30 results about "Strategy trees" patented technology

A method for applying a strategy to a dataset in a data mining system to address a business problem, comprising: receiving at least one conditional expression defining the strategy from a user through a graphical user interface (“GUI”) displayed on a display screen of the data mining system; applying the conditional expression to the dataset to split the dataset into segments; displaying the segments as nodes in a tree structure on the display screen; receiving a calculation expression for operating on one or more values in one or more of the segments; applying the calculation expression to one or more of the segments to generate one or more respective calculated values; displaying the one or more calculated values in respective nodes of the tree structure; receiving a treatment expression for operating on the calculated values; applying the treatment expression to one or more of the calculated values to generate respective responses to the business problem; and, displaying the responses in respective nodes of the tree structure to thereby address the business problem.

The firewall method based on the high-speed DS platform comprises: a system data exchange and support system, a data access and pre-process card for kMB-network data process and transfer, and a firewall data process card. Wherein, computing HASH to the data pack for further analysis, and dynamic-load-balance sending to multiple VPN card for process; parallel running for various data for further checkup, with gradient filter mechanism and strategy tree and protocol stack strategy transfer mechanism, organizing into a increasing-strength filter system in turn by the attack detection, message sorting, IP filter, connection state monitor, user certification, APMA, APCSC, and content security check.

The invention discloses a data processing method. The method comprises the following steps that collected information data serves as log data to be stored in an information system; log data in the information system is read, the log data is classified, and a strategy tree is encapsulated by combining strategy tree cache and other cache information; alarm event information and security event information are generated according to rule configuration information and strategy tree cache strategy information; the rule configuration information is subjected to dynamic adjustment according to the alarm event information and the security event information. According to the scheme, the framework is advanced, the technology is stable, and the compatibility is high; the large data volume condition issupported; the rule system configurability is high.

The invention provides a social privacy protection method of a multi-level attribute management center based on characteristic encryption. A friend making sponsor uploads encrypted personal information files to a remote friend finder, and a certain kind of user-defined attribute characteristics and a friend making access control strategy are uploaded to the multi-level attribute management center; the multi-level attribute management center generates ciphertext according to the attribute characteristics and is associated with an access strategy tree corresponding to attributes; a friend making requester requests for checking friend making information of other users in the friend finder, when an attribute set of the friend making requester is matched with the access strategy tree of the friend making sponsor, the multi-level attribute management center sends a secrete key corresponding to the attribute of the friend making sponsor to the friend making requester, and the friend making requester decrypts data ciphertext of an information owner of the friend finder. The method solves the single-point fault and performance bottleneck problems of a single-authorization center, and also solves the problems that an intelligent terminal does not depend on an authorization center and accordingly monitoring difficulty is high and application scenes are limited.

The invention provides an FPGA (Field Programmable Gate Array) based high-speed network strategy matching method. The method realizes utilization of relatively small storage space, and simultaneously is suitable for fast strategy matching processing in hardware, and further enables the performance of strategy matching processing to be irrelevant with the data volume of strategies. The FPGA based high-speed network strategy matching method employs a quintuple, supports accurate matching of the quintuple and IP-port range matching , and is stronger in functions and more diversified in matching conditions and matching modes than the traditional pure hardware manner; the strategy tree of the FPGA based high-speed network strategy matching method is a specific tree structure with a node length fixed at 512 bytes and the height of the tree layer fixed at 13, and aims at FPGA strategy matching; the logic processing of direct address positioning is simple and efficient, so that consumption of lots of time on the logical operation is avoided and the matching efficiency is greatly improved.

It is a message division filter and management method based on strategy tree, which comprises the following steps: first to align strategy to users and pre-edit strategy tree; to generate strategy tree memory image by pre-edit files; to get network card subarea property through data pack pre-process; then to inject data pack into strategy tree analyze engineer to get next operation of data pack and its relative property.

A feature extraction and feature selection method oriented to background multi-source data is characterized by including the following steps that first, background data of a plurality of months are divided into a training set and a testing set; second, corresponding grouping features are extracted on the training set according to different source data; third, feature group selection is performed on the testing set through cross validation with a Group Lasso method. The feature extraction and feature selection method oriented to background multi-source data has the advantages that a C45 strategy tree is used for the selected group feature so that a classifier off-network user analysis classifier can be established, the prediction accuracy on off-network users reaches 45%, and the prediction accuracy on downtime users with the off-network tendency reaches 88%.

This invention discloses an information index method in a relation-type database including the following steps: structuring a dimensionality reduction strategy tree to be set in a database system, in which, said tree includes at least a sub-node and a root node, each sub-node includes the identification information of the local node number and enquiry condition combination information and lower level sub-node number, when searching said database according to user enquiry condition but does not get the necessary data, it structures new enquiry conditions based on the strategy tree, searching the database according to a new enquiry condition till getting the necessary data or enquiring the root node of the tree to feed back information of the undesired data. This invention discloses an information index device in a relation-type database including: an enquiry condition obtaining unit, an enquiry condition output unit, a strategy tree storage unit and an operation unit.

The invention relates to a NOT operational character supported characteristic-based CP-ABE method having CCA security. An access control structure is an access control tree; a NOT node is added in the access control tree; and k is equal to -1. Meanwhile, according to regulation, a father node of the NOT node must be an ''AND'' node and only one intermediate node is hung under the NOT node; and thus the intermediate node is used as a root node to set a strategy tree and the strategy tree expresses a related strategy, set by an encryption party, of a NOT attribute. A high one-time signature technology is added to further enhance a security level of the method from a CPA security level to a CCA security level. Strategy expression based on an attribute encryption algorithm is enriched; security of the existing method is enhanced, thereby building the high access controlling capability; and moreover, the method has an encryption method with a provable security.

The invention discloses a rapid fine-grained multi-domain network interconnection security control method, and belongs to the field of network space security. The method comprises: firstly, security policy language grammar specifications are established, each security policy is converted into a normal form script, a corresponding grammar parser is achieved, and the scripts without errors are stored; designing each strategy in a white list into a tree for scripts in which domains and semantics of services exist in the network, and merging the trees; designing each strategy in the blacklist intoa tree, and merging the trees into a white list; and designing each strategy in the range default action into a tree, and merging the tree with the black and white list merging tree to obtain a security strategy tree; and finally, converting the script describing inter-domain service communication into a seven-tuple security rule describing fine granularity, distributing/transmitting the seven-tuple security rule to a security Internet gateway at a high speed, and updating security control information of an execution unit according to the security rule. The method has the advantages of simplicity, convenience, flexibility and higher efficiency.

The invention discloses a vehicle machine non-perception upgrading system and method, and relates to the technical field of vehicle machines. The system comprises a server and a terminal upgrading center, and the terminal upgrading center is in wireless communication connection with the server. The server generates an application upgrading strategy tree according to the dependency level of the to-be-upgraded application; the terminal upgrading center periodically accesses the server in a preset time period; the terminal upgrading center obtains an application upgrading strategy tree; and the terminal upgrading center performs grading and upgrading on the application according to the application upgrading strategy. The distribution upgrading strategy and the application distribution are generated at the server side, the terminal upgrading center performs application non-perceptive upgrading according to the upgrading strategy, the flow consumed by upgrading is less, no vehicle owner operation is needed in the whole upgrading process, and the safety is higher.

The invention provides a locomotive intelligent operation optimization calculation method. The locomotive intelligent operation optimization calculation method comprises steps that step 1, information required by locomotive intelligent operation optimization is preprocessed; step 2, an initial optimization curve is generated; step 3, optimization adjustment is carried out based on speed limit; step 4, optimization adjustment is carried out based on time deviation; step 5, the optimization curve is adjusted based on a requirement on operation steady safe shifting, and a final optimization strategy is generated. The strategy of every step is affected by locomotive driving strategy influencing parameters, which are matched with different strategies under different conditions, and are used as strategy classification attribute for strategy classification. The strategy of every step can be used to form a strategy tree, the leaf node of which is the matched final optimization strategy according to the classification attribute. Every step of the whole scheme is capable of satisfying the schedule requirement, the speed limit requirement, the time table requirement, the steady operation requirement, the energy saving requirement, and other requirements of the locomotive operation, and optimization effect and integrated operation efficiency are guaranteed.

The invention discloses a 1553B bus message distribution configuration unit based on data characteristics. The 1553B bus message distribution configuration unit comprises a bus information mapping module, a data characteristic calibration module and a data distribution batch processing module. The bus information mapping module maps the bus address, the sub-address, the transceiving state and themessage data word length of the received message to form an instruction word; The data feature calibration module constructs a distribution strategy tree based on the data processing distribution requirement to form a strategy code, and obtains a mapping relation (a data distribution query table) between the instruction word and the strategy code; And the data distribution batch processing modulesearches the strategy code according to the instruction word of the message, and performs a message processing flow according to the strategy code to realize data processing and distribution. According to the invention, the data processing program is simplified, the bus message distribution efficiency is greatly improved, the processing program does not need to be changed when the demand changes,and the expansion is easy.

The invention belongs to the technical field of information retrieval, and relates to a query rewriting method of a database. According to the method, firstly, a strategy tree is used for representing a large number of equivalent queries of any query statement and a corresponding rewriting sequence, secondly, a high-yield rewriting sequence is selected on the strategy tree based on a Monte Carlo tree search algorithm, and balance of query overhead and rewriting frequency is achieved; in order to more accurately estimate the overall income of the query after rewriting, the deep attention network is utilized to learn the incidence relation between the current query and the rewriting rule and the access data, and deep learning is utilized to fit the overall income of the current query according to the query features and the extracted incidence relation. And finally, aiming at a complex query with a large-scale strategy tree, the invention provides a parallel node selection algorithm, an optimal algorithm is utilized on the strategy tree to select a plurality of nodes which have no ancestor-descendant relationship and have the highest total income, and the strategy tree is expanded in parallel, so that the query rewriting efficiency is greatly improved.

The invention provides an XACML-based access control policy conflict detection method. The method comprises the following steps: 1, establishing a policy tree based on rules, policies and a policy set; 2, establishing an index structure according to the established strategy tree and the effect of the rule; 3, acquiring rules in the to-be-detected strategy, and comparing the rules with rules which are in different strategy trees and have opposite effects one by one; and 4, if the comparison result is that the rule pair has intersection, indicating that the rule pair conflicts, storing the rule pair in the queue to be output, repeating the steps 3-4 until all rules in the strategy to be detected are compared, and outputting all rule pairs in the queue to be output. According to the scheme provided by the invention, a brand new strategy index structure is adopted, conflict detection of complex strategies can be realized based on expression tree comparison, and reasons for conflict generation can be marked. A user can flexibly configure and eliminate strategy conflicts according to conflict marks, and efficient and accurate strategy management and control are achieved.