123results about How to "Rich dimension" patented technology

The embodiment of the invention discloses a management node of a load balancing control method. The management node and a plurality of physical hosts are contained in a trunking system, and each physical host comprises a hardware layer, a host machine running on the hardware layer and a virtual machine running on the host machine. The method comprises the steps that the current comprehensive load balancing value of the trunking system is calculated; when the comprehensive load balancing value is larger than a balance degree threshold value, candidate virtual machines to be transferred are selected from the virtual machines running on all the physical hosts in the trunking system to be added to a candidate virtual machine set based on computing resource dimensionality and network resource dimensionality respectively; a target virtual machine to be transferred in the candidate virtual machine set and a target physical host hosted by the target virtual machine are determined based on comprehensive dimensionality; the comprehensive dimensionality comprises the computing resource dimensionality and the network resource dimensionality; a source physical host hosted by the target virtual machine to be transferred is informed to transfer the target virtual machine to be transferred to the target physical host; the source host and the target physical host are both the physical hosts in the trunking system.

The invention discloses a method and a system for performing intervention according to information of chronic disease patients. The method comprises the steps that out-hospital health reminding schemes of the patients are generated according to hospital leaving advices given by a doctor; in-hospital diagnosis and treatment information, out-hospital medication information, vital sign information and healthy living mode information of the patients are collected, the collected information is processed, and labels are set; the patients are dynamically grouped according to the labels; and the doctor sends intervention information to the patients in a whole group and pushes the intervention information to each patient in the group. According to the embodiment, the collected patient information is more comprehensive in terms of dimension, contains patient diagnosis and treatment information, advice scheme information and self-examination information and also comprises patient daily behavior evaluation information, a data basis is provided for precise intervention by the doctor, precise dynamic label identification of the patients and intelligent grouping depending on the labels are realized, the doctor can quickly know the current health states of the patients by spending little time, and increase of the intervention workload of the doctor due to increase of the number of the patients managed by the doctor is avoided.

The invention discloses a malicious domain name detection method based on a knowledge graph, which comprises the following steps: S1, extracting key features for constructing a domain name reputation knowledge graph according to domain name information features; S2, integrating the information contained in the domain name by using the obtained existing data, and providing a data source for the creation of the knowledge graph; amplifying domain name knowledge by the data through a knowledge graph construction engine; enabling the knowledge graph construction engine to construct a domain name reputation knowledge graph by using the obtained domain name data; and S3, for a given specific domain name, giving a prediction result by using the multi-source information and a plurality of method judgment models. The method has the advantages that 1, domain name information is richer in dimension, and contained information is more complete; 2, outbreak growth of domain name data can be better coped with, and mass data can be processed more quickly through construction of a knowledge graph; and 3, a relational network is formed by the domain name information and the behavior pattern, and the information in the relational network is deeply mined by utilizing the sensitivity of the knowledge graph to the relational network.

The invention discloses a question labelling method based on deep reinforcement learning for an online question and answer platform. The method is based on a deep neural network and a reinforcement learning technology. According to the technical scheme, the method comprises the following steps: firstly, establishing a model, innovatively adding an index for measuring the diversity of problem labels while ensuring the labeling accuracy when designing model rewards, and meanwhile, considering a tail label effect, namely, long labels which are described in a complex and detailed manner during labeling, so that the labels can describe the problems deeply and more detailedly. According to the method, the accuracy and diversity of labels are comprehensively considered, the influence of the taillabel effect on problem labeling is reduced, the training efficiency and accuracy of the reinforcement learning model are improved by introducing the deep neural network, and the matching error rangecan be ensured under a certain confidence degree. According to the scheme provided by the invention, mass questions and labels in the question and answer platform can be accurately and diversely matched.

The invention discloses a pre-loan risk analysis method, and the method comprises the steps: obtaining first return data of a user, building a variable pool based on the first return data, and the variable pool comprising time slice variables of multiple dimensions; performing time calibration on the time slice variables in the variable pool to obtain calibrated time slice variables after time calibration; screening out a target time slice variable meeting a preset condition from the calibration time slice variables, and performing grouping variable combination on the target time slice variable to obtain a variable short list combination; carrying out in-mold variable screening on the variable short list combination to obtain a model group; and obtaining each variable index, obtaining a prediction model meeting each variable index in the model group, and carrying out pre-loan risk analysis according to the prediction model. The invention further discloses a pre-loan risk analysis device and equipment and a readable storage medium. According to the method, the information dimension is enriched, the situation of causal mismatching is prevented, and the accuracy of a risk analysis result is improved.

The invention provides a human-computer interaction method and system based on a multi-modal historical response result. The human-computer interaction method based on the multi-modal historical response result includes the steps of receiving a voice instruction of a user; performing voice recognition on the voice instruction to obtain a plurality of response results; acquiring a plurality of pieces of input feature information, and calculating the joint probability of each various response result according to a probability model and the plurality of pieces of input feature information; and determining the response result having the largest joint probability to respond to the voice instruction, wherein the probability model is a stochastic mathematical model established on the basis of the historical response results, and wherein each joint probability is the product of the probabilities of the response result under the plurality of pieces of input feature information. The embodiment of the invention enriches the dimension of the feature information inputted in determining the response result, comprehensively considers the influence of various factors on the true intention of the user, improves the accuracy of the response result, and improves the user experience in the human-computer interaction process.

A data processing system and method are provided. The system comprises a distributed scheduler, a data processing device and a storage component with a plurality of preset databases, wherein the distributed scheduler is adapted to acquire offline logs to be processed, extract log metadata from the offline logs, and transmit the offline logs and log metadata to the data processing device; the dataprocessing device is adapted to classify and merge the received offline logs to generate a corresponding virtual table, and statistics the log metadata to obtain the corresponding statistical information; the data processing device is further adapted to store virtual tables and statistical information in a storage component including a plurality of preset databases. The invention realizes the centralized index of the log, also improves the searching efficiency of the log, and can directly obtain the characteristic data of each dimension of the log, thereby greatly saving the flow of analysts in analyzing the log.

The invention discloses a multidimensional visual shopping guide system and method. The system comprises an input module, an intelligent shopping guide module and an air display screen, wherein the input module is used for acquiring at least one input signal of a user, wherein the input signal includes an image signal, a touch signal, a gesture signal, a voice signal and a brain wave signal; the intelligent shopping guide module is used for carrying out intelligent shopping guide according to at least one input signal, wherein the intelligent shopping guide includes the eyeball tracking of theuser, the voice recognition of the user, AI intelligent voice preach and the virtual model display control of an item; and the air display screen is used for displaying the virtual model of the itemthrough an air imaging mode. In the invention, all-weather commodity explanation and consultation services are provided for the user, service quality is high and a user experience is good; through theintelligent shopping guide, a focal point that the user pays attention to and the voice instruction of the user are identified; the virtual model which can carry out AI intelligent voice preach for the user and dynamically display the item at 360 degrees without dead angles for the user is provided; and the system is intuitive and has abundant dimensions and high interactivity, and the system andthe method can be widely applied in the intelligent control field.

The invention relates to an electric power credit investigation evaluation method based on a big data model. The method comprises the steps of collecting internal enterprise power data and external enterprise operation data; building credit loss power utilization enterprise features and high-risk credit loss power utilization enterprise features, building an enterprise credit investigation evaluation model for enterprise credit scoring by adopting an AHP analytic hierarchy process and a TOPSIS comprehensive evaluation method, determining a threshold value, and determining a user credit ratingaccording to a score range. According to the method, the dimension is more comprehensive, and the data reliability is higher; the method is advanced, the result is more accurate, and the enterprise credit analysis dimension is increased.

The invention discloses a method and a device for pushing user information. The method comprises: obtaining SNS network relation chain information and non-SNS network relation chain information of a user for whom the information is pushed; querying the user information in SNS server according to the obtained non-SNS network relation chain information, and obtaining identification information corresponding to the identifications in the SNS network relation chain information from the non-SNS network relation chain information; and eliminating the identification information identical to that of the SNS network relation chain information from the obtained identification information, obtaining the filtering result, and pushing the filtering result to the user for whom the information is pushed. According to the method and the device, the accuracy of pushing the information for the user is improved.

The invention relates to a medical information processing method. The method comprises the steps of importing all medical information; preprocessing the medical information to obtain preprocessed medical information; processing the preprocessed medical information according to rule information to obtain first medical information; extracting second medical information from the first medical information according to an extraction rule; and obtaining sampled data from the second medical information. According to the medical information processing method provided by the invention, all recipes/medical advices are automatically analyzed by utilizing software, and rich recipes/medical advice sampling methods are provided, so that personalized special comment of the recipes/medical advices is realized.

The invention discloses a financial credit risk identification model construction method and device, a financial credit risk identification method and device, and a computer readable storage medium, and the method comprises the steps: generating a financial credit knowledge graph according to the information of a loan subject; generating a financial cause map according to the financial event dataand the financial domain ontology; based on the information of the loan subject, the financial credit knowledge graph and the financial affair graph, generating a financial credit risk identificationmodel through machine learning; and updating the financial affair map according to a real-time financial event, and outputting financial credit risk information of the real-time financial event for the to-be-predicted loan subject through the financial credit risk identification model according to the updated financial affair map and the information of the to-be-predicted loan subject. According to the method of the invention, the financial credit risk can be predicted according to the real-time financial event, the dimensionality of risk identification is enriched, and the anticipation of thecredit risk of the credit subject is improved.

The invention belongs to a network space surveying and mapping method based on service verification, and belongs to the technical field of network security. The method comprises the following steps of: A, analyzing a flow log, mirror flow and offline data in sequence to obtain an analysis result; b, verifying the analysis result through a script to obtain a verification result; and C, perfecting the space surveying and mapping information through a verification result. The space surveying and mapping efficiency is improved by utilizing server IP, port and protocol information acquired by dataacquisition, dimensions of space surveying and mapping asset description are enriched, and validity of verification numbers is actively detected, so that reliability of whole space surveying and mapping data is improved, and huge burden on a network caused by an active scanning mode is avoided.

The embodiment of the invention discloses a publication issuing analysis system and method. The system comprises a data collection subsystem, a cleaning layer, a data layer, a processing layer, an application analysis subsystem and a display layer, wherein the data collection subsystem is used for acquiring publication-related Internet crawler data and/or ground sale data; the cleaning layer is used for cleaning the acquired data to obtain original data; the data layer is used for storing the original data; the processing layer is used for reading the original data in the data layer and performing scheduling and mining analysis on the original data to obtain basic analysis data; the application analysis subsystem is used for obtaining publication issuing analysis result data according to the basic analysis data; and the display layer is used for displaying an analysis result according to the publication issuing analysis result data. According to the technical scheme in the embodiment, the precise and reliable analysis result is provided for publication issuing by the adoption of big data processing relying on mass Internet data and/or ground sale data, and precise and effective decision information is provided for publication issuing.

The invention discloses an infant cry detection deep learning method, and relates to the technical field of voice signal processing. The method comprises the following steps: a, acquiring a voice signal; b, framing the voice signal segment and performing cochlear voice feature extraction on each frame; and c, inputting the adjacent N frames of voice features into a pre-trained baby cry detection deep learning model for reasoning and judging whether cry exists or not and d, voting the N frames of classification results by applying a majority-first voting principle, and judging whether baby cryexists in the N frames or not. The cochlear speech features adopted by the method are speech feature parameters more conforming to auditory perception analysis of people, and the convolutional networkand the long-short-term memory recurrent neural network are adopted as acoustic inference models for infant cry detection, so that the method can adapt to a speech environment with a low signal-to-noise ratio, and has higher accuracy compared with a traditional method.

The embodiment of the invention discloses a risk decision method and a risk decision device, wherein, the risk decision method comprises the following steps: monitoring an electronic device of a userand obtaining risk-related data in the electronic device; parameterizing the risk-related data to obtain the standard parameters suitable for the rule engine; importing the standard parameters into the rule engine to obtain the risk rating; making risk decisions based on risk ratings. By adopting the invention, the generation of risk decision can be influenced through standardization of risk related data, rule engine processing and risk rating, and the adopted parameters and rule engine improve the comprehensiveness and normalization of risk rating, and the accuracy and referenceability of risk decision can be improved through risk rating.

The invention relates to a dynamic video tactile feature extraction and rendering method, and belongs to the field of virtual reality and human-computer interaction. The method comprises the steps ofdecompressing a received video, preprocessing the video, segmenting shots based on inter-frame color histogram features, extracting saliency maps fused with time-space domain tactile saliency featuresfrom all the segmented frames in each shot, and performing pixel-level tactile rendering according to the saliency maps of the video frames. According to the method, the video content is divided intothe salient region and the non-salient region by extracting the video frame salient features fused with the space-time characteristics. And pixel-level tactile stimulation is applied to the video frame by adopting a one-to-one mapping relationship between the visual channels and the tactile channels. And real-time tactile feedback is generated through the terminal, so that the realistic experience of watching the video by the user is enriched. The method can be widely applied to video education, multimedia entertainment and human-computer interaction.

The invention relates to a battery remaining capacity detection method and device, electronic equipment and a storage medium. The method comprises the following steps: acquiring an accumulated consumed electric energy value of a battery of the electronic equipment from a discharge starting moment to a target moment; and acquiring a full-charge electric energy value of the battery, and taking a difference value between the full-charge electric energy value and the accumulated consumed electric energy value as a residual electric energy value of the battery at the target moment, wherein the residual electric energy value is positively correlated with the residual electric quantity of the battery at the target moment. The method can improve the detection accuracy of the residual electric quantity of the battery.

A computer system adapted to store a record of a patient's data received from at least one of doctor, patient, hospital, therapist, wherein the data is in at least one format selected from audio, visual, text, and wherein a patient can access the patient's own record and optionally add comments in one of said formats is disclosed. Also disclosed is a method of storing patient's data, comprising inputting data from at least one of doctor, hospital, or therapist in at least one format selected from audio, visual, text and authorizing the patient to access that patient's record so that patient can review the accuracy of the data and to input additional data or corrected data in one of said formats if the patient wishes to add comments to the record.

The invention provides a robot positioning method and device, and a storage medium. The robot positioning method comprises the following steps of: acquiring the laser point cloud information and grayscale image of an environment in which a robot is located, wherein the laser point cloud information comprises the position information of multiple laser points; projecting each laser joint into the grayscale image of the environment in which the robot is located according to the position information of each laser point, and obtaining the grayscale value of a projection point corresponding to eachlaser point; determining the probability of the observation value of each laser point position according to the grayscale value of the projection point corresponding to each laser point and the distance from each laser point to the nearest obstacle in the environmental map where the robot is located, wherein the environmental map where the robot is located is established and stored in advance, andthe distance from each laser point position to the nearest obstacle in the corresponding environmental map is calculated and stored in advance; and locating the robot according to a particle filter positioning algorithm and the probability of the observation value of each laser point position.

The embodiment of the invention provides an academic relationship prediction method and device based on a neural network introducing semantic information, and the method comprises the steps: determining a node information combination of two to-be-predicted nodes, enabling the node information of scholarship nodes to comprise the scholarship personal information, and enabling the node information of paper nodes to comprise a paper title and a paper abstract; and inputting the node information combination into a relationship prediction model, and outputting a relationship type vector of two nodes. The relationship prediction model is obtained by training based on sample node information and a predetermined relationship type vector label between sample nodes; and the feature vectors of the sample scholarship nodes and the sample paper nodes in training are respectively composed of scholarship personal information features and random additional features, and paper title information features and paper abstract information features. According to the method and the device provided by the embodiment of the invention, paper semantic information is considered, so that the academic relationship prediction accuracy is improved, and more types of academic relationships can be mined.