CN109934628A - Characteristic processing method and device - Google Patents
Characteristic processing method and device Download PDFInfo
- Publication number
- CN109934628A CN109934628A CN201910176079.6A CN201910176079A CN109934628A CN 109934628 A CN109934628 A CN 109934628A CN 201910176079 A CN201910176079 A CN 201910176079A CN 109934628 A CN109934628 A CN 109934628A
- Authority
- CN
- China
- Prior art keywords
- feature
- type
- data segments
- characteristic processing
- cryptographic hash
- 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.)
- Granted
Links
Landscapes
- Storage Device Security (AREA)
- Information Retrieval, Db Structures And Fs Structures Therefor (AREA)
Abstract
This disclosure relates to a kind of characteristic processing method and devices.According to present disclosure embodiment, this method comprises: being object distribution object ID；For the type distribution type ID of the feature of object；Using and the corresponding operator of type generate cryptographic Hash corresponding with the feature of object；And object-based object ID, type ID and cryptographic Hash generate the signature for indicating the feature of object.The method and device of present disclosure at least has one of following advantageous effects: the ability to express of feature-rich vector, and guarantees that characteristic processing process has preferable ease for use and scalability.
Description
Technical field
This disclosure relates to computer technologies, and in particular, to a kind of characteristic processing method and device.
Background technique
Characteristic processing is important a part during model training and service arrangement, can be understood as from
The attribute values such as family, context, article (Category Attributes or connection attribute) arrive the mapping process of several feature vectors.Single spy
Sign is easier to indicate, but during model construction, needs to have a variety of different types using a large amount of attribute value,
Finally formed vector dimension may be ten thousand or even hundred million this ranks, cause to exist in terms of the characteristics of management and processing feature certain
Difficulty.During characteristic processing and service arrangement, to the consistency of strict guarantee data handling procedure, need using tight
Therefore the consistent logic module of lattice works as newly-increased feature to carry out this operation, such as using common class libraries or same set of code
When generally require code intervention, cause ease for use limited with scalability.
Summary of the invention
It will be given for the brief overview of present disclosure, hereinafter in order to provide certain sides about present disclosure
The basic comprehension in face.It should be appreciated that this general introduction is not the exhaustive general introduction about present disclosure.It is not intended to determine
The key or pith of present disclosure, nor intended limitation scope of the present disclosure.Its purpose is only with simplification
Form provide certain concepts, taking this as a prelude to a more detailed description discussed later.
According to present disclosure in a first aspect, providing a kind of characteristic processing method, comprising:
For object distribution object ID；
For the type distribution type ID of the feature of object；
Using and the corresponding operator of type generate cryptographic Hash corresponding with the feature of object；And
Object-based object ID, type ID and cryptographic Hash generate the signature for indicating the feature of object.
According to the second aspect of the present disclosure, a kind of characteristic processing device is provided, comprising:
First allocation unit is configured to object distribution object ID；
Second allocation unit is configured to the type distribution type ID of the feature of object；
Computing unit, be configured to using and the corresponding operator of type generate cryptographic Hash corresponding with the feature of object；With
And
Generation unit is configured to object-based object ID, type ID and cryptographic Hash and generates the feature for indicating object
Signature.
According to the third aspect of present disclosure, a kind of storage medium is provided, is stored thereon with and realizes above-mentioned first aspect
Characteristic processing method program.
The technical solution of present disclosure one of at least has the following technical effects: enriching the expression energy of feature vector
Power, and guarantee that characteristic processing process has preferable ease for use and scalability.
Detailed description of the invention
The disclosure can by reference to being better understood below in association with description given by attached drawing, attached drawing together with
It is following to be described in detail together comprising in the present specification and forming a part of this specification.In the accompanying drawings:
Fig. 1 is the flow diagram of characteristic processing method according to one embodiment of the present disclosure；And
Fig. 2 is the structural block diagram of characteristic processing device according to one embodiment of the present disclosure.
Specific embodiment
It is described hereinafter in connection with exemplary embodiment of the attached drawing to present disclosure.It rises for clarity and conciseness
See, does not describe all features of practical embodiments in the description.It should be understood, however, that any this practical real developing
Much decisions specific to embodiment can be made during applying example, to realize the objectives of developer, and this
It is a little to determine to change with the difference of embodiment.
Here, and also it should be noted is that, in order to avoid having obscured present disclosure because of unnecessary details, attached
Illustrate only in figure with the apparatus structure closely related according to the scheme of present disclosure, and be omitted and present disclosure close
It is little other details.
It should be understood that present disclosure is not compromised by the following description referring to attached drawing and is only limited to described implementation
Form.Herein, in feasible situation, embodiment be can be combined with each other, the feature replacement between different embodiments or borrow
With, omit one or more features in one embodiment.
During being trained to model and service arrangement, the operation of characteristic processing is inevitably carried out, it is right
In with different attribute value and different types of feature, the vector dimension formed can reach ten thousand or even hundred million rank；Especially
When being newly-increased feature, code is needed to intervene, causes the ease for use of characteristic processing and scalability limited.
According to one embodiment of the present disclosure, different characteristic is divided according to object, right as unit of object
Feature therein carries out characteristic processing, for each feature, forms the corresponding cryptographic Hash of this feature；Based on object ID, type ID
The corresponding signature of feature is formed with cryptographic Hash, and the above process makes characteristic processing process have preferable ease for use and expansible
Property.
Specifically, Fig. 1 is the flow diagram of characteristic processing method 100 according to one embodiment of the present disclosure.
It is object distribution object ID at step S101.
Wherein, object can be that user object, audience and content object etc. will be special according to the attribute of feature difference
Sign is divided into different objects；Such as the feature with this attribute of user is divided into user object, with this attribute of advertisement
Feature is divided into audience, and the feature with this attribute of content is divided into content characteristic, wherein has this attribute of user
Feature include the types such as gender, age and height feature, include advertisement type, advertisement with the feature of this attribute of advertisement
The feature of the types such as duration and advertising source, the feature with this attribute of content includes content type, content region and content
The feature of the types such as time.Object ID is indicated with block id, for example, the block id=1 of user object, audience
Block id=2, the block id=3 of content object.
It is the type distribution type ID of the feature of object at step S102.
One object has the feature of at least one type, and the feature for such as belonging to audience includes gender, age and body
The feature of high three types, the feature distribution type ID of respectively each type, type ID is indicated with slot id, for example, gender
The slot id=1 of feature, the slot id=2 of age characteristics, the slot id=3 of height feature.
At step S103, using and the corresponding operator of type generate cryptographic Hash corresponding with the feature of object；
At step S104, object-based object ID, type ID and cryptographic Hash generate the signature of the feature of object.
The present embodiment divides different characteristic according to object, as unit of object, carries out feature to feature therein
Processing forms the corresponding cryptographic Hash of this feature for each feature；Feature pair is formed based on object ID, type ID and cryptographic Hash
The signature answered, the above process make characteristic processing process have preferable ease for use and scalability.
For example, using and the corresponding operator of type generate cryptographic Hash corresponding with the feature of object and include:
One-dimensional rarefaction is carried out using value of the operator corresponding with type to feature, obtains feature vector；To feature to
Amount carries out Hash calculation, obtains cryptographic Hash.In the present embodiment, the process of one-dimensional rarefaction and Hash calculation dependent on configuration and is calculated
Son, configuration can specify an operator corresponding with the type, and taking the feature in the type for the feature of each type
Value and corresponding parameter input to operator, and the value of feature obtains cryptographic Hash after the calculating of operator.
Carrying out one-dimensional rarefaction using value of the operator corresponding with type to feature for example can be one-dimensional using one-hot
The mode of rarefaction carries out one-dimensional rarefaction to feature, to obtain feature vector.The one-dimensional rarefaction of one-hot i.e. solely compile by heat
Code, for each feature, if there are m probable values for this feature, then having reformed into m after the one-dimensional rarefaction of one-hot
A binary feature, and these feature mutual exclusions, only one activation, so that data become sparse, rises to a certain extent every time
The effect of augmented features is arrived.It should be noted that can also be used when carrying out one-dimensional rarefaction to feature in this disclosure
The method of other rarefactions, however it is not limited to the one-dimensional rarefaction of one-hot.
Carrying out Hash calculation for example to the corresponding feature vector of the value of feature can be carried out using murmur hash method
Hash calculation obtains the corresponding cryptographic Hash of this feature, which is the integer of 64 bits, while obtaining the corresponding spy of this feature
The value in vector is levied, which is floating-point values, and the value in cryptographic Hash and the corresponding feature vector of this feature forms numerical value
It is right.For example, for feature (user A, gender male) carrying out that cryptographic Hash-is calculated to feature vector using murmur Hash
2366580462178760140, while obtaining the value in the corresponding feature vector of this feature is 1.0, then the numerical value pair exported
For (- 2366580462178760140,1.0).
The signature for the feature that object-based object ID, type ID and cryptographic Hash generate object for example may include:
The binary number of the first presetting digit capacity is continuously extracted since the lowest order of the binary representation of object ID as the
One data segments；
The binary number of the second presetting digit capacity is continuously extracted since the lowest order of the binary representation of type ID as the
Two data segments；
The binary number of third presetting digit capacity is continuously extracted since the lowest order of the binary representation of cryptographic Hash as the
Three data segments；And
First data segments, the second data segments and third data segments are spliced in a predetermined order as signature.Example
Such as, predetermined order can be used as signature successively to splice the first data segments, the second data segments, third data segments.
The sum of first presetting digit capacity, the second presetting digit capacity and third presetting digit capacity are equal to 64, and third presetting digit capacity is greater than
32.For example, the first presetting digit capacity is 6, the second presetting digit capacity is 10, and third presetting digit capacity is 48.
In a schematic example, the corresponding signature of feature includes the number of 64 bits, and 6 bits of starting are the marks of object
Know, intermediate 10 bits are the marks of type, and 48 final bits are the hash spaces of signature, can express about 280,000,000,000,000
Hash space, cope with the value hash space of most of type feature.For example, for feature (user A, gender male),
Object ID is 1, and type ID is 1, is number -2366580462178760140 and value 1.0 by value " male " Hash, will be right
As rear 6 digital bit of the binary representation of ID, rear 10 digital bit and number-of the binary representation of type ID
2366580462178760140 rear 48 digital bit is successively spliced into signature 288854468109568564.It finally can be defeated
Numerical value expresses (288854468109568564,1.0) this male feature of gender out.For every sample, including multiple spies
Sign can obtain a series of signature after above scheme to express the sample, the object of its character pair is contained in signature
Information, type information and discrete value information, so that the length of feature vector is extended to 10,000,000,000 or even hundred billion or more, significantly
Enrich the ability to express of feature vector.
In the examples described above, due to being extracted rear 6 digital bit of object ID, rear 10 digital bit of type ID, therefore,
The value range of object ID is 0~63, and the value range of type ID is 0~1023.
In this disclosure, it can be a configuration file by characteristic processing map procedures, can be convenient deployment and change
Generation.There is versatility for the calculating of the feature of each type, also promote the reusability of code, reduce the complexity of code intervention
Property；Meanwhile the aforesaid way specification process of characteristic processing.
The second aspect of present disclosure also provides a kind of characteristic processing device.Fig. 2 is provided according to present disclosure
The structural block diagram of characteristic processing device 200 in one embodiment.As shown in Figure 2, characteristic processing device 200 includes: first
Allocation unit 201, the second allocation unit 202, computing unit 203 and generation unit 204.
First allocation unit 201, for being object distribution object ID.
Wherein, object can be that user object, audience and content object etc. will be special according to the attribute of feature difference
Sign is divided into different objects；Such as the feature with this attribute of user is divided into user object, with this attribute of advertisement
Feature is divided into audience, and the feature with this attribute of content is divided into content characteristic, wherein has this attribute of user
Feature include the types such as gender, age and height feature, include advertisement type, advertisement with the feature of this attribute of advertisement
The feature of the types such as duration and advertising source, the feature with this attribute of content includes content type, content region and content
The feature of the types such as time.Object ID is indicated with block id, for example, the block id=1 of user object, audience
Block id=2, the block id=3 of content object.
Second allocation unit 202, the type distribution type ID for the feature for object.
One object has the feature of at least one type, and the feature for such as belonging to audience includes gender, age and body
The feature of high three types, the feature distribution type ID of respectively each type, type ID is indicated with slot id, for example, gender
The slot id=1 of feature, the slot id=2 of age characteristics, the slot id=3 of height feature.
Computing unit 203, for using and the corresponding operator of type generates cryptographic Hash corresponding with the feature of object.
Generation unit 204 generates the signature of the feature of object for object-based object ID, type ID and cryptographic Hash.
The present embodiment divides different characteristic according to object, as unit of object, carries out feature to feature therein
Processing forms the corresponding cryptographic Hash of this feature for each feature；Feature pair is formed based on object ID, type ID and cryptographic Hash
The signature answered, the above process make characteristic processing process have preferable ease for use and scalability.
For example, computing unit 203 is used for:
One-dimensional rarefaction is carried out using value of the operator corresponding with type to feature, obtains feature vector；To feature to
Amount carries out Hash calculation, obtains cryptographic Hash.
In the present embodiment, the process of one-dimensional rarefaction and Hash calculation can be each dependent on configuration and operator, configuration
The feature of type specifies an operator corresponding with the type, and the value of the feature in the type and corresponding parameter are inputted
To operator, the value of feature obtains cryptographic Hash after the calculating of operator.
For example, generation unit 204 is used for:
The binary number of the first presetting digit capacity is continuously extracted since the lowest order of the binary representation of object ID as the
One data segments；
The binary number of the second presetting digit capacity is continuously extracted since the lowest order of the binary representation of type ID as the
Two data segments；
The binary number of third presetting digit capacity is continuously extracted since the lowest order of the binary representation of cryptographic Hash as the
Three data segments；And
First data segments, the second data segments and third data segments are spliced in a predetermined order as signature.Example
Such as, predetermined order can be used as signature successively to splice the first data segments, the second data segments, third data segments.
The sum of first presetting digit capacity, the second presetting digit capacity and third presetting digit capacity are equal to 64, and third presetting digit capacity is greater than
32.For example, the first presetting digit capacity is 6, the second presetting digit capacity is 10, and third presetting digit capacity is 48.
In a schematic example, the corresponding signature of feature includes the number of 64 bits, and 6 bits of starting are the marks of object
Know, intermediate 10 bits are the marks of type, and 48 final bits are the hash spaces of signature, can express about 280,000,000,000,000
Hash space, cope with the value hash space of most of type feature.For example, for feature (user A, gender male),
Object ID is 1, and type ID is 1, is number -2366580462178760140 and value 1.0 by value " male " Hash, will be right
As rear 6 digital bit of the binary representation of ID, rear 10 digital bit and number-of the binary representation of type ID
2366580462178760140 rear 48 digital bit is successively spliced into signature 288854468109568564.It finally can be defeated
Numerical value expresses (288854468109568564,1.0) this male feature of gender out.
In terms of present disclosure third, a kind of storage medium is provided, is stored thereon with for realizing above-mentioned first aspect
Characteristic processing method program
Illustratively, any usable medium that computer readable storage medium can be that computer can access either is wrapped
The data storage devices such as server, the data center integrated containing one or more usable mediums.The usable medium can be magnetism
Medium, (for example, floppy disk, hard disk, tape), optical medium (for example, DVD) or semiconductor medium are (for example, solid state hard disk Solid
State Disk (SSD)) etc.).
In several embodiments provided by present disclosure, it should be understood that disclosed device and method, Ke Yitong
Other modes are crossed to realize.The apparatus embodiments described above are merely exemplary, for example, the division of module or unit,
Only a kind of logical function partition, there may be another division manner in actual implementation, for example, in feasible situation, it is more
A unit or assembly can be combined or can be integrated into another system, or some features can be ignored or not executed.
In addition, each functional unit in each embodiment of present disclosure can integrate in one processing unit,
It can be each unit to physically exist alone, can also be integrated in one unit with two or more units.It is above-mentioned integrated
Unit both can take the form of hardware realization, can also realize in the form of software functional units.
The following one or more technical effect of this method example implementation:
1, Features Management function: feature vector is abstracted as a series of numerical value pair, each numerical value compares comprising one 64
Special integer (referred to as signing) and a floating-point values, for example numerical value is to (24343434323423423,1.0),
24343434323423423 be the signature of some position inside feature vector, and 1.0 be some position inside feature vector
Specific value.In this way, the length of feature vector has just been extended to 10,000,000,000 or even hundred billion or more, feature vector has been greatly enriched
Ability to express.
2, plug-in unitization is realized: being an engine by feature extraction procedural abstraction, engine executes operator, and operator and parameter are
It is obtained by reading configuration, that is to say, that engine defines to execute by reading configuration extracts logic accordingly, in this way, tool
Body extract which type of feature just only it is related to configuration, facilitate feature change and iteration.
3, the concept mark input of block is whole: in deep learning model, generally requires multiple feature vectors and be used as input,
The concept of block (block) is introduced, to identify the entirety of an input, each piece may include multiple slot positions (slot), this is whole
Body can be used as a feature vector and carry out input model.For example, the feature of user can integrally be used as a block, and property therein
Other feature is one of slot position.Inside the input of model, we can have multiple pieces, such as user characteristics, content
Feature, characteristic of advertisement etc., each section can be expressed as an independent feature vector.
4, introduce the concept of slot position (slot): a slot position indicates a feature slot position, for example gender is exactly a slot position,
But the corresponding value of gender is specific signature (sign), and final signature is the number of 64 bits, can be divided into three
Part: 6 bits of starting are the marks of block (block), so the range of block ID needs in the range of 0 to 63；And then 10 ratio
Spy is the mark of slot position (slot), so the range of slot position ID needs in the range of 0 to 1023；48 final bits are signatures
Hash space, about 280,000,000,000,000 hash space can be expressed, it is sufficient to which the value for coping with most of slot position feature is breathed out
Uncommon space, says that crash space is minimum from probability.
More than, although describing multiple embodiments of present disclosure, scope of the present disclosure be not limited to this
A little embodiments, anyone skilled in the art can readily occur in change without departing from present disclosure
Change or replace, scheme corresponding to these change or replacement should all cover within the protection scope of present disclosure.Therefore, originally
The protection scope of disclosure should be subject to the claims.
Claims (10)
1. a kind of characteristic processing method, comprising:
For object distribution object ID；
For the type distribution type ID of the feature of the object；
Using and the corresponding operator of the type generate cryptographic Hash corresponding with the feature of the object；And
The object ID, the type ID and the cryptographic Hash based on the object generate the label for indicating the feature of the object
Name.
2. characteristic processing method according to claim 1, wherein corresponding with the type operator of use generate with it is described
The corresponding cryptographic Hash of the feature of object, comprising:
One-dimensional rarefaction is carried out using value of the operator corresponding with the type to the feature, obtains feature vector；And
Hash calculation is carried out to described eigenvector, obtains the cryptographic Hash.
3. characteristic processing method according to claim 2, wherein the one-dimensional rarefaction includes onehot one-dimensional sparse
Change.
4. characteristic processing method according to claim 2, wherein the Hash calculation includes murmur Hash calculation.
5. characteristic processing method according to claim 1, wherein the value range of the object ID is 0~63, the class
The value range of type ID is 0~1023.
6. characteristic processing method described in any one of -6 according to claim 1, wherein object ID based on the object,
Type ID and cryptographic Hash generation indicate that the signature of the feature of the object includes:
The binary number of the first presetting digit capacity is continuously extracted since the lowest order of the binary representation of the object ID as the
One data segments；
The binary number of the second presetting digit capacity is continuously extracted since the lowest order of the binary representation of the type ID as the
Two data segments；
The binary number of third presetting digit capacity is continuously extracted since the lowest order of the binary representation of the cryptographic Hash as the
Three data segments；And
First data segments, second data segments and the third data segments are spliced into conduct in a predetermined order
The signature.
7. characteristic processing method according to claim 6, wherein
The sum of first presetting digit capacity, second presetting digit capacity and described third presetting digit capacity be equal to 64, and
The third presetting digit capacity is greater than 32.
8. a kind of characteristic processing device, comprising:
First allocation unit is configured to object distribution object ID；
Second allocation unit is configured to the type distribution type ID of the feature of the object；
Computing unit, be configured to using and the corresponding operator of the type generate Hash corresponding with the feature of the object
Value；And
Generation unit is configured to the object ID, the type ID and the cryptographic Hash based on the object and generates expression
The signature of the feature of the object.
9. characteristic processing device according to claim 8, wherein the generation unit is configured to:
The binary number of the first presetting digit capacity is continuously extracted since the lowest order of the binary representation of the object ID as the
One data segments；
The binary number of the second presetting digit capacity is continuously extracted since the lowest order of the binary representation of the type ID as the
Two data segments；
The binary number of third presetting digit capacity is continuously extracted since the lowest order of the binary representation of the cryptographic Hash as the
Three data segments；And
First data segments, second data segments and the third data segments are spliced into conduct in a predetermined order
The signature.
10. a kind of storage medium is stored thereon with for realizing characteristic processing method described in any one of claim 1-6
Program.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201910176079.6A CN109934628B (en) | 2019-03-08 | 2019-03-08 | Feature processing method and device |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201910176079.6A CN109934628B (en) | 2019-03-08 | 2019-03-08 | Feature processing method and device |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN109934628A true CN109934628A (en) | 2019-06-25 |
| CN109934628B CN109934628B (en) | 2021-03-19 |
Family
ID=66986525
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201910176079.6A Active CN109934628B (en) | 2019-03-08 | 2019-03-08 | Feature processing method and device |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN109934628B (en) |
Citations (10)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN103810551A (en) * | 2012-11-09 | 2014-05-21 | 金蝶软件（中国）有限公司 | Organizational unit object generation method and device in ERP system |
| KR20160112862A (en) * | 2015-03-20 | 2016-09-28 | 이영희 | Slot machine game providing system based on user mobile device, and method thereof |
| CN106407201A (en) * | 2015-07-29 | 2017-02-15 | 腾讯科技（深圳）有限公司 | Data processing method and apparatus |
| CN107025233A (en) * | 2016-01-29 | 2017-08-08 | 苏宁云商集团股份有限公司 | A kind of processing method and processing device of data characteristics |
| CN107291454A (en) * | 2017-05-27 | 2017-10-24 | 北京思特奇信息技术股份有限公司 | A kind of method and Commentary Systems that comment is added in the comment list of event |
| CN108108436A (en) * | 2017-12-20 | 2018-06-01 | 东软集团股份有限公司 | Date storage method, device, storage medium and electronic equipment |
| CN109146083A (en) * | 2018-08-06 | 2019-01-04 | 阿里巴巴集团控股有限公司 | Feature coding method and apparatus |
| CN109409030A (en) * | 2018-10-19 | 2019-03-01 | 微梦创科网络科技（中国）有限公司 | A kind of digital publishing rights generation method and device |
| AU2017248417B2 (en) * | 2009-06-26 | 2019-06-06 | CounterTack, Inc. | Fuzzy hash algorithm |
| US10540338B2 (en) * | 2017-01-30 | 2020-01-21 | Alfresco Software, Inc. | Scalable fine grained access control within a search engine |
-
2019
- 2019-03-08 CN CN201910176079.6A patent/CN109934628B/en active Active
Patent Citations (10)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| AU2017248417B2 (en) * | 2009-06-26 | 2019-06-06 | CounterTack, Inc. | Fuzzy hash algorithm |
| CN103810551A (en) * | 2012-11-09 | 2014-05-21 | 金蝶软件（中国）有限公司 | Organizational unit object generation method and device in ERP system |
| KR20160112862A (en) * | 2015-03-20 | 2016-09-28 | 이영희 | Slot machine game providing system based on user mobile device, and method thereof |
| CN106407201A (en) * | 2015-07-29 | 2017-02-15 | 腾讯科技（深圳）有限公司 | Data processing method and apparatus |
| CN107025233A (en) * | 2016-01-29 | 2017-08-08 | 苏宁云商集团股份有限公司 | A kind of processing method and processing device of data characteristics |
| US10540338B2 (en) * | 2017-01-30 | 2020-01-21 | Alfresco Software, Inc. | Scalable fine grained access control within a search engine |
| CN107291454A (en) * | 2017-05-27 | 2017-10-24 | 北京思特奇信息技术股份有限公司 | A kind of method and Commentary Systems that comment is added in the comment list of event |
| CN108108436A (en) * | 2017-12-20 | 2018-06-01 | 东软集团股份有限公司 | Date storage method, device, storage medium and electronic equipment |
| CN109146083A (en) * | 2018-08-06 | 2019-01-04 | 阿里巴巴集团控股有限公司 | Feature coding method and apparatus |
| CN109409030A (en) * | 2018-10-19 | 2019-03-01 | 微梦创科网络科技（中国）有限公司 | A kind of digital publishing rights generation method and device |
Non-Patent Citations (1)
| Title |
|---|
| 范远超 等: "FDSS：基于HDFS的海量音乐特征数据存储***", 《计算机研究与发展》 * |
Also Published As
| Publication number | Publication date |
|---|---|
| CN109934628B (en) | 2021-03-19 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| KR102667837B1 (en) | High-precision privacy-preserving real-value function evaluation | |
| Bressert | SciPy and NumPy: an overview for developers | |
| Deutsch | A storeless model of aliasing and its abstractions using finite representations of right-regular equivalence relations | |
| Marin et al. | Bayesian essentials with R | |
| CN107562775A (en) | A kind of data processing method and equipment based on block chain | |
| CN110825363B (en) | Intelligent contract acquisition method and device, electronic equipment and storage medium | |
| US9860063B2 (en) | Code analysis tool for recommending encryption of data without affecting program semantics | |
| Aharoni et al. | Helayers: A tile tensors framework for large neural networks on encrypted data | |
| CN105706092A (en) | Methods and systems of four-valued simulation | |
| WO2020003109A1 (en) | Facet-based query refinement based on multiple query interpretations | |
| Bürgisser et al. | Fundamental invariants of orbit closures | |
| CN111488277A (en) | Node matching method, device, equipment and system | |
| US11775517B2 (en) | Query content-based data generation | |
| CN109271526A (en) | Method for text detection, device, electronic equipment and computer readable storage medium | |
| CN111221690A (en) | Model determination method and device for integrated circuit design and terminal | |
| CN109934628A (en) | Characteristic processing method and device | |
| Gupta et al. | Continuous reachability for unordered data Petri nets is in PTime | |
| Bonnet et al. | Ordinal theory for expressiveness of well-structured transition systems | |
| US20220405474A1 (en) | Method, computing device and computer-readable medium for classification of encrypted data using neural network | |
| US11481212B2 (en) | Automatic identification of reference data | |
| CN113887201A (en) | Text fixed-length error correction method, device, equipment and storage medium | |
| Hampe | Algorithmic aspects of tropical intersection theory | |
| Cortes et al. | On-line learning algorithms for path experts with non-additive losses | |
| Molteni et al. | A relation calculus for reasoning about t-probing security | |
| CN111723247A (en) | Graph-based hypothetical computation |
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 |