CN116186020B - Feature information processing method, device, electronic equipment and computer readable medium - Google Patents

Feature information processing method, device, electronic equipment and computer readable medium Download PDF

Info

Publication number
CN116186020B
CN116186020B CN202310479432.4A CN202310479432A CN116186020B CN 116186020 B CN116186020 B CN 116186020B CN 202310479432 A CN202310479432 A CN 202310479432A CN 116186020 B CN116186020 B CN 116186020B
Authority
CN
China
Prior art keywords
feature
information
target
feature matrix
matrix
Prior art date
Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
Active
Application number
CN202310479432.4A
Other languages
Chinese (zh)
Other versions
CN116186020A (en
Inventor
贾智杰
张耘菡
侯敏
肖法鲁
胡浩楠
王渊
方兴
李学寿
石鑫
潘禹辛
Current Assignee (The listed assignees may be inaccurate. Google has not performed a legal analysis and makes no representation or warranty as to the accuracy of the list.)
Citic Securities Co ltd
Original Assignee
Citic Securities Co ltd
Priority date (The priority date is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the date listed.)
Filing date
Publication date
Application filed by Citic Securities Co ltd filed Critical Citic Securities Co ltd
Priority to CN202310479432.4A priority Critical patent/CN116186020B/en
Publication of CN116186020A publication Critical patent/CN116186020A/en
Application granted granted Critical
Publication of CN116186020B publication Critical patent/CN116186020B/en
Active legal-status Critical Current
Anticipated expiration legal-status Critical

Links

Images

Classifications

    • GPHYSICS
    • G06COMPUTING; CALCULATING OR COUNTING
    • G06FELECTRIC DIGITAL DATA PROCESSING
    • G06F16/00Information retrieval; Database structures therefor; File system structures therefor
    • G06F16/20Information retrieval; Database structures therefor; File system structures therefor of structured data, e.g. relational data
    • G06F16/21Design, administration or maintenance of databases
    • G06F16/215Improving data quality; Data cleansing, e.g. de-duplication, removing invalid entries or correcting typographical errors
    • GPHYSICS
    • G06COMPUTING; CALCULATING OR COUNTING
    • G06FELECTRIC DIGITAL DATA PROCESSING
    • G06F12/00Accessing, addressing or allocating within memory systems or architectures
    • G06F12/02Addressing or allocation; Relocation
    • G06F12/08Addressing or allocation; Relocation in hierarchically structured memory systems, e.g. virtual memory systems
    • G06F12/0802Addressing of a memory level in which the access to the desired data or data block requires associative addressing means, e.g. caches
    • G06F12/0877Cache access modes
    • GPHYSICS
    • G06COMPUTING; CALCULATING OR COUNTING
    • G06FELECTRIC DIGITAL DATA PROCESSING
    • G06F16/00Information retrieval; Database structures therefor; File system structures therefor
    • G06F16/20Information retrieval; Database structures therefor; File system structures therefor of structured data, e.g. relational data
    • G06F16/22Indexing; Data structures therefor; Storage structures
    • G06F16/2228Indexing structures
    • GPHYSICS
    • G06COMPUTING; CALCULATING OR COUNTING
    • G06FELECTRIC DIGITAL DATA PROCESSING
    • G06F17/00Digital computing or data processing equipment or methods, specially adapted for specific functions
    • G06F17/10Complex mathematical operations
    • G06F17/16Matrix or vector computation, e.g. matrix-matrix or matrix-vector multiplication, matrix factorization
    • GPHYSICS
    • G06COMPUTING; CALCULATING OR COUNTING
    • G06QINFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES; SYSTEMS OR METHODS SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES, NOT OTHERWISE PROVIDED FOR
    • G06Q40/00Finance; Insurance; Tax strategies; Processing of corporate or income taxes
    • G06Q40/04Trading; Exchange, e.g. stocks, commodities, derivatives or currency exchange
    • YGENERAL TAGGING OF NEW TECHNOLOGICAL DEVELOPMENTS; GENERAL TAGGING OF CROSS-SECTIONAL TECHNOLOGIES SPANNING OVER SEVERAL SECTIONS OF THE IPC; TECHNICAL SUBJECTS COVERED BY FORMER USPC CROSS-REFERENCE ART COLLECTIONS [XRACs] AND DIGESTS
    • Y02TECHNOLOGIES OR APPLICATIONS FOR MITIGATION OR ADAPTATION AGAINST CLIMATE CHANGE
    • Y02DCLIMATE CHANGE MITIGATION TECHNOLOGIES IN INFORMATION AND COMMUNICATION TECHNOLOGIES [ICT], I.E. INFORMATION AND COMMUNICATION TECHNOLOGIES AIMING AT THE REDUCTION OF THEIR OWN ENERGY USE
    • Y02D10/00Energy efficient computing, e.g. low power processors, power management or thermal management

Abstract

Embodiments of the present disclosure disclose a feature information processing method, apparatus, electronic device, and computer readable medium. One embodiment of the method comprises the following steps: determining a target feature matrix corresponding to the target virtual article according to the article index label and the feature construction time period corresponding to the target virtual article; determining an object circulation feature matrix of the target virtual object in a feature construction time period; performing feature association on the target feature matrix and the object circulation feature matrix to obtain an associated feature matrix; determining strategy information corresponding to a target virtual object in a pre-constructed strategy information tree as target strategy information to obtain a target strategy information set; generating evaluation information aiming at a target virtual object according to the associated feature matrix and the target strategy information set; and caching the evaluation information and the associated feature matrix into a cache. The embodiment reduces the occupation of the cache resources and the waste of the cache resources.

Description

Feature information processing method, device, electronic equipment and computer readable medium
Technical Field
Embodiments of the present disclosure relate to the field of computer technology, and in particular, to a feature information processing method, a feature information processing device, an electronic device, and a computer readable medium.
Background
For virtual articles (e.g., stocks, bonds, etc.), the size of the corresponding data is gradually increasing as related businesses are continuously developed. How to convert the general data into standardized data features is particularly important for the analytical evaluation of virtual articles. Currently, in data feature processing, the following methods are generally adopted: and loading all the data into a cache, and carrying out data characteristic combing in a manual mode.
However, the inventors found that when the above manner is adopted, there are often the following technical problems:
firstly, data often has certain data redundancy, a mode of loading the whole data into a cache is adopted, occupied cache resources are increased along with the increase of the volume of the data, and meanwhile, the redundant data also causes the waste of the cache resources;
secondly, the collection frequency is often required to be set for data feature extraction, and when the collection frequency is higher, the volume of data is extremely large, so that more storage resources are occupied;
third, when virtual item recommendation is performed, risk resistance of an audience is often not combined, so that accurate recommendation cannot be performed on the virtual item, and recommendation flow is effectively distributed.
The above information disclosed in this background section is only for enhancement of understanding of the background of the inventive concept and, therefore, may contain information that does not form the prior art that is already known to those of ordinary skill in the art in this country.
Disclosure of Invention
The disclosure is in part intended to introduce concepts in a simplified form that are further described below in the detailed description. The disclosure is not intended to identify key features or essential features of the claimed subject matter, nor is it intended to be used to limit the scope of the claimed subject matter.
Some embodiments of the present disclosure propose feature information processing methods, apparatuses, electronic devices, and computer-readable media to solve one or more of the technical problems mentioned in the background section above.
In a first aspect, some embodiments of the present disclosure provide a feature information processing method, the method including: determining a target feature matrix corresponding to a target virtual article according to an article index label and a feature construction time period corresponding to the target virtual article, wherein the target feature matrix comprises a plurality of feature vectors changing with time; determining an object circulation feature matrix of the target virtual object in the feature construction time period; performing feature association on the target feature matrix and the object circulation feature matrix to obtain an associated feature matrix; determining strategy information corresponding to the target virtual object in a pre-constructed strategy information tree as target strategy information to obtain a target strategy information set; generating evaluation information for the target virtual article according to the associated feature matrix and the target strategy information set, wherein the evaluation information comprises: first and second evaluation information; and caching the evaluation information and the associated feature matrix into a cache.
In a second aspect, some embodiments of the present disclosure provide a feature information processing apparatus, the apparatus including: a first determining unit configured to determine a target feature matrix corresponding to a target virtual article according to an article index tag and a feature construction time period corresponding to the target virtual article, wherein the target feature matrix includes a plurality of feature vectors that change with time; a second determining unit configured to determine an object circulation feature matrix of the target virtual object in the feature construction period; the feature association unit is configured to perform feature association on the target feature matrix and the object circulation feature matrix to obtain an associated feature matrix; a third determining unit configured to determine policy information corresponding to the target virtual article in a policy information tree constructed in advance as target policy information, and obtain a target policy information set; a generating unit configured to generate evaluation information for the target virtual article according to the associated feature matrix and the target policy information set, wherein the evaluation information includes: first and second evaluation information; and the caching unit is configured to cache the evaluation information and the associated feature matrix into a cache.
In a third aspect, some embodiments of the present disclosure provide an electronic device comprising: one or more processors; a storage device having one or more programs stored thereon, which when executed by one or more processors causes the one or more processors to implement the method described in any of the implementations of the first aspect above.
In a fourth aspect, some embodiments of the present disclosure provide a computer readable medium having a computer program stored thereon, wherein the program, when executed by a processor, implements the method described in any of the implementations of the first aspect above.
The above embodiments of the present disclosure have the following advantageous effects: by the feature information processing method of some embodiments of the present disclosure, occupation of the redundant data to the cache resource is reduced, and use efficiency of the cache resource is improved. Specifically, the reason why there is waste of cache resources is that: the data often has certain data redundancy, a mode of loading the whole data into the cache is adopted, the occupied cache resources can be increased along with the increase of the volume of the data, and meanwhile, the waste of the cache resources can be caused by the redundant data. Based on this, in the feature information processing method of some embodiments of the present disclosure, first, a target feature matrix corresponding to a target virtual article is determined according to an article index tag and a feature construction time period corresponding to the target virtual article, where the target feature matrix includes a plurality of feature vectors that change with time. In practice, since the number of virtual products is numerous, and the feature data corresponding to the virtual products is also increasing with time, it is necessary to obtain the feature data of the target virtual product in the feature construction time period according to the index tag and the feature construction time period corresponding to the virtual article. The volume of the cache resources is primarily reduced, so that occupation of subsequent cache resources is reduced. And secondly, determining an object circulation feature matrix of the target virtual object in the feature construction time period. In practice, virtual products often correspond to valuable related operations, and therefore, it is desirable to determine an item flow feature matrix over a feature build period. Meanwhile, the data volume of the object circulation feature is restrained through the feature construction time period, so that occupation of subsequent cache resources is further reduced. And then, carrying out feature association on the target feature matrix and the object circulation feature matrix to obtain an associated feature matrix. Through feature association, feature matrix combination is realized, so that the quantity of redundant information is reduced. Further, determining policy information corresponding to the target virtual object in a pre-constructed policy information tree as target policy information, and obtaining a target policy information set. In practice, different virtual articles often correspond to different policy information, and therefore, the policy information corresponding to the target virtual article needs to be acquired. Then, generating evaluation information for the target virtual article according to the associated feature matrix and the target strategy information set, wherein the evaluation information comprises the following components: first evaluation information and second evaluation information. Thereby, evaluation information for the target article is generated. And finally, caching the evaluation information and the associated feature matrix into a cache. In practice, because the evaluation information is often corresponding to a higher access frequency, the evaluation information is loaded into the cache, and the time consumption from the reading of the evaluation information from the external memory to the cache is avoided. Meanwhile, compared with the full data, the feature matrix after correlation is the feature matrix subjected to redundant information rejection, and the volume of the data is far smaller than that of the full data, so that the occupation of cache resources and the waste of the cache resources can be greatly reduced.
Drawings
The above and other features, advantages, and aspects of embodiments of the present disclosure will become more apparent by reference to the following detailed description when taken in conjunction with the accompanying drawings. The same or similar reference numbers will be used throughout the drawings to refer to the same or like elements. It should be understood that the figures are schematic and that elements and components are not necessarily drawn to scale.
FIG. 1 is a flow chart of some embodiments of a feature information processing method according to the present disclosure;
FIG. 2 is a schematic tree structure of a policy information tree;
FIG. 3 is a schematic diagram of the structure of some embodiments of a feature information processing apparatus according to the present disclosure;
fig. 4 is a schematic structural diagram of an electronic device suitable for use in implementing some embodiments of the present disclosure.
Detailed Description
Embodiments of the present disclosure will be described in more detail below with reference to the accompanying drawings. While certain embodiments of the present disclosure are shown in the drawings, it should be understood that the present disclosure may be embodied in various forms and should not be construed as limited to the embodiments set forth herein. Rather, these embodiments are provided so that this disclosure will be thorough and complete. It should be understood that the drawings and embodiments of the present disclosure are for illustration purposes only and are not intended to limit the scope of the present disclosure.
It should be noted that, for convenience of description, only the portions related to the present invention are shown in the drawings. Embodiments of the present disclosure and features of embodiments may be combined with each other without conflict.
It should be noted that the terms "first," "second," and the like in this disclosure are merely used to distinguish between different devices, modules, or units and are not used to define an order or interdependence of functions performed by the devices, modules, or units.
It should be noted that references to "one", "a plurality" and "a plurality" in this disclosure are intended to be illustrative rather than limiting, and those of ordinary skill in the art will appreciate that "one or more" is intended to be understood as "one or more" unless the context clearly indicates otherwise.
The names of messages or information interacted between the various devices in the embodiments of the present disclosure are for illustrative purposes only and are not intended to limit the scope of such messages or information.
The present disclosure will be described in detail below with reference to the accompanying drawings in conjunction with embodiments.
Referring to fig. 1, a flow 100 of some embodiments of a feature information processing method according to the present disclosure is shown. The characteristic information processing method comprises the following steps:
Step 101, determining a target feature matrix corresponding to the target virtual article according to the article index label and the feature construction time period corresponding to the target virtual article.
In some embodiments, an execution subject (e.g., a computing device) of the feature information processing method may determine a target feature matrix corresponding to the target virtual item according to the item index tag and the feature build time period corresponding to the target virtual item. Wherein the target virtual item may be a virtual item that may perform value-related operations. In practice, the target virtual item may be a financial product. For example, the target virtual article may be, but is not limited to, any of the following: stocks, bonds, funds, futures, options. For example, the value-related operation may be a buy operation. The value related operation may also be a sell operation. The item index tag may be an index tag for identifying the uniqueness of the virtual item. In practice, the item index tag may be composed of virtual item numbers and value-related operation executable area codes. In particular, the value-related operations executable area code may characterize a trading market code of the virtual product. For example, the item index tag may be "600030.SH," where "600030" is a virtual item number. "SH" is a value-related operation executable area code. The feature build time period may be a feature intercept time period. For example, the feature build time period may include: start time and end time. As another example, the feature build time period may be composed of a time series. For example, the feature build time period may be [20230101,20230102,20230103]. The target feature matrix may include a plurality of feature vectors that vary over time. In particular, the target feature matrix may characterize a plurality of feature vectors of the target virtual article over a feature build time period. The feature vector may be composed of the underlying attributes of the target virtual item. In practice, the base attributes corresponding to the target virtual item may include, but are not limited to, at least one of: data acquisition time, item index tag, virtual item name, issuer name, virtual item type, trade market name, market rating, virtual item valuation, virtual item equity, virtual item stock rate. For example, the feature vector may be [ "1681797326.0126445", "600030.SH", "600030", "XX institution", "stock", "XX exchange", "A-stock", "19.9", "1.23", "37.02" ].
As an example, first, the execution subject may retrieve all feature vectors corresponding to the target virtual article according to the article index tag. Then, when the feature build period includes: and when the starting time and the ending time are the starting time and the ending time, the execution main body can take the starting time to the ending time as a time range, select the corresponding feature vector of which the data acquisition time is in the feature construction time range from all the feature vectors corresponding to the target virtual object, and form the target feature matrix according to the sequence of the data acquisition time. When the feature construction time period may be formed by a time sequence, the execution body may sequentially determine feature vectors corresponding to each time in the time sequence, and form the target feature matrix according to a sequence of data acquisition times.
The computing device may be hardware or software. When the computing device is hardware, the computing device may be implemented as a distributed cluster formed by a plurality of servers or terminal devices, or may be implemented as a single server or a single terminal device. When the computing device is embodied as software, it may be installed in the hardware devices listed above. It may be implemented as a plurality of software or software modules, for example, for providing distributed services, or as a single software or software module. The present invention is not particularly limited herein. It should be appreciated that the number of computing devices may have any number, as desired for implementation.
In some optional implementations of some embodiments, the determining the target feature matrix corresponding to the target virtual article according to the article index tag and the feature construction time period corresponding to the target virtual article may include the following steps:
and determining a feature vector corresponding to the target virtual object in a pre-constructed virtual object feature matrix according to the object index label, and obtaining a candidate feature vector sequence by taking the feature vector as a candidate feature vector.
Wherein the virtual article feature matrix may include a plurality of feature vectors of a plurality of virtual articles over time. In practice, the execution subject may use the article index tag as a search term, and search the feature vector corresponding to the target virtual article from the virtual article feature matrix, as a candidate feature vector, to obtain a candidate feature vector sequence. Wherein the candidate feature vectors in the candidate feature vector sequence are ordered.
And secondly, according to the characteristic construction time period, vector interval interception is carried out on the candidate characteristic vector sequence, and an intercepted candidate characteristic vector sequence is obtained.
As an example, the candidate feature vector sequence may be [ candidate feature vector a, candidate feature vector B, candidate feature vector C, candidate feature vector D ]. The data acquisition time corresponding to the candidate feature vector a may be "20230101". The data acquisition time corresponding to the candidate feature vector B may be "20230102". The data acquisition time corresponding to the candidate feature vector C may be "20230103". The data acquisition time corresponding to the candidate feature vector D may be "20230104". The feature build time period may be [20230102, 20230104]. Thus, the truncated candidate feature vector sequence may be [ candidate feature vector B, candidate feature vector C, candidate feature vector D ].
And thirdly, generating the target feature matrix according to the intercepted candidate feature vector sequence.
The execution body may matrix the truncated candidate feature vector in the truncated candidate feature vector sequence according to the sequence of the data acquisition time corresponding to the truncated candidate feature vector in the truncated candidate feature vector sequence, so as to obtain the target feature matrix.
As an example, the truncated candidate feature vector sequence may be [ candidate feature vector B, candidate feature vector C, candidate feature vector D ]. The target feature matrix may be [ [ candidate feature vector B ], [ candidate feature vector C ], [ candidate feature vector D ] ].
Step 102, determining an object circulation feature matrix of the target virtual object in the feature construction time period.
In some embodiments, the executing entity may determine an item circulation feature matrix of the target virtual item within the feature construction period. Wherein the item flow feature matrix may characterize a plurality of value-related operational feature vectors of the target virtual item over time. The value-related operational feature vector may be composed of value-related operational base attributes of the target virtual item. In practice, the value-related operational base attributes may include, but are not limited to, at least one of: value-related operations execution time, item index tag, issuer name, settlement subject name, transaction account number, circulation identification, virtual item holding status, settlement currency, quantity, cost, and profit-loss value.
As an example, first, the execution subject may determine attribute values of basic attributes of value-related operations when the target virtual article performs the value-related operations in the feature construction period, so as to obtain a plurality of value-related operation feature vectors, and construct the article circulation feature matrix according to the sequence of the value-related operation execution times included in the value-related operation feature vectors.
And 103, performing feature association on the target feature matrix and the object circulation feature matrix to obtain an associated feature matrix.
In some embodiments, the executing body may perform feature association on the target feature matrix and the object circulation feature matrix to obtain an associated feature matrix.
As an example, first, the execution subject may splice feature vectors with the same data acquisition time and the same value-related operation execution time with value-related operation feature vectors to obtain a candidate associated feature matrix. Then, the execution subject may perform attribute value deduplication on the attribute value corresponding to the basic attribute of the value-related operation in the candidate associated feature matrix and the attribute value corresponding to the basic attribute, so as to generate the associated feature matrix.
In some optional implementations of some embodiments, the performing body performs feature association on the target feature matrix and the object circulation feature matrix to obtain an associated feature matrix, and may include the following steps:
and under the condition of time alignment, determining the Cartesian product of the target feature matrix and the object circulation feature matrix as the post-correlation feature matrix.
As an example, the target feature matrix may be matrix a. The item flow feature matrix may be matrix B. The post-correlation feature matrix may be matrix C. Where matrix c=matrix a×matrix B.
And 104, determining strategy information corresponding to the target virtual article in a pre-constructed strategy information tree as target strategy information to obtain a target strategy information set.
In some embodiments, the executing body may determine policy information corresponding to the target virtual article in the pre-constructed policy information tree, as target policy information, to obtain a target policy information set. Wherein the policy information tree may include a plurality of policy information. The execution subject can determine the position of the policy information in the policy information tree according to the membership or execution sequence of the policy information. The policy information may characterize an execution policy of the virtual article when performing the value-related operation. In particular, the policy information may include at least one value-related operation for at least one virtual item for the purpose of the value-related operation. For example, the target policy information may include: and (5) turning the stock to obtain a arbitrage strategy.
As an example, the executing body may perform policy information retrieval on the policy information tree according to the item index identifier, so as to obtain the target policy information set.
Optionally, the policy information tree is a multi-way tree. The tree nodes in the policy information tree may include: policy information and parent node identification. Wherein the parent node identification characterizes a node identification of a parent node of the tree node. The information types corresponding to the target policy information in the target policy information set include: a first information type and a second information type. Wherein the first information type characterizes the target virtual article as being directly applicable to the policy information. The second information type characterizes the type of policy information included by the tree nodes on the path between the tree nodes of the first information type and the root node.
In some optional implementations of some embodiments, the executing body determines policy information corresponding to the target virtual object in a pre-built policy information tree, and obtains a target policy information set as target policy information, which may include the following steps:
and a first step of determining a set of determination conditions corresponding to the target virtual article.
The determination condition set may include a preset determination condition for determining whether the policy information is applicable to the target virtual article. For example, the execution subject may read, from the configuration file, a set of determination conditions corresponding to the target virtual article. For another example, the executing body may further obtain, in real time, a set of judgment conditions configured by the user in real time for the target virtual article.
And a second step of determining, for each policy information in the policy information tree, the policy information as target policy information in response to determining that the policy information satisfies the set of determination conditions.
The information type of the policy information satisfying the above-mentioned determination condition set is a first information type.
As an example, a tree structure diagram of the policy information tree shown in fig. 2 is shown, wherein the policy information tree shown in fig. 2 includes: policy information a, policy information B, policy information C, policy information D, policy information E, policy information F, policy information G, policy information H, policy information I, and policy information J. The tree node corresponding to the policy information A is a root node. The nodes corresponding to the strategy information A are tree nodes corresponding to the strategy information B, tree nodes corresponding to the strategy information C, tree nodes corresponding to the strategy information D and father nodes of tree nodes corresponding to the strategy information E. The tree node corresponding to the policy information C is the father node of the tree node corresponding to the policy information F. The tree nodes corresponding to the policy information F are tree nodes corresponding to the policy information G, tree nodes corresponding to the policy information H and father nodes of the tree nodes corresponding to the policy information I. The tree node corresponding to the policy information H is the father node corresponding to the tree node of the policy information J. The policy information J may satisfy the above-described determination condition set, and thus, the target policy information may be the policy information J, that is, the information type of the policy information J is the first information type.
And secondly, taking a tree node corresponding to the target strategy information with the information type of the first information type in the target strategy information set as a starting node and a root node of the strategy information tree as an ending node, and carrying out recursion backtracking on the strategy information tree according to a father node identifier included in the tree node in the strategy information tree to obtain the target strategy information.
And the information type of the strategy information included by the tree nodes on the path corresponding to the starting node to the ending node is a second information type. In practice, the execution subject can carry out recursive backtracking on the strategy information tree through a recursive algorithm to obtain target strategy information.
As an example, with further reference to fig. 2, the tree node to which policy information J corresponds may be the originating node. The tree node to which policy information a corresponds may be an end node. The tree path from the start node to the end node may include: policy information a, policy information C, policy information F, policy information H, policy information J. The information types of the policy information A, the policy information C, the policy information F and the policy information H are second information types.
And 105, generating evaluation information aiming at the target virtual article according to the associated feature matrix and the target strategy information set.
In some embodiments, the executing entity may generate the evaluation information for the target virtual article according to the associated feature matrix and the target policy information set. Wherein, the evaluation information may include: first evaluation information and second evaluation information. The first evaluation information may characterize a feature evaluation result of the target virtual article over a feature build period. The second evaluation information may represent a policy evaluation result of the target policy information corresponding to the target virtual article in the feature construction period.
As an example, the executing entity may determine, according to the post-association feature matrix, a total profit of the asset feature of the target virtual object in the feature construction period. And when the total profit of the asset characteristics is greater than the total profit threshold of the asset characteristics, generating first evaluation information of forward evaluation. And when the total profit of the asset characteristics is smaller than or equal to the total profit threshold of the asset characteristics, generating first evaluation information of negative evaluation.
As yet another example, for each target policy information in the target policy information set, the above-described enforcement agent may determine a rate of return of the target policy information, and when the rate of return is greater than a reference rate of return, generate second rating information for the forward rating. And when the yield is smaller than or equal to the reference yield, generating second evaluation information of negative evaluation. Specifically, the executing body may further calculate a tracking error, a summer ratio, an information ratio, and the like corresponding to the target policy information, so as to generate second evaluation information for evaluating the target policy thank you.
In some optional implementations of some embodiments, the generating, by the executing entity, evaluation information for the target virtual object according to the post-association feature matrix and the target policy information set may include the following steps:
and determining the value attribute net value of the target virtual object in the feature construction time period according to the associated feature matrix.
In practice, the net value of the value attribute may characterize the net profit value of the target virtual item over the feature build period. For example, the executing entity may determine a sum of a plurality of profit and loss values included in the post-correlation feature matrix as a net value of the value attribute.
And a second step of determining the first evaluation information according to the net value of the value attribute.
As an example, when the net value of the value attribute is greater than a net value threshold value of the value attribute, first rating information of the forward rating is generated. And when the value attribute net value is smaller than or equal to the value attribute net value threshold value, generating first evaluation information of forward evaluation.
Third, for each target policy information in the target policy information set, the following processing steps are performed:
and a first sub-step of determining a reference evaluation curve corresponding to the target strategy information.
For example, the benchmark evaluation curve may characterize a benchmark performance curve corresponding to the trading market.
And a second sub-step of generating a characteristic curve aiming at the target evaluation information according to the correlated characteristic matrix.
For example, the execution subject may determine, as the feature curve, an actual performance curve after executing the target policy information policy based on the post-association feature matrix.
And a third sub-step of generating second evaluation information corresponding to the target policy information according to the reference evaluation curve and the characteristic curve.
For example, when the characteristic curve is located above the reference evaluation curve, second evaluation information of the forward evaluation is generated. And generating second evaluation information of negative evaluation when the characteristic curve is positioned below the reference evaluation curve.
And step 106, caching the evaluation information and the associated feature matrix into a cache.
In some embodiments, the execution body may cache the evaluation information and the associated feature matrix in a cache. The cache may be a memory included in the execution body.
Optionally, the method further comprises:
and firstly, generating an article circulation characteristic curve aiming at the correlated characteristic matrix.
The article circulation characteristic curve can represent a profit and loss value curve of the target virtual article in the characteristic construction time period. In practice, the horizontal axis of the item flow characteristic corresponds to time. The vertical axis corresponds to the profit and loss value.
And a second step of determining the curve slope of each feature construction sampling point included in the feature construction time period of the object circulation feature curve to obtain a curve slope sequence.
The feature construction sampling points correspond to time points included in a feature construction time period. For example, the feature build time period may be [20230101, 20230102, 20230103, 20230104]. The execution body may determine a slope of the curve corresponding to the time point "20230101" in the article transfer characteristic curve, a slope of the curve corresponding to the time point "20230102" in the article transfer characteristic curve, a slope of the curve corresponding to the time point "20230103" in the article transfer characteristic curve, and a slope of the curve corresponding to the time point "20230104" in the article transfer characteristic curve, to obtain the curve slope sequence, where the curve slope sequence is ordered.
Thirdly, performing matrix segmentation on the associated feature matrix through a segmentation model and the curve slope sequence, which are included in a pre-trained target prediction model, so as to obtain a segmented feature matrix sequence.
The segmentation model is used for determining segmentation of the associated feature matrix. In practice, the segmentation model may be one comprising: AE (Auto Encoder) model and discriminant model. Wherein the discriminant model may be a classification model. The classification result of the discrimination model may include: and compressing the classification result by the features and expanding the classification result by the features. The feature matrix sequence after segmentation comprises: a segmented feature matrix set to be feature compressed and a segmented feature matrix set to be feature extended.
As an example, first, the execution subject may cluster the slope of the curve in the sequence of slopes of the curve by dividing the model. For example, the curve slope sequence may be [0.1,0.11,0.4,0.7]. Then "0.1" and "0.11" correspond to one class center. "0.4" corresponds to a class center. "0.7" corresponds to a class center. Wherein, class labels corresponding to 0.1 and 0.11 are characteristic compression classification results. The class label corresponding to "0.4" is the feature extended classification result. The class label corresponding to "0.7" is the feature extended classification result. Therefore, the execution body may determine the sub-matrix corresponding to the time interval [20230101, 20230102] as the segmented feature matrix to be feature compressed. And determining the submatrix corresponding to the time [20230103] as a segmented feature matrix to be subjected to feature expansion. And determining the submatrix corresponding to the time [20230104] as a segmented feature matrix to be subjected to feature expansion.
And fourthly, compressing the feature matrix of each segmented feature matrix to be subjected to feature compression in the segmented feature matrix group to be subjected to feature compression through a feature compression model included in the target prediction model so as to generate a compressed feature matrix, and obtaining a compressed feature matrix group.
In practice, the feature compression model may be a FPN (Feature Pyramid Networks, feature pyramid network) model.
And fifthly, performing feature matrix expansion on each feature matrix after feature expansion in the feature matrix group after feature expansion through a feature expansion model included in the target prediction model so as to generate an expanded feature matrix, and obtaining an expanded feature matrix group.
In practice, the feature extension model may be a feature insertion model. Specifically, for every two adjacent feature vectors in the feature matrix after segmentation to be feature-extended, the feature extension model may determine a mean feature vector according to the two adjacent feature vectors, and insert the mean feature vector between the two adjacent feature vectors.
As an example, two adjacent feature vectors may be feature vector a and feature vector B, where the mean feature vector= (feature vector a+feature vector B)/2.
And sixthly, carrying out feature fusion on the compressed feature matrix set and the expanded feature matrix set through a feature fusion model included in the target prediction model to obtain a fused feature matrix.
The feature fusion model is used for splicing the compressed feature matrix and the expanded feature matrix according to time sequence to obtain the fused feature matrix.
As an example, the compressed feature matrix set includes a compressed feature matrix a. The extended feature matrix group comprises an extended feature matrix B and an extended feature matrix C. The time interval corresponding to the compressed characteristic matrix A is [20230101, 20230102]. The corresponding time of the characteristic matrix B after expansion is [20230103]. The corresponding time of the characteristic matrix C after expansion is [20230104].
And seventhly, inputting the fused feature matrix into a time sequence type prediction model included in the target prediction model, and generating predicted article circulation information aiming at the target virtual article.
The time-series prediction model may be an RNN (Recurrent Neural Network ) model. The predicted item flow information may characterize a predicted flow amount or predicted profit and loss value of the target virtual item.
The first to seventh steps, as an invention point of the present disclosure, solve the second technical problem mentioned in the background art, that is, "the extraction of data features often needs to set the collection frequency, and when the collection frequency is higher, the volume of data is extremely large, so that more storage resources are occupied. In practice, the data features may be acquired at a daily acquisition frequency, or at a smaller acquisition frequency, such as an hour. However, both the above two modes are to set a fixed sampling frequency to perform data feature sampling. However, the change of the data characteristic corresponding to the virtual article often has uncertainty, that is, the data change may not be obvious in a larger time scale or may change frequently in a smaller time scale, so that it may be found that the fixed acquisition frequency may cause the data characteristic to be oversampled, so that the volume of the data characteristic is extremely large, and thus more storage resources are occupied. Undersampling of the data features may also result in the resulting data features not being indicative of the changing conditions of the target virtual article. Therefore, the present disclosure designs a segmentation model to segment a matrix corresponding to a slower change and a more change of an article transfer curve according to a change condition of the article transfer curve, that is, for a matrix corresponding to a slower change portion, it may be understood that fluctuation of data features is not obvious, feature compression may be performed, for a matrix corresponding to a more change portion, it may be understood that fluctuation of data features is obvious, and feature expansion may be performed to highlight feature change details. And then, performing feature matrix compression on each segmented feature matrix to be subjected to feature compression in the segmented feature matrix group to be subjected to feature compression through a feature compression model included in the target prediction model so as to generate a compressed feature matrix. And expanding the feature matrix of each segmented feature matrix to be feature-expanded in the segmented feature matrix set to be feature-expanded through a feature expansion model included in the target prediction model so as to generate an expanded feature matrix, thereby obtaining an expanded feature matrix set. And further, carrying out feature fusion on the compressed feature matrix set and the expanded feature matrix set through a feature fusion model included in the target prediction model to obtain a fused feature matrix. And thus obtaining the fused feature matrix obtained by time sequence splicing. And finally, inputting the fused feature matrix into a time sequence type prediction model included in the target prediction model to generate predicted article circulation information aiming at the target virtual article. Considering that the characteristic matrix after fusion is a typical matrix containing time sequence characteristics, a time sequence type prediction model is designed for generating predicted article circulation information. The method solves the problem of unstable volume and quality of the data characteristics caused by fixed acquisition frequency, namely, when the volume of the data characteristics is large, the characteristic compression can be carried out so as to reduce the occupation of storage resources. When the data volume is smaller, feature expansion can be performed so as to better characterize the feature change of the target virtual object, the generated fused feature matrix can well express the feature change of the target virtual object, and the accuracy of prediction is improved on the side face.
Optionally, the method further comprises:
firstly, performing feature rough extraction on the associated feature matrix through a feature rough extraction model included in a pre-trained recommendation evaluation information generation model to obtain a feature matrix after feature rough extraction.
The feature coarse extraction model may multiplex the target prediction model to include: a segmentation model, a feature compression model and a feature expansion model.
And secondly, carrying out feature fine extraction on the feature matrix after feature coarse extraction through a feature fine extraction model included in the recommendation evaluation information generation model to obtain a feature matrix after feature fine extraction.
The feature extraction model may be an LSTM (Long short-term memory) model.
And thirdly, inputting the predicted article circulation information and the feature matrix after feature fine extraction into a recommendation degree determining model included in the recommendation degree information generating model to generate recommendation degree information aiming at the target virtual article.
Wherein the recommendation-evaluating information includes: recommendation level value and recommendation risk level. In practice, the recommendation level value may be an accuracy corresponding to the recommended risk level. Specifically, the recommendation degree determining model may be a residual neural network model.
Fourth, an initial set of object information is determined.
Wherein, the initial object information in the initial object information set includes: object representation and object identification. In practice, the executing entity may acquire the initial object information set from a user portrait library constructed in advance.
And fifthly, for each initial object information in the initial object information set, determining the risk resistance information of the object corresponding to the initial object information according to the object portrait included in the initial object information and a pre-trained risk resistance prediction model.
The anti-risk capability information characterizes the anti-risk capability of the object corresponding to the initial object information, and can be specifically characterized by the anti-risk capability level. The risk resistance prediction model comprises: a graph neural network model and a classification layer. The result of the classification layer is a risk resistance level.
And step six, screening out initial object information of which the corresponding risk resistance information is matched with the recommended risk level from the initial object information set, and taking the initial object information as candidate object information to obtain a candidate object information set.
And seventh, determining the recommended flow of the target virtual article according to the recommendation degree value.
In practice, first, the execution body may determine, according to the recommendation value, a position of the target virtual object in the virtual object sequence to be recommended. Then, the execution body may allocate the total recommended flow from the start position of the virtual article sequence to be recommended to the attenuated recommended flow, and when allocating the total recommended flow to the target virtual article, determine the allocated recommended flow as the recommended flow of the target virtual article.
Eighth, for each candidate object information in the candidate object information set, determining a recommendation weight according to the risk resistance capability information included in the candidate object information.
In practice, the executing entity may determine the recommendation weight according to a pre-configured anti-risk capability-recommendation weight mapping table.
And ninth, determining recommended sub-flow corresponding to each candidate object information in the candidate object information set according to the obtained recommended weight set and recommended flow.
As an example, the recommendation weight set includes: recommendation weight a, recommendation weight B, and recommendation weight C. And if the recommended weight a corresponding to the candidate object information a is equal to or greater than the recommended sub-flow=recommended flow× (recommended weight a)/(recommended weight a+recommended weight b+recommended weight C).
Tenth, for each candidate object information in the candidate object information set, pushing target recommendation information to a terminal bound to an object identifier included in the candidate object information.
The target recommendation information is information for introducing the target virtual article. And the information recommended flow of the target recommended information is the same as the recommended sub-flow corresponding to the candidate object information.
As an invention point of the present disclosure, the first to tenth steps described above solve the third technical problem mentioned in the background art, that is, "when recommending a virtual article, the risk resistance of the audience is often not combined, so that accurate recommendation cannot be performed on the virtual article, and the recommended flow is effectively distributed. Based on this, firstly, the present disclosure performs feature rough extraction on the feature matrix after association through the feature rough extraction model included in the recommendation evaluation information generation model trained in advance, so as to achieve the purpose of feature rough extraction, and meanwhile, considering that the segmentation model, the feature compression model and the feature expansion model included in the target prediction model can achieve the purpose of feature rough extraction, and meanwhile, model multiplexing can also reduce the consumption of training resources (such as computing resources) of the model and the model use cost, so that the feature extraction model of the present disclosure can multiplex the segmentation model, the feature compression model and the feature expansion model included in the target prediction model. And secondly, carrying out feature fine extraction on the feature matrix after the feature coarse extraction through a feature fine extraction model included in the recommendation evaluation information generation model to obtain a feature matrix after the feature fine extraction, so as to achieve the purposes of feature fine extraction and feature dimension and volume reduction. And then, inputting the predicted article circulation information and the feature matrix after feature extraction into a recommendation degree determining model included in the recommendation degree information generating model to generate recommendation degree information for the target virtual article. Thereby obtaining accurate recommendation evaluation information. Further, determining an initial object information set, wherein initial object information in the initial object information set includes: object representation and object identification. Then, for each initial object information in the initial object information set, determining risk resistance information of the object corresponding to the initial object information according to the object portrait included in the initial object information and a pre-trained risk resistance prediction model. Thereby enabling a determination of the risk resistance of the object. And selecting initial object information, of which corresponding risk resistance information is matched with the recommended risk level, from the initial object information set as candidate object information, and obtaining a candidate object information set. And then, determining the recommended flow of the target virtual article according to the recommendation degree value. In practice, when the recommendation degree is low, but more recommended flows are distributed, flow waste is caused, and when the recommendation degree is high, but less recommended flows are distributed, the exposure degree of the virtual article is low. Therefore, the corresponding recommended flow rate needs to be determined in combination with the recommendation degree value of the target virtual article. In addition, for each candidate object information in the candidate object information set, a recommendation weight is determined according to the risk resistance capability information included in the candidate object information, and a recommendation sub-flow corresponding to each candidate object information in the candidate object information set is determined according to the obtained recommendation weight set and recommendation flow. In practice, the risk resistance of different objects is different, that is, the acceptance of the same virtual object is often different, for example, the adoption of the uniform flow distribution also causes the waste of recommended flow. And finally, pushing target recommendation information to a terminal bound with the object identifier included in the candidate object information for each candidate object information in the candidate object information set, wherein the target recommendation information is information for introducing the object to the target virtual object, and the information recommendation flow of the target recommendation information is the same as the recommendation sub-flow corresponding to the candidate object information. Therefore, accurate recommendation of the virtual articles and effective distribution of recommended flow are achieved.
The above embodiments of the present disclosure have the following advantageous effects: by the feature information processing method of some embodiments of the present disclosure, occupation of the redundant data to the cache resource is reduced, and use efficiency of the cache resource is improved. Specifically, the reason why there is waste of cache resources is that: the data often has certain data redundancy, a mode of loading the whole data into the cache is adopted, the occupied cache resources can be increased along with the increase of the volume of the data, and meanwhile, the waste of the cache resources can be caused by the redundant data. Based on this, in the feature information processing method of some embodiments of the present disclosure, first, a target feature matrix corresponding to a target virtual article is determined according to an article index tag and a feature construction time period corresponding to the target virtual article, where the target feature matrix includes a plurality of feature vectors that change with time. In practice, since the number of virtual products is numerous, and the feature data corresponding to the virtual products is also increasing with time, it is necessary to obtain the feature data of the target virtual product in the feature construction time period according to the index tag and the feature construction time period corresponding to the virtual article. The volume of the cache resources is primarily reduced, so that occupation of subsequent cache resources is reduced. And secondly, determining an object circulation feature matrix of the target virtual object in the feature construction time period. In practice, virtual products often correspond to valuable related operations, and therefore, it is desirable to determine an item flow feature matrix over a feature build period. Meanwhile, the data volume of the object circulation feature is restrained through the feature construction time period, so that occupation of subsequent cache resources is further reduced. And then, carrying out feature association on the target feature matrix and the object circulation feature matrix to obtain an associated feature matrix. Through feature association, feature matrix combination is realized, so that the quantity of redundant information is reduced. Further, determining policy information corresponding to the target virtual object in a pre-constructed policy information tree as target policy information, and obtaining a target policy information set. In practice, different virtual articles often correspond to different policy information, and therefore, the policy information corresponding to the target virtual article needs to be acquired. Then, generating evaluation information for the target virtual article according to the associated feature matrix and the target strategy information set, wherein the evaluation information comprises the following components: first evaluation information and second evaluation information. Thereby, evaluation information for the target article is generated. And finally, caching the evaluation information and the associated feature matrix into a cache. In practice, because the evaluation information is often corresponding to a higher access frequency, the evaluation information is loaded into the cache, and the time consumption from the reading of the evaluation information from the external memory to the cache is avoided. Meanwhile, compared with the full data, the feature matrix after correlation is the feature matrix subjected to redundant information rejection, and the volume of the data is far smaller than that of the full data, so that the occupation of cache resources and the waste of the cache resources can be greatly reduced.
With further reference to fig. 3, as an implementation of the method shown in the above figures, the present disclosure provides some embodiments of a feature information processing apparatus, which correspond to those method embodiments shown in fig. 1, and which are particularly applicable to various electronic devices.
As shown in fig. 3, the feature information processing apparatus 300 of some embodiments includes: a first determination unit 301, a second determination unit 302, a feature associating unit 303, a third determination unit 304, a generating unit 305, and a buffering unit 306. Wherein, the first determining unit 301 is configured to determine a target feature matrix corresponding to a target virtual article according to an article index tag and a feature construction time period corresponding to the target virtual article, where the target feature matrix includes a plurality of feature vectors that change with time; a second determining unit 302 configured to determine an object circulation feature matrix of the target virtual object in the feature construction period; a feature association unit 303, configured to perform feature association on the target feature matrix and the object circulation feature matrix, so as to obtain an associated feature matrix; a third determining unit 304 configured to determine policy information corresponding to the target virtual object in a policy information tree constructed in advance, as target policy information, to obtain a target policy information set; a generating unit 305 configured to generate evaluation information for the target virtual article according to the associated feature matrix and the target policy information set, where the evaluation information includes: first and second evaluation information; the caching unit 306 is configured to cache the evaluation information and the associated feature matrix into a cache.
It will be appreciated that the elements described in the feature information processing apparatus 300 correspond to the respective steps in the method described with reference to fig. 1. Thus, the operations, features and advantages described above for the method are equally applicable to the feature information processing apparatus 300 and the units contained therein, and are not described herein.
Referring now to fig. 4, a schematic diagram of an electronic device (e.g., computing device) 400 suitable for use in implementing some embodiments of the present disclosure is shown. The electronic device shown in fig. 4 is merely an example and should not impose any limitations on the functionality and scope of use of embodiments of the present disclosure.
As shown in fig. 4, the electronic device 400 may include a processing means (e.g., a central processing unit, a graphics processor, etc.) 401, which may perform various suitable actions and processes according to programs stored in a read-only memory 402 or programs loaded from a storage 408 into a random access memory 403. In the random access memory 403, various programs and data necessary for the operation of the electronic device 400 are also stored. The processing means 401, the read only memory 402 and the random access memory 403 are connected to each other by a bus 404. An input/output interface 405 is also connected to the bus 404.
In general, the following devices may be connected to the I/O interface 405: input devices 406 including, for example, a touch screen, touchpad, keyboard, mouse, camera, microphone, accelerometer, gyroscope, etc.; an output device 407 including, for example, a Liquid Crystal Display (LCD), a speaker, a vibrator, and the like; storage 408 including, for example, magnetic tape, hard disk, etc.; and a communication device 409. The communication means 409 may allow the electronic device 400 to communicate with other devices wirelessly or by wire to exchange data. While fig. 4 shows an electronic device 400 having various means, it is to be understood that not all of the illustrated means are required to be implemented or provided. More or fewer devices may be implemented or provided instead. Each block shown in fig. 4 may represent one device or a plurality of devices as needed.
In particular, according to some embodiments of the present disclosure, the processes described above with reference to flowcharts may be implemented as computer software programs. For example, some embodiments of the present disclosure include a computer program product comprising a computer program embodied on a computer readable medium, the computer program comprising program code for performing the method shown in the flow chart. In such embodiments, the computer program may be downloaded and installed from a network via communications device 409, or from storage 408, or from read only memory 402. The above-described functions defined in the methods of some embodiments of the present disclosure are performed when the computer program is executed by the processing device 401.
It should be noted that, the computer readable medium described in some embodiments of the present disclosure may be a computer readable signal medium or a computer readable storage medium, or any combination of the two. The computer readable storage medium can be, for example, but not limited to, an electronic, magnetic, optical, electromagnetic, infrared, or semiconductor system, apparatus, or device, or a combination of any of the foregoing. More specific examples of the computer-readable storage medium may include, but are not limited to: an electrical connection having one or more wires, a portable computer diskette, a hard disk, a Random Access Memory (RAM), a read-only memory (ROM), an erasable programmable read-only memory (EPROM or flash memory), an optical fiber, a portable compact disc read-only memory (CD-ROM), an optical storage device, a magnetic storage device, or any suitable combination of the foregoing. In some embodiments of the present disclosure, a computer readable storage medium may be any tangible medium that can contain, or store a program for use by or in connection with an instruction execution system, apparatus, or device. In some embodiments of the present disclosure, however, the computer-readable signal medium may comprise a data signal propagated in baseband or as part of a carrier wave, with the computer-readable program code embodied therein. Such a propagated data signal may take any of a variety of forms, including, but not limited to, electro-magnetic, optical, or any suitable combination of the foregoing. A computer readable signal medium may also be any computer readable medium that is not a computer readable storage medium and that can communicate, propagate, or transport a program for use by or in connection with an instruction execution system, apparatus, or device. Program code embodied on a computer readable medium may be transmitted using any appropriate medium, including but not limited to: electrical wires, fiber optic cables, RF (radio frequency), and the like, or any suitable combination of the foregoing.
In some implementations, the clients, servers may communicate using any currently known or future developed network protocol, such as HTTP (Hyper Text Transfer Protocol ), and may be interconnected with any form or medium of digital data communication (e.g., a communication network). Examples of communication networks include a local area network ("LAN"), a wide area network ("WAN"), the internet (e.g., the internet), and peer-to-peer networks (e.g., ad hoc peer-to-peer networks), as well as any currently known or future developed networks.
The computer readable medium may be contained in the electronic device; or may exist alone without being incorporated into the electronic device. The computer readable medium carries one or more programs which, when executed by the electronic device, cause the electronic device to: determining a target feature matrix corresponding to a target virtual article according to an article index label and a feature construction time period corresponding to the target virtual article, wherein the target feature matrix comprises a plurality of feature vectors changing with time; determining an object circulation feature matrix of the target virtual object in the feature construction time period; performing feature association on the target feature matrix and the object circulation feature matrix to obtain an associated feature matrix; determining strategy information corresponding to the target virtual object in a pre-constructed strategy information tree as target strategy information to obtain a target strategy information set; generating evaluation information for the target virtual article according to the associated feature matrix and the target strategy information set, wherein the evaluation information comprises: first and second evaluation information; and caching the evaluation information and the associated feature matrix into a cache.
Computer program code for carrying out operations for some embodiments of the present disclosure may be written in one or more programming languages, including an object oriented programming language such as Java, smalltalk, C ++ and conventional procedural programming languages, such as the "C" programming language or similar programming languages. The program code may execute entirely on the user's computer, partly on the user's computer, as a stand-alone software package, partly on the user's computer and partly on a remote computer or entirely on the remote computer or server. In the case of a remote computer, the remote computer may be connected to the user's computer through any kind of network, including a Local Area Network (LAN) or a Wide Area Network (WAN), or may be connected to an external computer (for example, through the Internet using an Internet service provider).
The flowcharts and block diagrams in the figures illustrate the architecture, functionality, and operation of possible implementations of systems, methods and computer program products according to various embodiments of the present disclosure. In this regard, each block in the flowchart or block diagrams may represent a module, segment, or portion of code, which comprises one or more executable instructions for implementing the specified logical function(s). It should also be noted that, in some alternative implementations, the functions noted in the block may occur out of the order noted in the figures. For example, two blocks shown in succession may, in fact, be executed substantially concurrently, or the blocks may sometimes be executed in the reverse order, depending upon the functionality involved. It will also be noted that each block of the block diagrams and/or flowchart illustration, and combinations of blocks in the block diagrams and/or flowchart illustration, can be implemented by special purpose hardware-based systems which perform the specified functions or acts, or combinations of special purpose hardware and computer instructions.
The units described in some embodiments of the present disclosure may be implemented by means of software, or may be implemented by means of hardware. The described units may also be provided in a processor, for example, described as: a processor includes a first determination unit, a second determination unit, a feature association unit, a third determination unit, a generation unit, and a cache unit. The names of these units do not constitute a limitation on the unit itself in some cases, and for example, the second determining unit may also be described as "a unit that determines an item flow feature matrix of the above-described target virtual item in the above-described feature construction period".
The functions described above herein may be performed, at least in part, by one or more hardware logic components. For example, without limitation, exemplary types of hardware logic components that may be used include: a Field Programmable Gate Array (FPGA), an Application Specific Integrated Circuit (ASIC), an Application Specific Standard Product (ASSP), a system on a chip (SOC), a Complex Programmable Logic Device (CPLD), and the like.
The foregoing description is only of the preferred embodiments of the present disclosure and description of the principles of the technology being employed. It will be appreciated by those skilled in the art that the scope of the invention in the embodiments of the present disclosure is not limited to the specific combination of the above technical features, but encompasses other technical features formed by any combination of the above technical features or their equivalents without departing from the spirit of the invention. Such as the above-described features, are mutually substituted with (but not limited to) the features having similar functions disclosed in the embodiments of the present disclosure.

Claims (8)

1. A feature information processing method, comprising:
determining a target feature matrix corresponding to a target virtual article according to an article index label and a feature construction time period corresponding to the target virtual article, wherein the target feature matrix comprises a plurality of feature vectors changing along with time;
determining an object circulation feature matrix of the target virtual object in the feature construction time period;
performing feature association on the target feature matrix and the object circulation feature matrix to obtain an associated feature matrix;
determining strategy information corresponding to the target virtual object in a pre-constructed strategy information tree as target strategy information to obtain a target strategy information set;
generating evaluation information for the target virtual article according to the associated feature matrix and the target strategy information set, wherein the evaluation information comprises: first and second evaluation information;
and caching the evaluation information and the associated feature matrix into a cache.
2. The method of claim 1, wherein the determining the target feature matrix corresponding to the target virtual item according to the item index tag and the feature construction time period corresponding to the target virtual item comprises:
Determining a feature vector corresponding to the target virtual article in a pre-constructed virtual article feature matrix according to the article index label, and taking the feature vector as a candidate feature vector to obtain a candidate feature vector sequence;
according to the feature construction time period, vector interval interception is carried out on the candidate feature vector sequence, and an intercepted candidate feature vector sequence is obtained;
and generating the target feature matrix according to the intercepted candidate feature vector sequence.
3. The method of claim 2, wherein the policy information tree is a multi-way tree, and the tree nodes in the policy information tree comprise: the information types corresponding to the target strategy information in the target strategy information set comprise: a first information type and a second information type; and
determining policy information corresponding to the target virtual object in the pre-constructed policy information tree as target policy information to obtain a target policy information set, including:
determining a judging condition group corresponding to the target virtual article;
for each policy information in the policy information tree, determining the policy information as target policy information in response to determining that the policy information satisfies the set of decision conditions, wherein the information type of the policy information satisfying the set of decision conditions is a first information type;
And performing recursive backtracking on the strategy information tree according to a father node identifier included in the tree node in the strategy information tree by taking a tree node corresponding to the target strategy information with the information type of a first information type in the target strategy information set as a starting node and a root node of the strategy information tree as an ending node, so as to obtain target strategy information, wherein the information type of the strategy information included in the tree node on a path from the starting node to the ending node is a second information type.
4. The method of claim 3, wherein the generating of the rating information for the target virtual item from the post-association feature matrix and the target policy information set comprises:
according to the correlated feature matrix, determining a value attribute net value of the target virtual article in the feature construction time period;
determining the first evaluation information according to the net value of the value attribute;
for each target policy information in the set of target policy information, performing the following processing steps:
determining a reference evaluation curve corresponding to the target strategy information;
generating a characteristic curve aiming at the evaluation information according to the associated characteristic matrix;
And generating second evaluation information corresponding to the target strategy information according to the reference evaluation curve and the characteristic curve.
5. The method of claim 4, wherein the performing feature association on the target feature matrix and the item circulation feature matrix to obtain an associated feature matrix comprises:
and under the condition of time alignment, determining the Cartesian product of the target feature matrix and the object circulation feature matrix as the associated feature matrix.
6. A feature information processing apparatus comprising:
a first determining unit configured to determine a target feature matrix corresponding to a target virtual article according to an article index tag and a feature construction time period corresponding to the target virtual article, wherein the target feature matrix comprises a plurality of feature vectors which change with time;
a second determining unit configured to determine an item circulation feature matrix of the target virtual item within the feature construction period;
the feature association unit is configured to perform feature association on the target feature matrix and the object circulation feature matrix to obtain an associated feature matrix;
a third determining unit configured to determine policy information corresponding to the target virtual article in a policy information tree constructed in advance as target policy information, and obtain a target policy information set;
A generating unit configured to generate evaluation information for the target virtual article according to the associated feature matrix and the target policy information set, wherein the evaluation information includes: first and second evaluation information;
and the caching unit is configured to cache the evaluation information and the associated feature matrix into a cache.
7. An electronic device, comprising:
one or more processors;
a storage device having one or more programs stored thereon;
when executed by the one or more processors, causes the one or more processors to implement the method of any of claims 1 to 5.
8. A computer readable medium having stored thereon a computer program, wherein the computer program, when executed by a processor, implements the method of any of claims 1 to 5.
CN202310479432.4A 2023-04-28 2023-04-28 Feature information processing method, device, electronic equipment and computer readable medium Active CN116186020B (en)

Priority Applications (1)

Application Number Priority Date Filing Date Title
CN202310479432.4A CN116186020B (en) 2023-04-28 2023-04-28 Feature information processing method, device, electronic equipment and computer readable medium

Applications Claiming Priority (1)

Application Number Priority Date Filing Date Title
CN202310479432.4A CN116186020B (en) 2023-04-28 2023-04-28 Feature information processing method, device, electronic equipment and computer readable medium

Publications (2)

Publication Number Publication Date
CN116186020A CN116186020A (en) 2023-05-30
CN116186020B true CN116186020B (en) 2023-07-04

Family

ID=86452737

Family Applications (1)

Application Number Title Priority Date Filing Date
CN202310479432.4A Active CN116186020B (en) 2023-04-28 2023-04-28 Feature information processing method, device, electronic equipment and computer readable medium

Country Status (1)

Country Link
CN (1) CN116186020B (en)

Citations (5)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN108280104A (en) * 2017-02-13 2018-07-13 腾讯科技(深圳)有限公司 The characteristics information extraction method and device of target object
CN111597356A (en) * 2020-05-27 2020-08-28 上海乂学教育科技有限公司 Intelligent education knowledge map construction system and method
CN112632397A (en) * 2021-01-04 2021-04-09 同方知网(北京)技术有限公司 Personalized recommendation method based on multi-type academic achievement portrait and mixed recommendation strategy
CN113435960A (en) * 2021-06-04 2021-09-24 北京沃东天骏信息技术有限公司 Virtual article display method and device, electronic equipment and computer readable medium
TW202247079A (en) * 2021-05-27 2022-12-01 鄭匡祐 Investment transaction system

Family Cites Families (1)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
US20170243295A1 (en) * 2016-02-22 2017-08-24 AlphaVee Solutions Ltd. Graphical user interface for displaying dynamic strategies

Patent Citations (5)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN108280104A (en) * 2017-02-13 2018-07-13 腾讯科技(深圳)有限公司 The characteristics information extraction method and device of target object
CN111597356A (en) * 2020-05-27 2020-08-28 上海乂学教育科技有限公司 Intelligent education knowledge map construction system and method
CN112632397A (en) * 2021-01-04 2021-04-09 同方知网(北京)技术有限公司 Personalized recommendation method based on multi-type academic achievement portrait and mixed recommendation strategy
TW202247079A (en) * 2021-05-27 2022-12-01 鄭匡祐 Investment transaction system
CN113435960A (en) * 2021-06-04 2021-09-24 北京沃东天骏信息技术有限公司 Virtual article display method and device, electronic equipment and computer readable medium

Non-Patent Citations (3)

* Cited by examiner, † Cited by third party
Title
Entropy trading strategies reveal inefficiencies in Japanese stock market;Levan Efremidze et al.;《International Review of Economics & Finance》;464-477 *
基于中国股票股价关联效应的量化投资策略研究;韩博阳;《中国优秀硕士学位论文全文数据库 经济与管理科学辑》;J159-227 *
面向监管报送数据治理的元数据血缘图谱研究;王哲 等;《中国金融电脑》;58-62 *

Also Published As

Publication number Publication date
CN116186020A (en) 2023-05-30

Similar Documents

Publication Publication Date Title
CN110363449B (en) Risk identification method, device and system
JP2023501469A (en) Fusion of multimodal data using recurrent neural networks
CN113393306A (en) Product recommendation method and device, electronic equipment and computer readable medium
CN114049072B (en) Index determination method and device, electronic equipment and computer readable medium
CN116186020B (en) Feature information processing method, device, electronic equipment and computer readable medium
CN111259975B (en) Method and device for generating classifier and method and device for classifying text
JP2022534160A (en) Methods and devices for outputting information, electronic devices, storage media, and computer programs
CN115730591A (en) User service method, device, equipment and storage medium based on knowledge graph
CN111754984B (en) Text selection method, apparatus, device and computer readable medium
US11734602B2 (en) Methods and systems for automated feature generation utilizing formula semantification
CN112948584A (en) Short text classification method, device, equipment and storage medium
CN113760178A (en) Cache data processing method and device, electronic equipment and computer readable medium
CN113284027B (en) Training method of partner recognition model, abnormal partner recognition method and device
CN111783808A (en) Method and apparatus for generating information
CN116823407B (en) Product information pushing method, device, electronic equipment and computer readable medium
US11880765B2 (en) State-augmented reinforcement learning
CN117077982B (en) Project material scheduling method and device, electronic equipment and readable medium
CN116384945B (en) Project management method and system
CN117689252A (en) Data product related information storage method, device, equipment and medium
CN112860999A (en) Information recommendation method, device, equipment and storage medium
CN117520410A (en) Service data processing method, device, electronic equipment and computer readable medium
CN117951517A (en) Training method and device of node classification model, electronic equipment and storage medium
CN114911925A (en) Block chain address classification method and device
CN116975580A (en) Information evaluation method, device and equipment
CN116823386A (en) Commodity mining method, commodity mining device, commodity mining equipment and storage medium

Legal Events

Date Code Title Description
PB01 Publication
PB01 Publication
SE01 Entry into force of request for substantive examination
SE01 Entry into force of request for substantive examination
GR01 Patent grant
GR01 Patent grant