing multi-threshold segmentation to divide into multiple second categories, and using factor analysis to obtain the information content of each unique factor; calculating the separation necessity value, separation requirement value, and separation value for each second category; and determining the user's electricity consumption behavior profile based on the separation value. This invention finely classifies users, obtains personalized needs for different user profiles, and achieves the goal of accurate recommendations.

The application relates to a point of interest recommendation method based on a knowledge graph and a graph attention convolution network, and has the technical characteristics that a user-point of interest interaction knowledge graph is constructed; a user geographic preference coefficient is calculated to generate a user geographic preference matrix H; a graph neural network is initialized to obtain a user vector and a point of interest vector, and the geographic preference coefficient of the user is taken as message transmission; entity embedding and neighborhood entity embedding are carried out; convolution is carried out on the user-point of interest interaction knowledge graph to obtain a trained model and updated user vectors and point of interest vectors; a preference score of the user for the point of interest is calculated, and a POI recommendation list interested by the user is generated in combination with TOP-K sorting. The application is reasonable in design, effectively improves the defects that a traditional point of interest recommendation method ignores social information of the user and geographic attribute information of the user, improves the accuracy of a point of interest recommendation task, and greatly reduces the computing power cost.