A cloud data collection method and system based on index tree

Abstract

The present invention provides a cloud data collection method and system based on an index tree, the method comprising: constructing a plurality of category sets, the category sets corresponding to a plurality of benchmark keywords; performing byte splitting on the benchmark keywords to obtain unique identification vectors; classifying and processing a plurality of unique identification vectors within the category sets to obtain a plurality of optimal sets, generating classification points based on the optimal sets, and constructing an index tree through the category sets, classification points, and unique identification vectors; when collecting local data, generating a link path and associating it to the unique identification vector; when a user performs data retrieval, obtaining a second keyword and generating a guidance vector, and obtaining a link path based on the guidance vector. By comparing the guidance vector obtained based on byte splitting and the unique identification vector, it is ensured that an approximate target identification vector can still be retrieved based on the correct part when a deviation is input, avoiding the situation where no data is obtained, improving the accuracy of the retrieval, and thus improving the utilization rate of the cloud.

Description

Technical Field

[0001] The present invention relates to the technical field of data processing, and in particular to a cloud data collection method and system based on an index tree. Background Art

[0002] With the integration and development of Internet technology and the big data industry, the traditional computing model can no longer meet the current needs for data storage and processing due to its limited computing resources and storage capabilities. The emergence of cloud computing technology provides users with powerful computing resources and storage resources.

[0003] As an important part of cloud computing, cloud storage allows remote storage devices to collect data from users and allows users to access remote storage devices through the network to achieve data collection, storage and sharing. Compared with traditional local storage technology, cloud storage has the advantages of low cost, easy expansion and high security.

[0004] However, to ensure data security, after data is collected, it is generally stored in the cloud in an encrypted form. This also requires users to accurately enter query keywords when searching for data in the cloud in order to distinguish encrypted data from random strings and obtain the required cloud data. When there is a deviation in the search keywords, it will often lead to the inability to obtain cloud data, which reduces the availability of the cloud. Summary of the invention

[0005] In view of the shortcomings of the prior art, the purpose of the present invention is to provide a cloud data collection method and system based on an index tree, aiming to solve the technical problem in the prior art that after collecting data in the cloud, if there is a deviation in the input of keywords when the user performs data retrieval, it will lead to the inability to obtain cloud data, thereby reducing the availability of the cloud.

[0006] In order to achieve the above objectives, in a first aspect, an embodiment of the present application provides a cloud data collection method based on an index tree, comprising the following steps:

[0007] Constructing a plurality of category sets, each of the category sets corresponding to a plurality of benchmark keywords;

[0008] Performing byte splitting on the benchmark keyword to obtain a plurality of benchmark feature vectors corresponding to the benchmark keyword, wherein the plurality of benchmark feature vectors all include the same number of vector values, and generating a unique identification vector corresponding to the benchmark keyword based on the plurality of benchmark feature vectors, wherein the unique identification vector includes a plurality of update weight values;

[0009] In the category set, a plurality of the unique identification vectors are classified to obtain a plurality of optimal sets, classification points are generated based on the optimal sets, the classification points correspond to a plurality of the unique identification vectors, and an index tree is constructed based on the category set, the classification points and the unique identification vectors;

[0010] When collecting local data, extracting a first keyword of the local data, encrypting the local data into ciphertext data, storing the ciphertext data, and generating a link path corresponding to the reference keyword, and associating the link path to the unique identification vector through a corresponding relationship between the first keyword and the reference keyword;

[0011] When the user performs data retrieval, a second keyword input by the user is obtained, a target set is selected from the plurality of category sets based on the second keyword, a guidance vector is generated through the second keyword, a target point is selected from the plurality of classification points based on the guidance vector, and a target identification vector is determined from the plurality of unique identification vectors corresponding to the target point to obtain the link path.

[0012] Compared with the prior art, the beneficial effects of the present invention are as follows: by performing byte splitting on the benchmark keyword, it is possible to ensure that the byte order relationship between the unique identification vector generated by the benchmark keyword and the benchmark keyword corresponds, thereby improving the strong correlation between the benchmark keyword and the unique identification vector; by constructing the category set and the classification point, it is avoided that in the subsequent retrieval process, all of the unique identification vectors need to be traversed for data retrieval, thereby improving the efficiency of data retrieval; by comparing the guidance vector with the unique identification vector, it is ensured that in the case of a deviation in the input of the second keyword, an approximate target identification vector can still be retrieved based on the correct part, thereby avoiding the situation where no data is obtained, improving the accuracy of the retrieval, and thereby improving the utilization rate of the cloud; by converting the second keyword and the benchmark keyword into the guidance vector and the unique identification vector respectively, the complexity of the retrieval is reduced, the computing efficiency is improved, and thereby the retrieval efficiency is improved.

[0013] Furthermore, the step of performing byte splitting on the benchmark keyword to obtain a plurality of benchmark feature vectors corresponding to the benchmark keyword includes:

[0014] The benchmark keyword is split once in single-byte units to obtain a plurality of single-byte features;

[0015] The benchmark keyword is split twice in double-byte units to obtain a plurality of double-byte features;

[0016] Hash mapping is performed on the single-byte feature and the double-byte feature respectively to obtain a plurality of reference feature vectors.

[0017] Furthermore, the step of generating a unique identification vector corresponding to the benchmark keyword based on the plurality of benchmark feature vectors, wherein the unique identification vector includes a plurality of updating weight values ​​comprises:

[0018] Mapping the vector values ​​at the same position in a plurality of the reference feature vectors into weight component values ​​according to a preset mapping rule;

[0019] The weight component value is compared with a weight threshold to obtain an updated weight value, and a plurality of the updated weight values ​​are combined into the unique identification vector.

[0020] Furthermore, the step of mapping the vector values ​​at the same position in the plurality of reference feature vectors into weight component values ​​according to a preset mapping rule is specifically as follows:

[0021] Determine whether the vector value at the same position in a plurality of the reference feature vectors is 1;

[0022] If the vector value is 1, a sub-weight value of 1 is generated;

[0023] If the vector value is not 1, a sub-weight value of -1 is generated;

[0024] Superimposing a plurality of said sub-weight values ​​into a weight component value;

[0025] The step of comparing the weight component value with the weight threshold to obtain an updated weight value is specifically as follows:

[0026] Determine whether the weight component value is less than 0;

[0027] If the weight component value is less than 0, an updated weight value of 0 is generated;

[0028] If the weight component value is greater than or equal to 0, an updated weight value of 1 is generated.

[0029] Furthermore, the step of classifying the plurality of unique identification vectors to obtain a plurality of optimal sets includes:

[0030] Selecting a first number of first reference vectors from the plurality of unique identification vectors, and dividing the plurality of unique identification vectors into a plurality of initial sets according to the Euclidean distance between the first reference vectors and other unique identification vectors;

[0031] setting one of the initial sets as a reference set, segmenting the updated weight value of the first reference vector in the reference set into an optimal value and a plurality of first neighboring values, and determining the updated weight value of the first reference vector in the other initial sets as a plurality of second neighboring values;

[0032] Obtaining a classification evaluation sub-coefficient of the benchmark set through the optimal value, the first neighboring value, and the second neighboring value, determining a total evaluation coefficient through a plurality of the classification evaluation sub-coefficients, and iteratively updating the first quantity through the total evaluation coefficient to obtain an optimal quantity;

[0033] The optimal number of second reference vectors is selected from the plurality of unique identification vectors, and the plurality of unique identification vectors are divided into a plurality of optimal sets according to the Euclidean distance between the second reference vector and other unique identification vectors.

[0034] Furthermore, the calculation formula of the classification evaluation sub-coefficient is:

[0035] ,

[0036] in, It represents the classification evaluation sub-coefficient of the benchmark set when the i-th initial set is set as the benchmark set. It indicates the minimum distance between the optimal value and several second neighboring values ​​when the i-th initial set is set as the benchmark set and the j-th updated weight value in the first benchmark vector in the benchmark set is the optimal value. It represents the average distance between the optimal value and several first neighboring values ​​when the i-th initial set is set as the benchmark set and the j-th updated weight value in the first benchmark vector in the benchmark set is the optimal value.

[0037] Furthermore, the step of generating classification points based on the optimal set is specifically:

[0038] Obtain the updated weight values ​​of all the unique identification vectors in the optimal set, and determine whether the updated weight values ​​at the same position are the same from left to right. If the updated weight values ​​at the same position are the same, extract the updated weight value as the classification weight value, and combine several classification weight values ​​into the classification point.

[0039] Furthermore, the step of determining a target identification vector from a plurality of the unique identification vectors corresponding to the target point comprises:

[0040] Selecting a plurality of the unique identification vectors corresponding to the target point as vectors to be selected;

[0041] Respectively obtain similarity values ​​of a plurality of the candidate vectors and the guide vector;

[0042] A maximum similarity value is determined from the plurality of similarity values, and the candidate vector corresponding to the maximum similarity value is determined as a target identification vector.

[0043] In a second aspect, an embodiment of the present application provides a cloud data collection system based on an index tree, which is applied to the cloud data collection method based on an index tree as described in the first aspect above, and the system includes:

[0044] A first construction module is used to construct a plurality of category sets, each of which corresponds to a plurality of benchmark keywords;

[0045] An analysis module, configured to perform byte splitting on the benchmark keyword to obtain a plurality of benchmark feature vectors corresponding to the benchmark keyword, wherein the plurality of benchmark feature vectors all include the same number of vector values, and to generate a unique identification vector corresponding to the benchmark keyword based on the plurality of benchmark feature vectors, wherein the unique identification vector includes a plurality of update weight values;

[0046] A second construction module is used to classify a plurality of the unique identification vectors in the category set to obtain a plurality of optimal sets, generate classification points based on the optimal sets, the classification points correspond to a plurality of the unique identification vectors, and construct an index tree based on the category set, the classification points and the unique identification vectors;

[0047] an associating module, configured to extract a first keyword of the local data when collecting local data, encrypt the local data into ciphertext data, store the ciphertext data, generate a link path corresponding to the reference keyword, and associate the link path with the unique identification vector through a corresponding relationship between the first keyword and the reference keyword;

[0048] The execution module is used to obtain a second keyword input by the user when the user performs data retrieval, select a target set from the plurality of category sets according to the second keyword, generate a guidance vector through the second keyword, select a target point from the plurality of classification points based on the guidance vector, and determine a target identification vector from the plurality of unique identification vectors corresponding to the target point to obtain the link path.

[0049] In a third aspect, an embodiment of the present application provides a computer, comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein when the processor executes the computer program, the index tree-based cloud data collection method as described in the first aspect above is implemented.

[0050] In a fourth aspect, an embodiment of the present application provides a storage medium on which a computer program is stored. When the computer program is executed by a processor, the index tree-based cloud data collection method as described in the first aspect above is implemented. BRIEF DESCRIPTION OF THE DRAWINGS

[0051] Figure 1 It is a flow chart of the cloud data collection method based on the index tree in the first embodiment of the present invention;

[0052] Figure 2 It is a structural block diagram of a cloud data acquisition system based on an index tree in a second embodiment of the present invention;

[0053] The following specific implementation manner will further illustrate the present invention in conjunction with the above-mentioned drawings. DETAILED DESCRIPTION

[0054] In order to facilitate the understanding of the present invention, the present invention will be described more fully below with reference to the relevant drawings. Several embodiments of the present invention are given in the drawings. However, the present invention can be implemented in many different forms and is not limited to the embodiments described herein. On the contrary, the purpose of providing these embodiments is to make the disclosure of the present invention more thorough and comprehensive.

[0055] It should be noted that when an element is referred to as being "fixed to" another element, it may be directly on the other element or there may be a central element. When an element is considered to be "connected to" another element, it may be directly connected to the other element or there may be a central element at the same time. The terms "vertical", "horizontal", "left", "right" and similar expressions used herein are for illustrative purposes only.

[0056] Unless otherwise defined, all technical and scientific terms used herein have the same meaning as those commonly understood by those skilled in the art to which the present invention belongs. The terms used herein in the specification of the present invention are only for the purpose of describing specific embodiments and are not intended to limit the present invention. The term "and / or" used herein includes any and all combinations of one or more of the related listed items.

[0057] See also Figure 1 The first embodiment of the present invention provides a cloud data collection method based on an index tree, comprising the following steps:

[0058] S10: constructing a plurality of category sets, each of which corresponds to a plurality of benchmark keywords;

[0059] The base keywords may be classified based on actual usage or academic dictionaries to complete the construction of the category set. For example, when accounting and exchange rate are the base keywords, commerce is the category set.

[0060] S20: performing byte splitting on the benchmark keyword to obtain a plurality of benchmark feature vectors corresponding to the benchmark keyword, wherein the plurality of benchmark feature vectors all include the same number of vector values, and generating a unique identification vector corresponding to the benchmark keyword based on the plurality of benchmark feature vectors, wherein the unique identification vector includes a plurality of update weight values;

[0061] The step S20 comprises:

[0062] S210: splitting the reference keyword once in single-byte units to obtain a plurality of single-byte features;

[0063] S220: splitting the reference keyword twice in double-byte units to obtain a plurality of double-byte features;

[0064] Taking accounting as an example, its single-byte features are a, c, c, o, u, n, t, i, n, g, and its double-byte features are: ac, cc, co, ou, un, nt, ti, in, ng. By performing the first split and the second split, it can be ensured that the benchmark feature vector obtained subsequently can not only fully represent all the features of the benchmark keyword, but also represent the order of its bytes.

[0065] S230: Perform hash mapping on the single-byte feature and the double-byte feature respectively to obtain a plurality of reference feature vectors.

[0066] The hash mapping is to convert the single-byte feature and the double-byte feature into a feature vector by setting a hash function, such as: a is converted into the reference feature vector [010100000110] with 12 vector values ​​after hash mapping. It can be understood that all the single-byte features and the double-byte features correspond to one reference feature vector. And each reference feature vector has 12 vector values.

[0067] S240: Mapping the vector values ​​at the same position in a plurality of the reference feature vectors into weight component values ​​according to a preset mapping rule;

[0068] Determine whether the vector value at the same position in several of the reference feature vectors is 1; if the vector value is 1, generate a sub-weight value with a value of 1; if the vector value is not 1, generate a sub-weight value with a value of -1; and superimpose several of the sub-weight values ​​into a weight component value.

[0069] For example, after a is hash mapped, the reference feature vector [010100000110] with 12 vector values ​​is formed, and after c is hash mapped, the reference feature vector [001101001000] with 12 vector values ​​is formed. When obtaining the sub-weight values, the vector values ​​at the same position in the two reference feature vectors are determined, and the set of the weight component values ​​formed by a and c is [-2002 -20 -2 -2000 -2]. When multiple reference feature vectors are analyzed comprehensively, the calculation method is consistent and will not be repeated here.

[0070] S250: Compare the weight component value with the weight threshold to obtain an updated weight value, and combine a plurality of the updated weight values ​​into the unique identification vector;

[0071] Specifically, it is determined whether the weight component value is less than 0; if the weight component value is less than 0, an updated weight value of 0 is generated; if the weight component value is greater than or equal to 0, an updated weight value of 1 is generated. When the set of weight component values ​​formed by a and c is [-2002 -20 -2 -2000 -2], its unique identification vector is [011101001110], by obtaining the updated weight value, it is converted into a hash-mapped form, which is convenient for subsequent corresponding data processing.

[0072] S30: in the category set, classify a plurality of the unique identification vectors to obtain a plurality of optimal sets, generate classification points based on the optimal sets, the classification points correspond to a plurality of the unique identification vectors, and construct an index tree based on the category set, the classification points and the unique identification vectors;

[0073] Specifically, the step S30 includes:

[0074] S310: selecting a first number of first reference vectors from the plurality of unique identification vectors, and dividing the plurality of unique identification vectors into a plurality of initial sets according to the Euclidean distance between the first reference vectors and other unique identification vectors;

[0075] The first number is greater than 1, and the first number is less than the total number of the unique identification vectors. After obtaining the first number of the first reference vectors, the first reference vector is used as the center point of the data, and based on the Euclidean distance between other unique identification vectors and different first reference vectors, the attribution of other unique identification vectors is divided. It can be understood that the number of the initial set is the first number.

[0076] S320: setting one of the initial sets as a reference set, segmenting the updated weight value of the first reference vector in the reference set into an optimal value and a plurality of first neighboring values, and determining the updated weight value of the first reference vector in the other initial sets as a plurality of second neighboring values;

[0077] S330: Obtain the classification evaluation sub-coefficient of the benchmark set through the optimal value, the first neighboring value, and the second neighboring value, determine the total evaluation coefficient through a plurality of the classification evaluation sub-coefficients, and iteratively update the first quantity through the total evaluation coefficient to obtain the optimal quantity;

[0078] The calculation formula of the classification evaluation sub-coefficient is:

[0079] ,

[0080] in, It represents the classification evaluation sub-coefficient of the benchmark set when the i-th initial set is set as the benchmark set. It represents the minimum distance between the optimal value and several second neighboring values ​​when the i-th initial set is set as the benchmark set and the j-th updated weight value in the first benchmark vector in the benchmark set is the optimal value. It represents the average distance between the optimal value and several first neighboring values ​​when the i-th initial set is set as the benchmark set and the j-th updated weight value in the first benchmark vector in the benchmark set is the optimal value.

[0081] All the initial sets are set as the reference sets one by one, so as to obtain the classification evaluation sub-coefficients corresponding to all the initial sets respectively, and the classification evaluation sub-coefficients of all the initial sets are weighted averaged to obtain the total evaluation coefficient. In this embodiment, the first quantity is set several times, and the first quantity set several times is all quantities within the range of 1 to the total number of the unique identification vectors. By sorting the total evaluation coefficients formed by the first quantities several times, the first quantity corresponding to the total evaluation coefficient with the highest value is selected as the optimal quantity. In some embodiments, by setting an evaluation threshold, the total evaluation coefficient is compared with the evaluation threshold to determine the optimal quantity.

[0082] S340: selecting the optimal number of second reference vectors from the plurality of unique identification vectors, and dividing the plurality of unique identification vectors into a plurality of optimal sets according to the Euclidean distance between the second reference vectors and other unique identification vectors;

[0083] The execution method of step S340 is consistent with that of step S310, and only the number of selected data center points is different, which will not be described again here, and the first reference vector and the second reference vector are both the unique identification vector.

[0084] S350: Obtain the updated weight values ​​of all the unique identification vectors in the optimal set, and determine whether the updated weight values ​​at the same position are the same from left to right. If the updated weight values ​​at the same position are the same, extract the updated weight value as a classification weight value, and combine several classification weight values ​​into the classification point;

[0085] Assuming that there are three unique identification vectors in the optimal set, which are: [011101001110], [011000011010], [011110000111], it is determined from left to right whether the updated weight values ​​in the three unique identification vectors are the same. If they are the same, they are extracted as the classification weight values ​​until a different updated weight value appears at a certain position. Based on this, the classification point is:

[011] . By setting the classification point, the unique identification vectors under the category set can be further divided by similarity, thereby improving the efficiency of subsequent retrieval.

[0086] S40: when collecting local data, extracting a first keyword of the local data, encrypting the local data into ciphertext data, storing the ciphertext data, generating a link path corresponding to the reference keyword, and associating the link path to the unique identification vector through a corresponding relationship between the first keyword and the reference keyword;

[0087] It is understandable that the first keyword is the same as a certain benchmark keyword. When collecting different local data, based on different first keywords, different link paths are linked to the unique identification vector under the index tree to further improve the index tree and form an overall retrieval system. Data encryption has been used a lot in cloud data storage and will not be described here. In some embodiments, the link path can be encrypted to form a ciphertext link, and the ciphertext link can be associated with the corresponding unique identification vector to further enhance the privacy and security of data storage.

[0088] S50: when the user performs data retrieval, a second keyword input by the user is obtained, a target set is selected from the plurality of category sets according to the second keyword, a guidance vector is generated through the second keyword, a target point is selected from the plurality of category points based on the guidance vector, and a target identification vector is determined from the plurality of unique identification vectors corresponding to the target point, so as to obtain the link path;

[0089] It can be understood that the second keyword is the same as the first keyword and the benchmark keyword. Through the correspondence between the benchmark keyword and the category set, the target set can be obtained through the second keyword. The step of generating the guidance vector based on the second keyword is consistent with the step of generating the unique identification vector based on the benchmark keyword, and will not be repeated here.

[0090] Specifically, the step S50 includes:

[0091] S510: Selecting a plurality of the unique identification vectors corresponding to the target point as vectors to be selected;

[0092] S520: Obtaining similarity values ​​of a plurality of the candidate vectors and the guide vector respectively;

[0093] Taking accounting as an example, if the unique identification vector converted into it is [011110000111], when the second keyword input is also accounting, one of the candidate vectors obtained is [011110000111], and the guidance vector is [011110000111]. By comparing the similarity between the candidate vector and the guidance vector, the link path corresponding to accounting can be obtained, and the data can be obtained after decryption by the secret key.

[0094] S530: determining a maximum similarity value from a plurality of similarity values, and determining the candidate vector corresponding to the maximum similarity value as a target identification vector;

[0095] When the user input is incorrect, such as accounting is input as acaounting, due to the steps of byte splitting, hash mapping, weight division and reverse hash mapping, the guidance vector formed by its transformation is [011010000111], which has only a small difference from [011110000111]. Therefore, fuzzy retrieval of data can be completed with a high similarity.

[0096] By performing byte splitting on the benchmark keyword, it can be ensured that the byte order relationship between the unique identification vector generated by the benchmark keyword and the benchmark keyword corresponds, thereby improving the strong correlation between the benchmark keyword and the unique identification vector; by constructing the category set and the classification point, it is avoided that in the subsequent retrieval process, all the unique identification vectors need to be traversed for data retrieval, thereby improving the efficiency of data retrieval; by comparing the guidance vector with the unique identification vector, it can be ensured that when there is a deviation in the input of the second keyword, the approximate target identification vector can still be retrieved based on the correct part, thereby avoiding the situation where no data is obtained, improving the accuracy of the retrieval, and further improving the utilization rate of the cloud; by converting the second keyword and the benchmark keyword into the guidance vector and the unique identification vector respectively, the complexity of the retrieval is reduced, the computing efficiency is improved, and further improving the retrieval efficiency.

[0097] See also Figure 2 , the second embodiment of the present invention provides a cloud data acquisition system based on an index tree, which is applied to the cloud data acquisition method based on an index tree in the above-mentioned embodiment, and will not be repeated hereafter. As used below, the terms "module", "unit", "sub-unit", etc. may be a combination of software and / or hardware that implements a predetermined function. Although the apparatus described in the following embodiments is preferably implemented in software, the implementation in hardware, or a combination of software and hardware, is also possible and conceivable.

[0098] The system comprises:

[0099] A first construction module 10 is used to construct a plurality of category sets, each of which corresponds to a plurality of benchmark keywords;

[0100] The analysis module 20 is used to perform byte splitting on the benchmark keyword to obtain a plurality of benchmark feature vectors corresponding to the benchmark keyword, wherein the plurality of benchmark feature vectors all include the same number of vector values, and to generate a unique identification vector corresponding to the benchmark keyword based on the plurality of benchmark feature vectors, wherein the unique identification vector includes a plurality of update weight values;

[0101] The analysis module 20 comprises:

[0102] The first unit is used to split the reference keyword once in single-byte units to obtain a plurality of single-byte features;

[0103] The second unit is used to split the reference keyword twice in double-byte units to obtain a plurality of double-byte features;

[0104] A third unit is used to perform hash mapping on the single-byte feature and the double-byte feature respectively to obtain a plurality of reference feature vectors;

[0105] A fourth unit is used to map the vector values ​​at the same position in a plurality of the reference feature vectors into weight component values ​​according to a preset mapping rule;

[0106] The fourth unit is specifically used to determine whether the vector value at the same position in a plurality of the reference feature vectors is 1; if the vector value is 1, a sub-weight value with a value of 1 is generated; if the vector value is not 1, a sub-weight value with a value of -1 is generated; and a plurality of the sub-weight values ​​are superimposed as a weight component value;

[0107] A fifth unit, configured to compare the weight component value with a weight threshold to obtain an updated weight value, and combine a plurality of the updated weight values ​​into the unique identification vector;

[0108] The fifth unit is specifically used to determine whether the weight component value is less than 0; if the weight component value is less than 0, generate an updated weight value of 0; if the weight component value is greater than or equal to 0, generate an updated weight value of 1;

[0109] A second construction module 30 is used to classify a plurality of the unique identification vectors in the category set to obtain a plurality of optimal sets, generate classification points based on the optimal sets, the classification points correspond to a plurality of the unique identification vectors, and construct an index tree based on the category set, the classification points and the unique identification vectors;

[0110] The second building block 30 comprises:

[0111] A sixth unit is used to select a first number of first reference vectors from the plurality of unique identification vectors, and divide the plurality of unique identification vectors into a plurality of initial sets according to the Euclidean distance between the first reference vectors and other unique identification vectors;

[0112] a seventh unit, configured to set one of the initial sets as a reference set, segment the updated weight value of the first reference vector in the reference set into an optimal value and a plurality of first neighboring values, and determine the updated weight value of the first reference vector in the other initial sets as a plurality of second neighboring values;

[0113] An eighth unit is used to obtain a classification evaluation sub-coefficient of the benchmark set through the optimal value, the first neighboring value, and the second neighboring value, determine a total evaluation coefficient through a plurality of the classification evaluation sub-coefficients, and iteratively update the first quantity through the total evaluation coefficient to obtain an optimal quantity;

[0114] A ninth unit is used to select the optimal number of second reference vectors from the plurality of unique identification vectors, and divide the plurality of unique identification vectors into a plurality of optimal sets according to the Euclidean distance between the second reference vector and other unique identification vectors;

[0115] A tenth unit is used to obtain the updated weight values ​​of all the unique identification vectors in the optimal set, and determine whether the updated weight values ​​at the same position are the same from left to right. If the updated weight values ​​at the same position are the same, the updated weight value is extracted as a classification weight value, and a plurality of the classification weight values ​​are combined into the classification point;

[0116] an associating module 40, for extracting a first keyword of the local data when collecting local data, encrypting the local data into ciphertext data, storing the ciphertext data, and generating a link path corresponding to the reference keyword, and associating the link path to the unique identification vector through a corresponding relationship between the first keyword and the reference keyword;

[0117] The execution module 50 is used for obtaining a second keyword input by the user when the user performs data retrieval, selecting a target set from the plurality of category sets according to the second keyword, generating a guidance vector through the second keyword, selecting a target point from the plurality of category points based on the guidance vector, and determining a target identification vector from the plurality of unique identification vectors corresponding to the target point, so as to obtain the link path;

[0118] The execution module 50 includes:

[0119] An eleventh unit is used to select a plurality of the unique identification vectors corresponding to the target point as vectors to be selected;

[0120] A twelfth unit is used to obtain similarity values ​​between a plurality of the candidate vectors and the guide vector respectively;

[0121] The thirteenth unit is used to determine a maximum similarity value from the plurality of similarity values, and determine the candidate vector corresponding to the maximum similarity value as a target identification vector.

[0122] The present invention also provides a computer, comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein when the processor executes the computer program, the index tree-based cloud data acquisition method as described in the above technical solution is implemented.

[0123] The present invention also provides a storage medium on which a computer program is stored. When the computer program is executed by a processor, the cloud data acquisition method based on the index tree as described in the above technical solution is implemented.

[0124] In the description of this specification, the description with reference to the terms "one embodiment", "some embodiments", "examples", "specific examples", or "some examples" means that the specific features, structures, materials or characteristics described in conjunction with the embodiment or example are included in at least one embodiment or example of the present invention. In this specification, the schematic representation of the above terms does not necessarily refer to the same embodiment or example. Moreover, the specific features, structures, materials or characteristics described may be combined in any one or more embodiments or examples in a suitable manner.

[0125] The above-mentioned embodiments only express several implementation methods of the present invention, and the description thereof is relatively specific and detailed, but it cannot be understood as limiting the scope of the patent of the present invention. It should be pointed out that, for ordinary technicians in this field, several variations and improvements can be made without departing from the concept of the present invention, which all belong to the protection scope of the present invention. Therefore, the protection scope of the patent of the present invention shall be subject to the attached claims.

Claims

1. A cloud data collection method based on an index tree, characterized in that: The following steps are involved: Constructing a plurality of category sets, each of the category sets corresponding to a plurality of benchmark keywords; Performing byte splitting on the benchmark keyword to obtain a plurality of benchmark feature vectors corresponding to the benchmark keyword, wherein the plurality of benchmark feature vectors all include the same number of vector values, and generating a unique identification vector corresponding to the benchmark keyword based on the plurality of benchmark feature vectors, wherein the unique identification vector includes a plurality of update weight values; The step of generating a unique identification vector corresponding to the benchmark keyword based on the plurality of benchmark feature vectors, wherein the unique identification vector includes a plurality of updating weight values ​​comprises: Mapping the vector values ​​at the same position in a plurality of the reference feature vectors into weight component values ​​according to a preset mapping rule; Comparing the weight component value with a weight threshold to obtain an updated weight value, and combining a plurality of the updated weight values ​​into the unique identification vector; In the category set, a plurality of the unique identification vectors are classified to obtain a plurality of optimal sets, classification points are generated based on the optimal sets, the classification points correspond to a plurality of the unique identification vectors, and an index tree is constructed based on the category set, the classification points and the unique identification vectors; The step of generating classification points based on the optimal set is specifically: Obtain the updated weight values ​​of all the unique identification vectors in the optimal set, and determine whether the updated weight values ​​at the same position are the same from left to right. If the updated weight values ​​at the same position are the same, extract the updated weight value as a classification weight value, and combine several classification weight values ​​into the classification point; When collecting local data, extracting a first keyword of the local data, encrypting the local data into ciphertext data, storing the ciphertext data, and generating a link path corresponding to the reference keyword, and associating the link path to the unique identification vector through a corresponding relationship between the first keyword and the reference keyword; When the user performs data retrieval, a second keyword input by the user is obtained, a target set is selected from the plurality of category sets based on the second keyword, a guidance vector is generated through the second keyword, a target point is selected from the plurality of classification points based on the guidance vector, and a target identification vector is determined from the plurality of unique identification vectors corresponding to the target point to obtain the link path.

2. The cloud data collection method based on index tree according to claim 1, characterized in that: The step of performing byte splitting on the benchmark keyword to obtain a plurality of benchmark feature vectors corresponding to the benchmark keyword comprises: The benchmark keyword is split once in single-byte units to obtain a plurality of single-byte features; The benchmark keyword is split twice in double-byte units to obtain a plurality of double-byte features; Hash mapping is performed on the single-byte feature and the double-byte feature respectively to obtain a plurality of reference feature vectors.

3. The cloud data collection method based on index tree according to claim 1, characterized in that: The step of mapping the vector values ​​at the same position in the plurality of reference feature vectors into weight component values ​​according to a preset mapping rule is specifically as follows: Determine whether the vector value at the same position in a plurality of the reference feature vectors is 1; If the vector value is 1, a sub-weight value of 1 is generated; If the vector value is not 1, a sub-weight value of -1 is generated; Superimposing a plurality of said sub-weight values ​​into a weight component value; The step of comparing the weight component value with the weight threshold to obtain an updated weight value is specifically as follows: Determine whether the weight component value is less than 0; If the weight component value is less than 0, an updated weight value of 0 is generated; If the weight component value is greater than or equal to 0, an updated weight value of 1 is generated.

4. The cloud data collection method based on index tree according to claim 1, characterized in that: The step of classifying the plurality of unique identification vectors to obtain a plurality of optimal sets comprises: Selecting a first number of first reference vectors from the plurality of unique identification vectors, and dividing the plurality of unique identification vectors into a plurality of initial sets according to the Euclidean distance between the first reference vectors and other unique identification vectors; setting one of the initial sets as a reference set, segmenting the updated weight value of the first reference vector in the reference set into an optimal value and a plurality of first neighboring values, and determining the updated weight value of the first reference vector in the other initial sets as a plurality of second neighboring values; Obtaining a classification evaluation sub-coefficient of the benchmark set through the optimal value, the first neighboring value, and the second neighboring value, determining a total evaluation coefficient through a plurality of the classification evaluation sub-coefficients, and iteratively updating the first quantity through the total evaluation coefficient to obtain an optimal quantity; The optimal number of second reference vectors is selected from the plurality of unique identification vectors, and the plurality of unique identification vectors are divided into a plurality of optimal sets according to the Euclidean distance between the second reference vector and other unique identification vectors.

5. The cloud data collection method based on index tree according to claim 4 is characterized in that: The calculation formula of the classification evaluation sub-coefficient is: ， in, It represents the classification evaluation sub-coefficient of the benchmark set when the i-th initial set is set as the benchmark set. It indicates the minimum distance between the optimal value and several second neighboring values ​​when the i-th initial set is set as the benchmark set and the j-th updated weight value in the first benchmark vector in the benchmark set is the optimal value. It represents the average distance between the optimal value and several first neighboring values ​​when the i-th initial set is set as the benchmark set and the j-th updated weight value in the first benchmark vector in the benchmark set is the optimal value.

6. The cloud data collection method based on index tree according to claim 1, characterized in that: The step of determining a target identification vector from a plurality of unique identification vectors corresponding to the target point comprises: Selecting a plurality of the unique identification vectors corresponding to the target point as vectors to be selected; Respectively obtain similarity values ​​of a plurality of the candidate vectors and the guide vector; A maximum similarity value is determined from the plurality of similarity values, and the candidate vector corresponding to the maximum similarity value is determined as a target identification vector.

7. A cloud data collection system based on an index tree, applied to the cloud data collection method based on an index tree as claimed in any one of claims 1 to 6, characterized in that: The system comprises: A first construction module is used to construct a plurality of category sets, each of which corresponds to a plurality of benchmark keywords; An analysis module, configured to perform byte splitting on the benchmark keyword to obtain a plurality of benchmark feature vectors corresponding to the benchmark keyword, wherein the plurality of benchmark feature vectors all include the same number of vector values, and to generate a unique identification vector corresponding to the benchmark keyword based on the plurality of benchmark feature vectors, wherein the unique identification vector includes a plurality of update weight values; The analysis module comprises: A fourth unit is used to map the vector values ​​at the same position in a plurality of the reference feature vectors into weight component values ​​according to a preset mapping rule; A fifth unit, configured to compare the weight component value with a weight threshold to obtain an updated weight value, and combine a plurality of the updated weight values ​​into the unique identification vector; A second construction module is used to classify a plurality of the unique identification vectors in the category set to obtain a plurality of optimal sets, generate classification points based on the optimal sets, the classification points correspond to a plurality of the unique identification vectors, and construct an index tree based on the category set, the classification points and the unique identification vectors; The second building block comprises: A tenth unit is used to obtain the updated weight values ​​of all the unique identification vectors in the optimal set, and determine whether the updated weight values ​​at the same position are the same from left to right. If the updated weight values ​​at the same position are the same, the updated weight value is extracted as a classification weight value, and a plurality of the classification weight values ​​are combined into the classification point; an associating module, configured to extract a first keyword of the local data when collecting local data, encrypt the local data into ciphertext data, store the ciphertext data, generate a link path corresponding to the reference keyword, and associate the link path with the unique identification vector through a corresponding relationship between the first keyword and the reference keyword; The execution module is used to obtain a second keyword input by the user when the user performs data retrieval, select a target set from the plurality of category sets according to the second keyword, generate a guidance vector through the second keyword, select a target point from the plurality of classification points based on the guidance vector, and determine a target identification vector from the plurality of unique identification vectors corresponding to the target point to obtain the link path.

8. A computer comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, characterized in that: When the processor executes the computer program, the index tree-based cloud data collection method according to any one of claims 1 to 6 is implemented.

Images