75results about How to "In line with expectations" patented technology

The invention discloses an intelligent dialogue generation method and a device, and a computer storage medium, relates to the technical field of artificial intelligence, and can comprehensively consider factors of dialogue contents and emotional emotions, so that answers of intelligent dialogues more accord with user expectations, and the interactivity of the intelligent dialogues is improved. Themethod comprises the steps of obtaining a question and answer corpus in an intelligent dialogue; based on a pre-trained emotion classifier, performing emotion marking on the question and answer corpus in the intelligent dialogue; inputting the question and answer corpus carrying the emotion label into a network model for training to obtain a dialogue control model; when a dialogue generation request is received, inputting a question statement requesting dialogue generation into a coding part of the dialogue control model to obtain a context semantic vector; and jointly inputting the emotion classification label requesting to generate the dialogue and the context semantic vector into a decoding part of a dialogue control model to obtain word vector representation of an answer statement matched with the emotion classification label, and outputting the corresponding answer statement after the word vector representation is formed.

The invention discloses a method and a system for core process knowledge intelligent pushing based on multi-model fusion. The method comprises: preprocessing existing corpus data, and then inputting the data into a classification algorithm model for pre-classification, improving effect of classification through model fusion, when user queries or user feeds back, performing similarity calculation on user input and text categories, to determine the categories to which keywords belong, taking first k1 most similar categories, just retrieving in the categories, for each category, using the input keywords to retrieve respectively using different models in the category, combining all previous results, performing relevancy sorting using a BM25 algorithm, taking first k2 results, and using Jaccardsimilarity to remove texts which are too similar in the results, finally, returning the results to users. The method and the system can further adjust user's keyword models according to the feedbackof the users, and better fit needs of the users, so as to optimize user pushing effect and matching degree in the next pushing.

The invention discloses a method for sending an instant message. A sending party can be allowed to carry out a tracking or processing operation on a sent instant message before the instant message arrives at a receiving party through optimizing and transforming an instant message system and controlling the position and the state of the instant message in the transmission process, thereby realizing that the sending party user has the right of knowing, regretting, modifying or even withdrawing the sent message. Misunderstanding or other accidental circumstances generated in instant communication between people can be reduced in the message transmission process by adopting the technical solution disclosed by the invention, thereby enabling transmission of messages to be more accurate, conforming to expectations of users, and achieving the best communication effect.

The invention discloses a method of intelligently generating a room design scheme. The method includes: acquiring room design partitioning rules and grouping rules of rooms of different room types onthe basis of big data and machine learning; and generating the room design scheme of a current room according to the partitioning rules, the grouping rules and room type information of the current room, wherein the room type information includes door positions, window positions and the area of the room, the partitioning rules include rules of carrying out function region dividing on the room according to use functions, and the grouping rules include rules of placing furniture groups at different positions in all function regions. The method of intelligently generating the room design scheme provided by the invention realizes the room design scheme on the basis of the massive data and a machine learning algorithm, does not need manual setting of attribute labels as compared with the prior art, and can intelligently and quickly generate the design scheme, which more accords with user expectation, through machine learning.

The invention provides an online analytical processing method and device. The method comprises the steps that query instructions are received, the query instructions include pending to query conditions; a column storage child library associated with the pending to query conditions in the preset column storage library is determined based on the pending to query conditions, the preset column storage library includes at least a column storage child library, the column storage child library has at least a record according to the column storage, each record includes an identifier and data corresponding to the identifier, data of different column storage child library have different properties; records meeting the pending to query conditions are searched and obtained from the column storage library associated with the pending to query conditions. According to the technical scheme, the disk IO bottleneck problem caused by big data can be better solved, the speed and efficiency of online analytical processing are effectively improved, and large-scale promotions and applications of online analytical processing models are facilitated.

The invention relates to an intelligent water-saving tap for hot water. The intelligent water-saving tap comprises two parts, namely, a water tap and a controller, wherein one end of the controller is communicated with a hot water pipe while the other end of the controller is communicated with a circulating pipe; the hot water pipe and the circulating pipe are respectively communicated with a water heater; a hot water outlet communicated with the water tap is formed in a joint at which the hot water pipe is connected with the controller; the controller comprises a solenoid valve, a temperature sensor, a water pump and an electrical control box, and the solenoid valve is arranged in a joint at which the hot water outlet is connected with the hot water pipe; the temperature sensor is arranged at the upstream of the solenoid valve on the hot water pipe; the water pump is arranged between the hot water pipe and the circulating pipe; the electrical control box is used for controlling the water pump and the solenoid valve; and a signal of the temperature sensor is linked to the electrical control box. By adopting the water-saving tap provided by the invention, waste of a water resource caused by discharging certain amount of cold water before the hot water reaches the hot water outlet during a hot water using process can be avoided, so that a good water saving effect can be realized, and a user can feel comfortable during using.

The invention discloses a method for complementing an object shape in an RGBD image, comprising steps of adopting a pre-cut object in a single RGBD image as an input, estimating a preliminary complete shape in virtue of a same-category three-dimensional model in a database, and obtaining an optimal complete shape through an optimization algorithm based on block matching. The method for complementing the object shape in the RGBD image simplifies a complex process of three-dimensional data collection, user does not need to hold an EGBD camera to surround the object for multiple scanning, instead, the user only needs to push one button just like taking picture. In the backstage, the method of the invention can perform automatic completion on the data of the single RGBD image shot by the user, the shape of the object after completion does not only meet the expectation of the user from the appearance, the local details are consistent with the detail in a visible part of an RGBD image.

The embodiment of the disclosure provides a navigation method and device, a computer readable storage medium and electronic equipment, and belongs to the technical field of a computer and communication. The method comprises the steps of: displaying a first planned route when a terminal is under first positioning accuracy, wherein the first planned route comprises a first starting point position, and the first starting point position is an automatic positioning position; before detecting that the terminal initiates a navigation operation, acquiring second positioning accuracy of the terminal; acquiring a second starting point position when the terminal is under the second positioning accuracy, wherein the second positioning accuracy is smaller than the first positioning accuracy; and according to a distance between the first starting point position and the second starting point position, acquiring a second planned route, wherein the distance is greater than a distance threshold. The technical scheme of the embodiment of the disclosure provides the navigation method, so that accuracy of the planned routes in a navigation scene can be improved.

The invention discloses a page display method based on a user level. The page display method comprises the steps that a client sends an access request to a server; the server generates a correspondingdisplay page based on the current user level of the user in response to the access request of the client; the client acquires the generated display page from the server and displays the display pageto a user; wherein the display page comprises first display information generated on the basis of the current user level of the user and second display information of a preset cash withdrawal commission fee calculation mathematical model, and the first display information comprises a preset daily single cash withdrawal amount threshold corresponding to the current user level of the user. Differentdisplay pages are displayed for users with different user levels, so that personalized display is realized, and the user experience is improved. Correspondingly, the invention further provides a corresponding page display device and system, electronic equipment and a computer readable storage medium.

The invention provides an intelligent question-answering method and device, a server and a computer readable storage medium, and belongs to the technical field of artificial intelligence. When word segmentation processing is carried out, the target domain map facing the specific domain is introduced, and word segmentation processing is carried out on the questions of the original user based on the target domain map and the first dictionary facing the general domain, so that the accuracy of a word segmentation result is improved; and then words with suspected input errors in the original user questions are further determined through error detection, and error correction processing on the original user questions is performed based on the words similar to the words with the suspected input errors, thereby generating answer data according to the target user questions obtained after error correction. Therefore, the obtained answer data better conforms to the expectation of the user, the accuracy of intelligent questioning and answering is improved, and then the intelligent questioning and answering quality is ensured.

The invention discloses a method based on cloud rendering, which comprises the following steps: rendering setting: rendering configuration: a user selects a locally stored to-be-rendered file in a local program, and sets a rendering environment and parameters of the to-be-rendered file; data uploading: the to-be-rendered file and the parameter data are synchronously uploaded to a cloud rendering server by the local program; cloud rendering: the cloud rendering server performs rendering operation on the to-be-rendered file according to the uploaded parameter data; completing downloading: afterrendering of the cloud rendering server is completed, a rendered file is automatically downloaded to a local computer of the user. In the present invention, the content to be rendered is sent to a cloud rendering server. The file content to be rendered is rendered at the cloud server by utilizing high running speed of the cloud server, so that the problem that a high-price professional rendering computer needs to be configured in advance and professionals need to be arranged for operation in the traditional rendering operation is solved, the capital investment of enterprises is reduced, and meanwhile, the rendering efficiency is higher.

The invention provides a navigation path recommendation method and device and electronic equipment. The method comprises the steps that: a clustering graph generated by a plurality of first objects based on the feedback of overselection bias is obtained, wherein the clustering graph comprises a plurality of clustering sub-graphs and travel mode selection probabilities corresponding to the clustering sub-graphs; a target sub-graph associated with a target object is determined from the clustering sub-graphs; and according to the travel mode selection probability corresponding to the target sub-graph, recommendation sorting is performed on available paths. Therefore, the path recommendation better conforms to the expectation of the target object, and the use experience of the target object isimproved while the intelligent degree of path recommendation is improved.

The embodiment of the invention discloses a method and device for determining the production cycle of a commodity, electronic equipment and a storage medium, and the method comprises the steps: determining the production delay time of the commodity according to a pre-production influence factor corresponding to the commodity, wherein the pre-production influence factors comprise at least one of the following items: production scheduling waiting time and intensive processing time; determining a production starting date according to the production delay time; obtaining an estimated total production number of the commodity on the production starting date; judging whether the raw material reserve quantity of the commodity meets a triggering condition of the raw material quantity required for estimating the total production quantity or not; if the trigger condition is satisfied, determining the production cycle of the commodity according to the production delay time and the online production time of the commodity. According to the technical scheme disclosed by the embodiment of the invention, the accurate commodity production cycle is obtained, disputes caused by inaccurate production cycle labeling are avoided, meanwhile, the accurate commodity production cycle is more in line with the user's psychological expectation, and the user experience is greatly improved.

The invention discloses a multi-user Skyline query method for a time-dependent road network, and belongs to the data management and query direction in the field of computers. According to the invention, firstly, in combination with a query demand of a user for carrying out personal demand service on a real road network, Skyline query of multiple users on the time-dependent road network is designed, and a shortest path tree based on mark points is established; on this basis, a multi-user Skyline query algorithm for the time-dependent road network is provided, and the algorithm effectively meetsthe query requirements of personal demand services in an actual road network. In the algorithm implementation process, the query efficiency is further improved by adopting geometric structures such as a shortest path tree and a convex hull, and a search area is established according to the convex hull to perform efficient expansion. According to the method, the position service requirement of a real road network of a user can be met, the query efficiency is guaranteed through the shortest path tree and the convex hull structure, and the method has very high practical application value.

The embodiment of the invention provides an interaction control method and device, electronic equipment and a storage medium, and the method comprises the steps: obtaining the orientation of a virtual object in a game scene, and dividing an interaction region of the virtual object into a plurality of sub-interaction regions according to the orientation, the plurality of sub-interaction regions having region priorities corresponding to the sub-interaction regions; and determining a target object from a plurality of interaction objects according to the region priorities of the sub-interaction regions and a position relationship between the plurality of interaction objects and the virtual object in the interaction region, and providing interaction information of the target interaction object. The target interaction object can be determined based on the current orientation of the virtual object and the position relationship between the interaction object and the virtual object, the game intuition and expectation of a player are better met, meanwhile, the operation that the player selects the target interaction object can be omitted, screen coverage is reduced, and game convenience and game experience are improved.

The invention relates to a garment color segmentation extraction method, device and equipment based on mathematical statistics. The method comprises the following steps: acquiring an image; wherein the image comprises a human body image, and the human body wears a garment to be subjected to color segmentation extraction; performing human body detection on the image to detect a human body position in the image, and segmenting the human body image to obtain human body image slices; segmenting and extracting main colors of the garment on the human body image slices based on a DBSCAN clustering algorithm to obtain a color segmentation and extraction result; and outputting the color segmentation extraction result. By means of the arrangement, compared with traditional image processing methods such as image matching and principal component analysis, the method is simple in design and high in universality, the DBSCAN clustering algorithm is adopted, the number of clustering centers does not need to be preset compared with the Kmeans algorithm, and therefore the effect is more visual and stable, and the method better meets the expectation. Therefore, the scheme of the invention has very high practicability.

The invention relates to the field of business security of computer security, and discloses a risk model training method and device, computer equipment and a readable storage medium, and the method comprises the steps: obtaining long-term historical data through a long-term extraction rule, and obtaining recent historical data through a recent extraction rule; summarizing the long-term historical data and the recent historical data to obtain comprehensive training data; adjusting data volumes of long-term historical data and short-term historical data in the comprehensive training data, and adjusting a positive-negative proportion between a sum of a long-term positive sample and a short-term positive sample in the comprehensive training data and a sum of a long-term negative sample and a short-term negative sample in the comprehensive training data to obtain comprehensive adjustment data; and training the initial intelligent model through comprehensive adjustment data to obtain a mature model. The invention further relates to a block chain technology, and information can be stored in a block chain node. According to the invention, the risk identification efficiency is high, the accuracy is high, and the problems that experience and rules become vulnerabilities and normal users are identified as risk users are avoided.

The present application provides a method and apparatus for monitoring conflict of affinity rules based on load balancing. The method includes determining a type of DRS rule requested to be created bythe DRS rule creation instruction when a DRS rule creation instruction is received. According to the type of the DRS rule created by the request, the conflict between the DRS rule created by the request and the DRS rule existing in the system is monitored. In the embodiment of the present application, affinity rule management is introduced in load balancing, and affinity rules can be set by users. The premise of load balancing scheduling is that the set affinity rules must be satisfied, so that the final scheduling is more intelligent and the scheduling result is more in line with the expectations and requirements of users.