CN110321546A - Account identification, display methods, device, server, terminal and storage medium - Google Patents

Account identification, display methods, device, server, terminal and storage medium Download PDF

Info

Publication number
CN110321546A
CN110321546A CN201810288695.6A CN201810288695A CN110321546A CN 110321546 A CN110321546 A CN 110321546A CN 201810288695 A CN201810288695 A CN 201810288695A CN 110321546 A CN110321546 A CN 110321546A
Authority
CN
China
Prior art keywords
account
dispatch
information
identified
sequence
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
Application number
CN201810288695.6A
Other languages
Chinese (zh)
Other versions
CN110321546B (en
Inventor
卓居超
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.)
Shenzhen Tencent Computer Systems Co Ltd
Original Assignee
Shenzhen Tencent Computer Systems 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 Shenzhen Tencent Computer Systems Co Ltd filed Critical Shenzhen Tencent Computer Systems Co Ltd
Priority to CN201810288695.6A priority Critical patent/CN110321546B/en
Publication of CN110321546A publication Critical patent/CN110321546A/en
Application granted granted Critical
Publication of CN110321546B publication Critical patent/CN110321546B/en
Active legal-status Critical Current
Anticipated expiration legal-status Critical

Links

Classifications

    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06F—ELECTRIC DIGITAL DATA PROCESSING
    • G06F40/00—Handling natural language data
    • G06F40/20—Natural language analysis
    • G06F40/279—Recognition of textual entities
    • H—ELECTRICITY
    • H04—ELECTRIC COMMUNICATION TECHNIQUE
    • H04L—TRANSMISSION OF DIGITAL INFORMATION, e.g. TELEGRAPHIC COMMUNICATION
    • H04L63/00—Network architectures or network communication protocols for network security
    • H04L63/12—Applying verification of the received information
    • H04L63/126—Applying verification of the received information the source of the received data

Landscapes

  • Engineering & Computer Science (AREA)
  • General Engineering & Computer Science (AREA)
  • Computer Security & Cryptography (AREA)
  • Theoretical Computer Science (AREA)
  • General Physics & Mathematics (AREA)
  • General Health & Medical Sciences (AREA)
  • Physics & Mathematics (AREA)
  • Computational Linguistics (AREA)
  • Health & Medical Sciences (AREA)
  • Audiology, Speech & Language Pathology (AREA)
  • Artificial Intelligence (AREA)
  • Computer Hardware Design (AREA)
  • Computing Systems (AREA)
  • Computer Networks & Wireless Communication (AREA)
  • Signal Processing (AREA)
  • Information Retrieval, Db Structures And Fs Structures Therefor (AREA)
  • Management, Administration, Business Operations System, And Electronic Commerce (AREA)

Abstract

The embodiment of the present application provides a kind of account identification, display methods, device, server, terminal and storage medium, passes through determination dispatch account to be identified；Obtain the first information relevant to the dispatch dispatch behavior of account to be identified；Dispatch account to be identified is identified using the first information, obtain the mode of the recognition result of dispatch account to be identified, so that realizing the identification to dispatch account to be identified based on the first information relevant to the dispatch dispatch behavior of account to be identified, when reducing the prior art and realizing the identification to dispatch account to be identified according to the recognition result individually to every article in dispatch account to be identified, the caused recognition result inaccuracy to dispatch account to be identified, cannot effectively identify rubbish send the documents account the problem of.

Description

Account identification, display methods, device, server, terminal and storage medium
Technical field
The present invention relates to computer classes technical fields, identification, display side more specifically to a kind of account of sending the documents Method, device, server, terminal and storage medium.
Background technique
With the development of social networks, dispatch account (for example, public platform etc.) is had been to be concerned by more and more people, Ren Menke Meet the reading requirement of oneself to read the article in different dispatch accounts according to the interest of oneself selection.Although however, hair Literary account can satisfy the reading requirement of user, but is also no lack of some dispatch accounts and there is transmission rubbish article (for example, carrying The article of the junk information such as SEO marketing, pornographic gambling, two dimensional code popularization), and then lead to the problem of user's reading experience difference.
Although the prior art can realize the identification to rubbish article in dispatch account based on word frequency, because of rubbish text Chapter and rubbish dispatch account between have no direct relation (there are the dispatch account of rubbish article be not necessarily rubbish dispatch account, There is no the dispatch accounts of rubbish article to be not necessarily non-junk dispatch account), therefore the prior art can not effectively be realized to rubbish The identification of rubbish dispatch account.
In view of this, identification, display methods, device, server, terminal and the storage medium of a kind of account of sending the documents are provided, Effectively to realize to the identification of rubbish dispatch account, it is a problem to be solved.
Summary of the invention
In view of this, the present invention provides a kind of account identification, display methods, device, server, terminal and storage medium, Effectively to realize the identification to rubbish dispatch account.
To achieve the above object, the embodiment of the present invention provides the following technical solutions:
A kind of account recognition methods, comprising:
Determine dispatch account to be identified；
Obtain the first information relevant to the dispatch dispatch behavior of account to be identified；
The dispatch account to be identified is identified using the first information, obtains the dispatch account to be identified Recognition result.
A kind of account display methods, comprising:
Receive dispatch account display request；
Obtain each dispatch account of the dispatch account display request instruction；
Determine that dispatch account number classification belonging to each dispatch account, the dispatch account number classification are rubbish dispatch account Classification or non-junk dispatch account number classification；
Rule is shown based on dispatch account relevant to the dispatch account number classification, shows each dispatch account.
A kind of account identification device, comprising:
Dispatch account determination unit to be identified, for determining dispatch account to be identified；
First information acquiring unit is believed for obtaining to the dispatch dispatch behavior of account relevant first to be identified Breath；
Recognition result determination unit is obtained for being identified using the first information to the dispatch account to be identified To the recognition result of the dispatch account to be identified.
A kind of server, comprising: at least one processor and at least one processor；The memory is stored with program, The processor calls the program of the memory storage, and described program is for realizing the account recognition methods.
A kind of account display device, comprising:
Account of sending the documents shows request reception unit, for receiving dispatch account display request；
Dispatch account acquiring unit, for obtaining each dispatch account of the dispatch account display request instruction；
Dispatch account number classification determination unit, it is described for determining dispatch account number classification belonging to each dispatch account Account number classification of sending the documents is rubbish dispatch account number classification or non-junk dispatch account number classification；
Account display unit of sending the documents is shown for showing rule based on dispatch account relevant to the dispatch account number classification Show each dispatch account.
A kind of terminal, comprising: processor and memory, the processor and memory are connected by communication bus； Wherein, the processor, for calling and executing the program stored in the memory；The memory, for storing program, Described program is for realizing the account display methods.
A kind of storage medium, for storing the program for realizing the account recognition methods；Or the account is realized in storage The program of display methods.
The embodiment of the present application provides a kind of account identification, display methods, device, server, terminal and storage medium, passes through Determine dispatch account to be identified；Obtain the first information relevant to the dispatch dispatch behavior of account to be identified；Utilize the first information Dispatch account to be identified is identified, obtain it is to be identified dispatch account recognition result mode so that based on it is to be identified Dispatch account the relevant first information of dispatch behavior realize to it is to be identified dispatch account identification, reduce the prior art according to It is caused when realizing the identification to dispatch account to be identified according to the recognition result individually to every article in dispatch account to be identified To it is to be identified dispatch account recognition result inaccuracy, cannot effectively identify rubbish send the documents account the problem of.
Detailed description of the invention
In order to more clearly explain the embodiment of the invention or the technical proposal in the existing technology, to embodiment or will show below There is attached drawing needed in technical description to be briefly described, it should be apparent that, the accompanying drawings in the following description is only this The embodiment of invention for those of ordinary skill in the art without creative efforts, can also basis The attached drawing of offer obtains other attached drawings.
Fig. 1 is a kind of hardware block diagram of server provided by the embodiments of the present application；
Fig. 2 is a kind of identification model generation method flow chart provided by the embodiments of the present application；
Fig. 3 is a kind of side of acquisition provided by the embodiments of the present application the second information relevant to the dispatch account attribute of account Method flow chart；
Fig. 4 is a kind of method flow diagram of the term vector sequence of determining dispatch account provided by the embodiments of the present application；
Fig. 5 is a kind of term vector sequence inputting by article sequence provided by the embodiments of the present application to Recognition with Recurrent Neural Network mould Type handles term vector sequence based on Recognition with Recurrent Neural Network model, obtains the method flow diagram of the first identification information；
Fig. 6 be it is provided by the embodiments of the present application it is a kind of based on attention mechanism module to the expression of article sequence and second believe Breath is handled, and the method flow diagram of the second identification information is obtained；
Fig. 7 is a kind of dispatch account recognition methods flow chart provided by the embodiments of the present application；
Fig. 8 is another account recognition methods flow chart of sending the documents provided by the embodiments of the present application；
Fig. 9 is a kind of hardware block diagram of terminal provided by the embodiments of the present application；
Figure 10 is a kind of dispatch account display methods flow chart provided by the embodiments of the present application；
Figure 11 (a)~Figure 11 (e) is the application of dispatch account display methods in a kind of terminal provided by the embodiments of the present application Page schematic diagram；
Figure 12 is a kind of structural block diagram of account identification device of sending the documents provided by the embodiments of the present application；
Figure 13 is a kind of detailed construction schematic diagram of identification model generation unit provided by the embodiments of the present application；
Figure 14 is a kind of structural block diagram of account display device of sending the documents provided by the embodiments of the present application.
Specific embodiment
The dispatch account recognition methods of the application can be used for being identified to obtain recognition result to dispatch account, with basis Recognition result determines whether dispatch account is rubbish dispatch account, and then sends out convenient for the management to dispatch account, such as shielding rubbish Literary account etc..
One dispatch account can generally have an at least article delivered, although the prior art can pass through article In word frequency realize to dispatch account in rubbish article identification.But the dispatch account including rubbish article is not necessarily (e.g., a dispatch account only delivered the rubbish article of a marketing property to rubbish dispatch account, it would be possible that this dispatch account Can not directly be regarded as being rubbish dispatch account), the dispatch account without including rubbish article might not be rubbish (e.g., a dispatch account did not delivered rubbish article to account of sending the documents, but identical/similar there are continuously delivering more in one day Article the case where, then this dispatch account may be identified as rubbish dispatch account).
Therefore, based in the prior art individually to dispatch account in every article recognition result, can not effectively realize Identification to rubbish dispatch account.In order to solve this problem, present inventor proposes a kind of dispatch account recognition methods, The available first information relevant to the dispatch dispatch behavior of account to be identified of this method；And using the first information to be identified Dispatch account is identified, the recognition result of dispatch account to be identified is obtained, in order to the identification based on dispatch account to be identified As a result determine whether the dispatch account to be identified is rubbish dispatch account.
The embodiment of the present application, it is preferred that can using the article sequence that deliver of dispatch account to be identified as with hair to be identified The relevant first information of dispatch behavior of literary account come realize to this it is to be identified dispatch account identification so as to hair to be identified The recognition result of literary account can embody the timing information of article that dispatch account to be identified is delivered, and then can effectively realize pair The identification of rubbish dispatch account.For example, can in order to do not delivered rubbish article but continuously deliver more it is identical/similar The identification of the rubbish dispatch account of article.
It describes in detail below to a kind of dispatch account recognition methods provided by the embodiments of the present application:
A kind of dispatch account recognition methods provided by the embodiments of the present application can be applied to server, which can be net Network side provides the service equipment of service for user, may be the server cluster of multiple servers composition, it is also possible to separate unit Server.
Optionally, Fig. 1 shows the hardware block diagram of server, and referring to Fig.1, the hardware configuration of server can wrap It includes: processor 11, communication interface 12, memory 13 and communication bus 14；
In embodiments of the present invention, processor 11, communication interface 12, memory 13, communication bus 14 quantity can be with For at least one, and processor 11, communication interface 12, memory 13 complete mutual communication by communication bus 14；
Processor 11 may be a central processor CPU or specific integrated circuit ASIC
(Application Specific Integrated Circuit), or be arranged to implement of the invention real Apply one or more integrated circuits etc. of example；
Memory 13 may include high speed RAM memory, it is also possible to further include nonvolatile memory (non-volatile Memory) etc., a for example, at least magnetic disk storage；
Wherein, memory is stored with program, the program that processor can call memory to store, and program is used for:
Determine dispatch account to be identified；
Obtain the first information relevant to the dispatch dispatch behavior of account to be identified；
Dispatch account to be identified is identified using the first information, obtains the recognition result of dispatch account to be identified.
Optionally, the refinement function of program and extension function can refer to above description.
For the ease of to it is provided by the above embodiment it is a kind of send the documents account recognition methods understanding, now first to the dispatch During identifying to dispatch account, the generation method for the identification model being applied to carries out detailed for account recognition methods It introduces.
Optionally, the identification model in the embodiment of the present application can be by dispatch account identification model and attention mechanism mould Type is constituted.Wherein, account identification model of sending the documents and attention Mechanism Model are by training sample to by Recognition with Recurrent Neural Network model What the neural network constituted with attention mechanism module was trained；After the completion of neural metwork training, obtained instruction Recognition with Recurrent Neural Network model after white silk can be dispatch account identification model, and the attention mechanism module after obtained training can To be attention Mechanism Model.
Optionally, Fig. 2 is a kind of identification model generation method flow chart provided by the embodiments of the present application.
As shown in Fig. 2, this method comprises:
S201, at least one dispatch account training sample is determined；
Optionally, dispatch account can be public platform, alternatively, other are used for the account published an article.Above is only this Apply for the preferred embodiment for the dispatch account that embodiment provides, the tool of dispatch account can be arbitrarily arranged in inventor according to their own needs Hold in vivo, it is not limited here.
Wherein, any one dispatch account can all be realized as a dispatch account training sample to neural network Training is to generate identification model.Furthermore, it is desirable to it is to be noted that: any one dispatch account also can be by be identified as one Dispatch account is identified to obtain recognition result with identified model, in order to determine the dispatch account to be identified based on recognition result Number whether be rubbish dispatch account.
S202, acquisition and the relevant first information of the dispatch dispatch behavior of account training sample and with dispatch account training sample This relevant second information of account attribute；
Optionally, the first information relevant to the dispatch dispatch behavior of account includes the article sequence that dispatch account is delivered. Wherein, the article sequence delivered of dispatch account includes: each piece article successively delivered by dispatch account in default historical time section The article sequence of composition.For example, if dispatch account creation time so far, dispatch account altogether has delivered 3 articles (respectively Article 1, article 2 and article 3, wherein the time of delivering of article 2 is later than delivering the time for article 3, and article 1 delivers evening time The time is delivered in article 2)；If default historical time section be the creation time of dispatch account so far, account of sending the documents is delivered Article sequence be made of the article 3, article 2 and article 1 being arranged successively.
Optionally, the number of the account attribute for account of sending the documents is at least one, specifically, the account attribute of dispatch account can Think whether the registration time limit of dispatch account, dispatch account are certified, the history for account of sending the documents is sent the documents number etc..
It is above only a kind of preferred embodiment of the account attribute of account of sending the documents provided by the embodiments of the present application, inventor can Particular content included by the account attribute of any setting dispatch account according to their own needs, it is not limited here.
Optionally, the second information relevant to the dispatch account attribute of account training sample can be for based on dispatch account instruction Practice the vector of the account attribute building of sample.
In the embodiment of the present application, it is preferred that obtain second information relevant to the dispatch account attribute of account to be identified Mode, be referred to a kind of method stream of acquisition and relevant second information of the account attribute of account of sending the documents provided by Fig. 3 Cheng Tu.
As shown in figure 3, this method comprises:
S301, each account attribute at least one account attribute and at least one account attribute of dispatch account is determined Attribute value；
Optionally, each account at least one account attribute and at least one account attribute of dispatch account is determined The attribute value of attribute.For example, account attribute is prescribing a time limit in registration year for dispatch account, the attribute value of account attribute is the registration time limit； When account attribute is whether dispatch account is certified, the attribute value of account attribute is "Yes" or "No"；Account attribute is dispatch When the history dispatch number of account, the attribute value of account attribute is history dispatch number.
It is above only the preferred embodiment of the attribute value of account attribute and account attribute provided by the embodiments of the present application, This is without limitation.
S302, property parameters corresponding with account attribute are determined according to the attribute value of account attribute；
Optionally, respectively each account attribute setting determines that the account attribute is corresponding based on the attribute value of the account attribute Property parameters rule；For example, account attribute is prescribing a time limit in registration year for dispatch account, the attribute value of account attribute is registration year Limit, the corresponding property parameters of account attribute are the attribute value (that is, registration time limit) of account attribute；Account attribute is dispatch account Whether when being certified, the attribute value of account attribute is "Yes" or "No", if the attribute value of account attribute is "Yes", account category Property corresponding property parameters be 1, if the attribute value of account attribute is "No", the corresponding property parameters of account attribute are 0；Account When attribute is the history dispatch number of dispatch account, the attribute value of account attribute is history dispatch number, and account attribute is corresponding Property parameters are the attribute value (that is, history dispatch number) of account attribute.
Be only above the corresponding property parameters of account attribute provided by the embodiments of the present application it is preferable to determine mode, inventions The method of determination of the corresponding property parameters of account attribute can be arbitrarily arranged in people according to their own needs, it is not limited here.
S303, the corresponding property parameters of each account attribute at least one account attribute based on determined by, generate with Second information, the corresponding property parameters of each account attribute are an element in the second information at least one account attribute.
Optionally, based on determining and the corresponding category of each account attribute at least one account attribute of dispatch account Property parameter, generate with the corresponding vector of account of sending the documents, this vector may be considered relevant to the account attribute of the dispatch account Second information；Wherein, the parameter of each account attribute at least one account attribute is an element in the vector.
In the embodiment of the present application, it is preferred that determine the sequence of each account attribute at least one account attribute, and according to According to the sequence of each account attribute at least one account attribute, category corresponding to account attribute each at least one account attribute Property parameter is ranked up, and obtains property parameters sequence；Regard each of property parameters sequence property parameters as raw respectively At vector in an element, obtained ranking results are the category after each element in the vector of generation successively sorts Property argument sequence.That is, sequence of the account attribute at least one account attribute property parameters corresponding with the account attribute exist Sequence in each element of vector is identical.
For example, if at least one account attribute is respectively the registration time limit (account attribute of account attribute 1[dispatch account 1) corresponding property parameters are], account attribute 2[account of sending the documents whether be certified that (the corresponding property parameters of account attribute are 0)] and the history of account attribute 3[dispatch account sends the documents number (the corresponding property parameters of account attribute are 5)]；If really The sequence of each account attribute is followed successively by account attribute 2, account attribute 3 and account attribute 1 at least one fixed account attribute, then The vector of generation is (0,5,1).
S203, the term vector sequence that dispatch account training sample is determined based on the dispatch account training sample first information, will Term vector sequence inputting handles term vector sequence based on Recognition with Recurrent Neural Network model, obtains to Recognition with Recurrent Neural Network model To the first identification information；
Optionally, it can be obtained according to the sequence of the article in the article sequence in the first information of dispatch account, sequence The each word successively to sort in every article generates the word sequence of this article sequence；Each word in word sequence is converted into word Vector obtains term vector sequence corresponding with word sequence.Wherein, obtained term vector sequence be the dispatch account word to Measure sequence.
Optionally, if article sequence be made of the article 1, article 2 and article 3 being arranged successively (article 1 deliver the time morning In delivering the time for article 2, the time of delivering of article 2 earlier than article 3 delivers the time), then can sequentially obtain in article 1 according to It successively sorts in the term vector and article 3 of each word successively to sort in the term vector of each word of minor sort, article 2 The term vector of each word, and the term vector sequence based on each term vector composition article sequence successively got, term vector sequence Each term vector in column puts in order as the acquisition sequence of each term vector.
For example, if article sequence is made of the article 1, article 2 and article 3 being arranged successively, sequence obtain in article 1 according to Term vector (term vector wherein, successively got from article 1 is term vector 1, term vector 2), the text of each word of minor sort The each word successively to sort in chapter 2 term vector (term vector wherein, successively got from article 2 be term vector 3, word to Amount 4, term vector 5), the term vector (term vector wherein, successively got from article 3 of each word that successively sorts in article 3 For term vector 6), generate article sequence term vector sequence (term vector sequence generated from each word for successively getting to Amount is constituted, then term vector sequence generated is the term vector 1 being arranged successively, term vector 2, term vector 3, term vector 4, term vector 5 and term vector 6).
It can be by the term vector sequence inputting for account training sample of sending the documents to Recognition with Recurrent Neural Network model, based on circulation nerve Network model handles term vector sequence, obtains the first identification information, which may include to the dispatch The expression of the article sequence of account training sample.
S204, by the first identification information and the second information input to attention mechanism module, be based on attention mechanism module First identification information and the second information are handled, the second identification information is obtained；
S205, classification prediction is carried out to the second identification information, obtains classification prediction result；
Optionally, the second identification information is input to softmax function, classification prediction is carried out to the second identification information, is obtained To classification prediction result.
S206, it is reversely adjusted in Recognition with Recurrent Neural Network model based on classification prediction result for term vector sequence For the parameter that is handled the first identification information and the second information in the parameter and attention mechanism module of reason, with generate with The corresponding dispatch account identification model of Recognition with Recurrent Neural Network model and attention mechanism mould corresponding with attention mechanism module Type.
Optionally, based on gradient descent algorithm using step S205 to dispatch account training sample classification prediction result and The classification results for the dispatch account sample that user demarcates in advance, it is reversed to adjust in Recognition with Recurrent Neural Network model for term vector Ginseng in the parameter and attention mechanism module that sequence is handled for being handled the first identification information and the second information Number；After execution multipass step S201-S205 provided by the embodiments of the present application, by Recognition with Recurrent Neural Network model and attention mechanism The neural network of module composition reaches convergence, can obtain dispatch account identification model corresponding with Recognition with Recurrent Neural Network model, With attention Mechanism Model corresponding with attention mechanism module.
Optionally, the classification results for the dispatch account training sample that user demarcates in advance can serve to indicate that the dispatch account Training sample is rubbish dispatch account or non-junk dispatch account.
For the ease of to it is provided by the embodiments of the present application it is a kind of send the documents account recognition methods understanding, now to above-mentioned the application A kind of method of the term vector sequence for determining dispatch account that embodiment provides is described further.
Optionally, Fig. 4 is a kind of method flow of the term vector sequence of determining dispatch account provided by the embodiments of the present application Figure.
As shown in figure 4, this method comprises:
S401, for dispatch account the first information in article sequence, respectively obtain this article sequence in every text The term vector sequence of the term vector sequence of chapter, article is made of the term vector of each word successively to sort in article；
Optionally, the first information for account of sending the documents is the first information relevant to the dispatch dispatch behavior of account；For For every article in article sequence, this article is segmented, obtains the word segmentation result of this article；Sequence obtains should The term vector of each word successively to sort in the word segmentation result of piece article, the sequence that each term vector successively got is formed Term vector sequence as this article.
For example, the word segmentation result to this article is the word 1 that successively sorts if the content of article is " could you tell me and have a meal not " (may I ask), word 2 (you), word 3 (having a meal) and word 4 (not having)；Successively obtain word 1 (may I ask) term vector (referred to herein as term vector 1), Term vector (referred to herein as term vector 2), the word 3 of word 2 (you) (are had a meal) term vector (referred to herein as term vector 3) and word 4 (not having) Term vector (referred to herein as term vector 4) will be made of the term vector 1, term vector 2, term vector 3 and the term vector 4 that successively get Term vector sequence as this article term vector sequence (that is, the term vector sequence of this article from the word that successively sorts to Amount 1, term vector 2, term vector 3 and term vector 4 are constituted).
Further, before segmenting to article, data cleansing can also be carried out to article, to remove in article The character of preset requirement is not met.Optionally, the character for not meeting preset requirement includes: punctuation mark, number etc..Above only It is the preferred embodiment provided by the embodiments of the present application that data cleansing is carried out to article, inventor can arbitrarily set according to their own needs The mode that data cleansing is carried out to article is set, it is not limited here.
S402, according to the sequence of the article in article sequence, the term vector sequence of each article got is ranked up, Generate the term vector sequence of article sequence.
Optionally, the row after the term vector sequence for determining every article in article sequence, according to article in article sequence Sequence is ranked up the term vector sequence of each article in the article sequence got, to generate the term vector of article sequence Sequence.
For example, article sequence includes the article 1, article 2 and article 3 successively to sort, the term vector sequence of article 1 is got 3, the term vector sequence 2 of the term vector sequence 1 of article 2 and article 3, according to the sequence of article in article sequence to the text got The term vector sequence of article in Zhang Xulie is ranked up, obtain article sequence term vector sequence (word of this article sequence to Amount sequence is made of the term vector sequence 3, term vector sequence 1 and term vector sequence 2 successively to sort；Specifically, if term vector sequence It include the term vector 1 and term vector 3 successively to sort in column 1；It include term vector 2, the term vector successively to sort in term vector sequence 2 4 and term vector 5；When term vector sequence 3 includes the term vector 7 and term vector 6 successively to sort, then the term vector 7, the word that successively sort The sequence that vector 6, term vector 1, term vector 3, term vector 2, term vector 4 and term vector 5 form is the term vector sequence of article sequence Column).
For the ease of to it is provided by the embodiments of the present application it is a kind of send the documents account recognition methods understanding, it is now above-mentioned to the application A kind of term vector sequence inputting by article sequence that embodiment provides is based on Recognition with Recurrent Neural Network to Recognition with Recurrent Neural Network model Model handles term vector sequence, and the method for obtaining the first identification information describes in detail.
Fig. 5 is a kind of term vector sequence inputting by article sequence provided by the embodiments of the present application to Recognition with Recurrent Neural Network mould Type handles term vector sequence based on Recognition with Recurrent Neural Network model, obtains the method flow diagram of the first identification information.
As shown in figure 5, this method comprises:
S501, using the term vector sequence of article sequence as the first hidden of input information input to Recognition with Recurrent Neural Network model Hiding layer obtains the expression of the word sequence of the article sequence as the first output result；
Optionally, Recognition with Recurrent Neural Network model may include three hidden layers, respectively the first hidden layer, the second hidden layer With third hidden layer.
In the embodiment of the present application, it is preferred that using the term vector sequence of article sequence as input information input to circulation First hidden layer of neural network model obtains the first output as a result, the first output result can be the word sequence of article sequence It indicates.
Optionally, the word sequence of article sequence are as follows: according to the sequence of article in article sequence, sequentially obtain in every article The each word successively to sort, and using the sequence of each word successively got composition as the word sequence of article sequence, wherein word Acquisition sequence and word putting in order in word sequence it is identical.
For example, if the term vector sequence of article sequence is by the term vector 1 of the word 1 successively to sort, the term vector 3 of word 3 and word 2 Term vector 2 constitute；After using the term vector sequence of this article sequence as input information input to Recognition with Recurrent Neural Network model, obtain The the first output result arrived is the expression of the word sequence of article sequence, wherein the expression of the word sequence of article sequence herein can Think the expression of the expression for the word 1 being arranged successively, the expression of word 3 and word 2.
In the embodiment of the present application, it is preferred that using the term vector sequence of article sequence as input information input to circulation First hidden layer of neural network model obtains the expression of the word sequence of the article sequence as the first output result, comprising: will The term vector sequence of article sequence, to the first hidden layer of Recognition with Recurrent Neural Network model, obtains article sequence as input information input The positive reverse presentation indicated with the word sequence of article sequence of the word sequence of column, and according to the positive expression of word sequence and word order The reverse presentations of column calculates the expression of the word sequence of article sequence, and the expression of the word sequence of the article sequence being calculated can be with First as the first hidden layer exports result.
Optionally, if the term vector sequence of article sequence is (wij1,...,wijT), wherein i indicates article, and j indicates sentence Son, T indicate word, wijTIndicate the T word in article sequence in j-th of sentence of i-th article；It is by term vector sequence (wij1,...,wijT) be input to first hidden layer of Recognition with Recurrent Neural Network model after；The word-based sequence vector of first hidden layer (wij1,...,wijT) the positive of word sequence that obtains article sequence indicate[that is,Wherein,Indicated for the positive of t-th of word of j-th of sentence of i-th article in article sequence], and And first hidden layer can also be according to the inverted order (w of term vector sequenceijT,...,wij1) obtain article sequence word sequence it is reversed It indicates[that is,Wherein,For the jth of i-th article in article sequence The reverse presentation of t-th of word of a sentence]；Further, the first hidden layer can be indicated based on the forward direction of word sequenceWith the reverse presentation of word sequenceThe expression of word sequence is calculatedWherein, hijtFor the expression of t-th of word in j-th of sentence in article sequence in i-th article, the calculating The expression of obtained word sequenceFirst output result of as the first hidden layer.
S502, the second hidden layer of the first output result input Recognition with Recurrent Neural Network model is obtained tying as the second output The expression of the sentence sequence of the article sequence of fruit；
Optionally, using the first output result as the input information of the second hidden layer of Recognition with Recurrent Neural Network model, input To the second hidden layer of the Recognition with Recurrent Neural Network model, the second output is obtained as a result, the second output result is article sequence The expression of sentence sequence.
In the embodiment of the present application, it is preferred that the sentence sequence of article sequence is: the row according to article in article sequence Sequence, sequence determine each sentence for successively sorting in every article, and using the molecular sequence of each sentence successively determined as The sentence sequence of article sequence.
In the embodiment of the present application, if the first output result of the first hidden layer is the expression of word sequence When, the first output result is exported to the second hidden layer, the second hidden layer handles the first output result, and it is defeated to obtain second Result out.Specifically, the mode that the second hidden layer handles the first output result is successively based on formula one, uijt= tanh(Wwhijt+bw), formula two,With formula three,Obtain the second output result (the alternatively referred to as expression of the sentence sequence of article sequence) hij；Wherein, WijtFor parameter matrix, bwFor bias vector, tanh () It is [- 1,1], h for nonlinear function valueijFor the expression of j-th of sentence in i-th article in article sequence.
S503, the third hidden layer of the second output result input Recognition with Recurrent Neural Network model is obtained exporting as third and is tied The first identification information of fruit, the first identification information include the expression of article sequence.
In the embodiment of the present application, if the second output result of the second hidden layer is the expression h of sentence sequenceijWhen, by Two output results obtain third output result as input information input to third hidden layer.
Optionally, the mode that third hidden layer handles the second output result are as follows: be based onFormula obtains The expression h of article sequencei, wherein hiFor the expression of i-th article in article sequence.Correspondingly, the article sequence that can also will be obtained (h is regarded in the expression of column as1,...,hi,...hT)。
For the ease of to it is provided by the embodiments of the present application it is a kind of send the documents account recognition methods understanding, it is now above-mentioned to the application Embodiment provides a kind of by the first information and the second information input to attention mechanism module, is based on attention mechanism module pair First identification information and the second information are handled, and the method for obtaining the second identification information describes in detail.
Optionally, the first identification information of the article sequence for account of sending the documents can be the table of the article sequence of dispatch account Show；Second information of account of sending the documents is related to the dispatch account attribute of account, and the form of expression can be vector.
It is provided by the embodiments of the present application a kind of by the first information and the second information input to attention mechanism module as a result, The first identification information and the second information are handled based on attention mechanism module, the method for obtaining the second identification information can Think by the second information input of the expression of the article sequence for account of sending the documents and account of sending the documents to attention mechanism module, based on note Meaning power mechanism module handles the expression of article sequence and the second information, the method for obtaining the second identification information.
Optionally, in the embodiment of the present application, based on attention mechanism module to the expression of article sequence and the second information into Row processing, the method for obtaining the second identification information can be found in Fig. 6.
As shown in fig. 6, this method comprises:
S601, the expression of article sequence of dispatch account is handled according to the second information of dispatch account, obtain with The scoring of the expression of every article in article sequence；
S602, using every article in the expression and article sequence of article sequence expression scoring, calculate as the The final expression of the article sequence of two identification informations, and by the final table for indicating to be determined as new article sequence of article sequence Show.
Optionally, if the second information relevant to the dispatch account attribute of account is vector a, by this vector a and article sequence Expression (the h of column1,...,hi,...hT) it is input to attention mechanism module together, the final expression of article sequence can be obtained, The final of this article sequence indicates to may be considered the expression (h based on attention mechanism module to article sequence1,...,hi, ...hT) and vector a handled, obtained second identification information.
Specifically, attention mechanism module is in the expression (h for receiving vector a and article sequence1,...,hi,...hT) after, By vector a to the expression (h of article sequence1,...,hi,...hT) handled, to obtain finally indicating for article sequence Mode includes: the expression (h according to vector a to article sequence1,...,hi,...hT) handled, obtain in article sequence The scoring of the expression of every article；Utilize the scoring of the expression of every article in the expression and article sequence of article sequence, meter Calculate the final expression of article sequence.
Wherein, according to vector a to the expression (h of article sequence1,...,hi,...hT) handled, it obtains and article sequence In every article expression scoring, comprising: according to formula mi=tanh (Waa+Whhi) to the table of vector a and article sequence Show (h1,...,hi,...hT) handled, obtain the scoring m of the expression of every articlei, wherein Wa,For parameter matrix, WhFor Parameter matrix, miFor the scoring of the expression of i-th article in article sequence.Correspondingly, the Grading sequence of the expression of article is (m1,...,mi,...,mT).It optionally, can be based on such mode according to vector, respectively to every in the expression of article sequence The expression of article is handled, and the scoring of the expression of article is obtained.
Correspondingly, the scoring of the expression using every article in the expression and article sequence of article sequence, calculates article The final expression of sequence, comprising: pass through formula h=∑ mihiRealize the Grading sequence (m of the expression using article1,..., mi,...,mT) to the expression (h of article sequence1,...,hi,...hT) be weighted summation obtain the final expression h of article sequence. Optionally, multiplying for the expression of article and the scoring of article can be calculated for every article in article sequence based on such mode Product result；By the summation of each result of product calculated, it is determined as the final expression of article sequence.
It is discussed in detail by above-mentioned, it can be in order to which those skilled in the art be to a kind of dispatch provided by the embodiments of the present application Identification model construction method in account recognition methods is understood in detail.Those skilled in the art will be appreciated that the application For embodiment when constructing identification model, applied dispatch account is properly termed as dispatch account training sample；The embodiment of the present application When being identified based on the identification model built to dispatch account, what the identification model for having been built was identified Dispatch account can be described as dispatch account to be identified.Although dispatch account can be claimed according to the difference of the application approach of dispatch account For dispatch account training sample or dispatch account to be identified, still, what those skilled in the art should learn is that the application is implemented Example in, to dispatch account training sample processing mode and to it is to be identified dispatch account processing mode can be found in above-mentioned implementation To the description of the processing mode of dispatch account in example, then this is not repeated them here.
In order to be more clearly illustrated to a kind of dispatch account recognition methods of the offer of the embodiment of the present application, now from The angle identified to dispatch account is introduced.
Optionally, Fig. 7 is a kind of dispatch account recognition methods flow chart provided by the embodiments of the present application.
As shown in fig. 7, this method comprises:
S701, dispatch account to be identified is determined；
S702, the first information relevant to the dispatch dispatch behavior of account to be identified is obtained；
Optionally, it is referred in relation to obtaining the mode of the first information relevant to the dispatch dispatch behavior of account to be identified The description of embodiment is stated, then this is not repeated them here.
S703, dispatch account to be identified is identified using the first information, obtains the identification knot of dispatch account to be identified Fruit.
Optionally, the identification model being trained based on dispatch account training sample to neural network, the knowledge are determined It include dispatch account identification model in other model；It determines the first information relevant to the dispatch dispatch behavior of account to be identified, leads to It crosses dispatch account identification model to identify the first information, to obtain the recognition result of dispatch account to be identified.In the application In embodiment, it is preferred that can be dispatch account identification model and carried out to the term vector sequence of the article sequence in the first information Identification, to obtain the recognition result of dispatch account to be identified.
In the embodiment of the present application, it is preferred that determine the first information relevant to the dispatch dispatch behavior of account to be identified, Using the term vector sequence of the article sequence in the first information as input information input to dispatch account identification model, to be sent out (output result herein may be considered the first identification letter that above-described embodiment is mentioned to the output result of literary account identification model Breath).It optionally, can be using first identification information as input information input to softmax function, by softmax function pair First identification information carries out classification prediction, obtains the recognition result of the dispatch account to be identified.
Optionally, the embodiment of the present application also provides another dispatch account recognition methods, refers to Fig. 8.
As shown in figure 8, this method comprises:
S801, dispatch account to be identified is determined；
S802, the first information relevant to the dispatch dispatch behavior of account to be identified is obtained；
In the embodiment of the present application, the executive mode of step S801-S802 and step S701- provided by the above embodiment The executive mode of S702 is identical, and the description of above-described embodiment, then this are referred in relation to the specific executive mode of step S801-S802 It does not repeat them here.
S803, second information relevant to the dispatch account attribute of account to be identified is obtained；
Optionally, it is referred in relation to obtaining the mode of second information relevant to the dispatch account attribute of account to be identified The description of embodiment is stated, then this is not repeated them here.
S804, dispatch account to be identified is identified using the first information and the second information, obtains dispatch account to be identified Number recognition result.
Optionally, the identification model being trained based on dispatch account training sample to neural network, the knowledge are determined Other model not only includes that dispatch account identification model further comprises attention mechanism model, and the first information is inputted as input letter Breath is input to dispatch account identification model, and obtaining output result, (output result herein may be considered above-described embodiment and mention The first identification information), and using the output result and the second information as the input information input of attention Mechanism Model to infusing It anticipates power Mechanism Model, with the output result that obtains the attention Mechanism Model, (output result herein may be considered above-mentioned reality Apply the second identification information that example is mentioned).In the embodiment of the present application, it is preferred that can believe the second identification information as input Breath is input to softmax function, carries out classification prediction to the second identification information by softmax function, obtains the dispatch to be identified The recognition result of account.
In the embodiment of the present application, it is preferred that the first information relevant to the dispatch dispatch behavior of account to be identified includes The article sequence that dispatch account to be identified is delivered.
In the embodiment of the present application, it is preferred that dispatch account to be identified is known using the first information and the second information Not, the recognition result of dispatch account to be identified is obtained, comprising: the sequence of the article in article sequence in the foundation first information, Sequence obtains each word successively to sort in every article, generates the word sequence of article sequence；Each word in word sequence is turned It changes term vector into, obtains term vector sequence corresponding with word sequence；By term vector sequence inputting to dispatch account identification model, base Term vector sequence is handled in dispatch account identification model, obtains the first identification information；Known based on the second information to first Other information is handled, and the second identification information is obtained.
Optionally, by term vector sequence inputting to dispatch account identification model, based on dispatch account identification model to word to Amount sequence is handled, and the first identification information is obtained, comprising: by the term vector sequence inputting of article sequence to dispatch account identification Model handles term vector sequence based on dispatch account identification model, exports the article sequence as the first identification information Expression.
Optionally, second information relevant to the dispatch account attribute of account to be identified is obtained, comprising: determine hair to be identified The attribute value of each account attribute at least one account attribute and at least one account attribute of literary account；According to account category Property attribute value determine corresponding with account attribute property parameters；Based on each account at least one identified account attribute The corresponding property parameters of attribute generate second information relevant to the dispatch account attribute of account to be identified, at least one account The corresponding property parameters of each account attribute are an element in the second information in attribute.
Optionally, the first identification information is handled based on the second information, obtains the second identification information, comprising: by One identification information and the second information input are to attention Mechanism Model, based on attention Mechanism Model to the first identification information and the Two information are handled, and the second identification information is obtained.
In the embodiment of the present application, it is preferred that the first identification information is handled based on the second information, obtains the second knowledge Other information, comprising: the expression of article sequence is handled according to the second information, obtains in article sequence commenting for every article Point；The expression of article sequence is handled based on the scoring of identified every article, is obtained as the second identification information The final expression of article sequence.
The embodiment of the present application provides a kind of account recognition methods and server, passes through determination dispatch account to be identified；It obtains The first information relevant to the dispatch dispatch behavior of account to be identified；Dispatch account to be identified is known using the first information Not, the mode for obtaining the recognition result of dispatch account to be identified, so that based on related to the dispatch dispatch behavior of account to be identified The first information realize the identification to dispatch account to be identified, reduce the prior art according to individually to dispatch account to be identified In the recognition result of every article when realizing the identification to dispatch account to be identified, the caused identification to dispatch account to be identified As a result inaccuracy, cannot effectively identify rubbish send the documents account the problem of.
On the basis of a kind of account recognition methods and server that the above embodiments of the present application provide, the embodiment of the present application A kind of account display methods can also be provided.Account display methods provided by the embodiment of the present application can be applied to terminal, It can be applied to server.
Server applied by account display methods provided by the embodiments of the present application can be provided by the above embodiment one Server applied by kind dispatch account recognition methods, being also possible to another can be with a kind of dispatch provided by the above embodiment The server that server applied by account recognition methods is communicated.When account display methods institute provided by the embodiments of the present application When the server of application is different from server applied by a kind of dispatch account recognition methods provided by the above embodiment, the application Server applied by a kind of account display methods that embodiment provides can be according to a kind of dispatch account provided by the above embodiment Server applied by recognition methods realizes the display to dispatch account to the recognition result of dispatch account.
In the embodiment of the present application, mainly by taking account display methods is applied to terminal as an example, the embodiment of the present application is provided A kind of account display methods be illustrated.A kind of account display methods provided by the embodiments of the present application is specifically as follows a kind of hair Literary account display methods.
In the embodiment of the present application, terminal can be mobile phone, tablet computer etc..Such as, referring to Fig. 9, it illustrates the application A kind of composed structure schematic diagram for the terminal that the dispatch account display methods of embodiment is applicable in.
In Fig. 9, which may include: processor 901, memory 902, communication interface 903, input unit 904 With display 905 and communication bus 906.
For storing one or more than one program in memory 902, program may include program code, program code Including computer operation instruction, it is at least stored with the journey for realizing following functions in the embodiment of the present application, in the memory Sequence:
Receive dispatch account display request；
Obtain each dispatch account being located within the scope of the dispatch account of dispatch account display request instruction；
Determine that dispatch account number classification belonging to each dispatch account, dispatch account number classification are rubbish dispatch account number classification or non- Rubbish dispatch account number classification；
Based on dispatch account display rule relevant to dispatch account number classification, each dispatch account is shown.
Processing module 901, communication interface 903, input unit 904, display 905, passes through communication always at memory 902 Line 906 completes mutual communication.
In the embodiment of the present application, the processor 901 can be central processing unit (Central Processing Unit, CPU), application-specific integrated circuit (application-specific integrated circuit, ASIC), number Signal processor (DSP), specific integrated circuit (ASIC), ready-made programmable gate array (FPGA) or other programmable logic devices Part etc..
The processor can call and execute the program stored in memory 902.
The communication interface 903 can be the interface of communication module, such as the interface of gsm module.
The application can also include input unit 905, which may include the touching incuded on touch display panel Touch touch sensing unit, the keyboard etc. of event.
The display 904 includes display panel, such as touch display panel.In the case where a kind of possible, it can use Liquid crystal display (Liquid Crystal Display, LCD), Organic Light Emitting Diode (Organic Light-Emitting Diode, OLED) etc. forms configure display panel.
Certainly, terminal structure shown in Fig. 9 does not constitute the restriction to terminal in the embodiment of the present application, in practical applications Terminal may include than more or fewer components shown in Fig. 9, or the certain components of combination.
Optionally, the refinement function of program and extension function can refer to and be described below.
Below with reference to the above general character of the application, terminal is applied to the dispatch account display methods of the embodiment of the present application Situation describes in detail.
Referring to Figure 10, it illustrates a kind of flow diagrams for account display methods of sending the documents provided by the embodiments of the present application.
As shown in Figure 10, this method comprises:
S1001, dispatch account display request is received；
Optionally, terminal can receive the dispatch account display request of user's transmission.
S1002, each dispatch account for obtaining dispatch account display request instruction；
In the embodiment of the present application, it is preferred that when dispatch account display request is dispatch account searching request, it is believed that Account of sending the documents display request carries dispatch account range, and the search condition that dispatch account searching request carries may be considered dispatch The dispatch account range that account display request carries；The each dispatch account for obtaining dispatch account display request instruction, can recognize For be obtain be located at dispatch account display request carry dispatch account within the scope of each dispatch account.Optionally, Ke Yicong Inquiry meets the dispatch account of search condition in already present all dispatch accounts, and using the dispatch account found as hair Each dispatch account of literary account display request instruction, that is, within the scope of the dispatch account that dispatch account display request carries Each dispatch account.
In the embodiment of the present application, preferred search condition can be dispatch account dispatch account title in whole/ Partial content etc..It is above only the preferred embodiment of search condition provided by the embodiments of the present application, the tool in relation to search condition Hold in vivo it is not limited here.
S1003, determine that dispatch account number classification belonging to each dispatch account, dispatch account number classification are rubbish dispatch account class Other or non-junk dispatch account number classification；
Optionally, after each dispatch account for getting dispatch account display request instruction, acquisition can be determined respectively Dispatch account number classification belonging to each dispatch account arrived, if dispatch account is rubbish dispatch account, which belongs to Rubbish dispatch account number classification；If account of sending the documents is non-junk dispatch account, which belongs to non-junk dispatch account class Not.
S1004, rule is shown based on dispatch account relevant to dispatch account number classification, shows each dispatch account.
In the embodiment of the present application, it is preferred that be provided with dispatch account display rule relevant to dispatch account number classification.It can Choosing, dispatch account display rule is used to indicate the dispatch account that shielding belongs to rubbish dispatch account number classification, alternatively, dispatch account The display priority that display rule is used to indicate the dispatch account for belonging to non-junk dispatch account number classification, which is higher than, belongs to rubbish dispatch The display priority of the dispatch account of account number classification.
In the embodiment of the present application, it is preferred that the dispatch account that shielding belongs to rubbish dispatch account number classification, which can be, not to be shown Show the dispatch account for belonging to rubbish dispatch account number classification, be also possible to the dispatch account that display belongs to rubbish dispatch account number classification, But the dispatch account for belonging to rubbish dispatch account number classification is shown with particular form, particular form, which can be, to be different from belonging to non-rubbish The display mode of the dispatch account of rubbish dispatch account number classification, for example, belonging to rubbish dispatch account class with what particular form was shown Other dispatch account can not be checked etc..
In the embodiment of the present application, it is preferred that belong to the display priority of the dispatch account of non-junk dispatch account number classification Higher than the display priority for the dispatch account for belonging to rubbish dispatch account number classification, it can be and belong to non-junk dispatch in the same page Display order of the display order of the dispatch account of account number classification earlier than the dispatch account for belonging to rubbish dispatch account number classification.
For example, if getting 3 dispatch accounts, account 1 of respectively sending the documents, dispatch account 2 and dispatch account 3；Wherein, it sends out Literary account 1 belongs to rubbish dispatch account number classification, and dispatch account 2 belongs to non-junk dispatch account number classification, and dispatch account 3 belongs to rubbish Dispatch account number classification；Dispatch account display rule can indicate to arrange each dispatch account got in the same page Column display belongs to display order of each dispatch account of non-junk dispatch account number classification in the page and sends the documents earlier than rubbish is belonged to Display order of each dispatch account of account number classification in the page.For example, showing the dispatch account got in the same page Number 1, dispatch account 2 and dispatch account 3, wherein display order of the dispatch account 2 in the page is earlier than dispatch account 1 and dispatch Display order of the account 3 in the page.
For the ease of to it is provided by the embodiments of the present application it is a kind of send the documents account display methods understanding, now in conjunction with a kind of dispatch A kind of dispatch account display methods provided by the embodiments of the present application is described in detail in the application scenarios of account display methods.
A kind of dispatch account display methods application scenarios provided by the embodiments of the present application are that dispatch account searches for scene.Such as figure 11 (a) -11 (e) show the application page signal for account display methods of sending the documents in a kind of terminal provided by the embodiments of the present application Figure.User is by being operated into the page as shown in Figure 11 (a) to terminal, in the page as shown in Figure 11 (a), clicks friend Friend's circle or public platform, are to scan for circle of friends or scan for public platform with determination.
In the embodiment of the present application, it is preferred that circle of friends can regard dispatch account as, and public platform can also regard hair as Literary account.It is above only the preferred embodiment of dispatch account provided by the embodiments of the present application, specific inventor can be according to oneself Demand the particular content of dispatch account is arbitrarily set, then this is without limitation.
Optionally, as shown in Figure 11 (a), when the user clicks when public platform, the page shown in Figure 11 (b) can be jumped to； As shown in Figure 11 (c), user can input search condition in the search box of the page of such as Figure 11 (b), in Figure 11 (c) " Xiamen "；And the search button as shown in Figure 11 (c) is clicked, can meet search condition from all already present public platforms is The public platform in " Xiamen " scans for.In the embodiment of the present application, it is preferred that is searched out meets the public affairs in search condition " Xiamen " Many numbers one share 5, respectively public platform 1, public platform 2, public platform 3, public platform 4 and public platform 5；Also, determine the public Number 1 belongs to non-junk public platform classification, and public platform 2 belongs to rubbish public platform classification, and public platform 3 belongs to non-junk public platform class Not, public platform 4 belongs to non-junk public platform classification, and public platform 5 belongs to rubbish public platform classification.
In the embodiment of the present application, it is preferred that the public platform for meeting search condition scanned for, and determines and searches out The process of classification belonging to public platform can be executed in terminal, can also be executed in server；If, can when server executes To be that implementing result is returned to terminal by server, in order to which terminal is shown；Alternatively, server is directly according to implementing result It determines the page to be jumped to, such as Figure 11 (d) or Figure 11 (e), and then the page jumped to is treated by server controlling terminal It is shown.Specific executive mode, inventor can be arbitrarily arranged according to their own needs, it is not limited here.
Further, based on dispatch account display rule relevant to dispatch account number classification, each hair searched out is shown Literary account；If dispatch account relevant to dispatch account number classification shows that rule is used to indicate shielding and belongs to rubbish dispatch account class When other dispatch account, it can realize from the page jump as shown in Figure 11 (c) to the page as shown in Figure 11 (d) to search The display of public platform out；The dispatch account that shielding herein belongs to rubbish dispatch account number classification, which can be, does not show that belonging to rubbish sends out The dispatch account of literary account number classification.
If dispatch account relevant to dispatch account number classification, which shows that rule is used to indicate, belongs to non-junk dispatch account class The display priority of other dispatch account is higher than the display priority for belonging to the dispatch account of rubbish dispatch account number classification, Ke Yicong The page jump as shown in Figure 11 (c) realizes the display to the public platform searched out to the page as shown in Figure 11 (e)；Herein Belong to non-junk dispatch account number classification dispatch account display priority be higher than belong to rubbish send the documents account number classification dispatch The display priority of account can be in the same page, belong to the dispatch account of non-junk dispatch account number classification in the page Display order is earlier than display order of the dispatch account in the page for belonging to rubbish dispatch account number classification；It further, can be with The dispatch account for belonging to rubbish dispatch account number classification shown in the page is marked with special color；It further, can be with The dispatch account for belonging to rubbish dispatch account number classification shown in prohibiting page is activated by a user to jump to and the dispatch account Number corresponding page.
It is above only the embodiment of the present application in order to illustrate to a kind of dispatch account identification side provided by the embodiments of the present application Method is described in detail, and scene applied by a kind of dispatch account recognition methods provided, related the embodiment of the present application provide Dispatch account identification/display methods concrete application scene inventor can arbitrarily be arranged according to their own needs, do not do herein It limits.
The embodiment of the present application provides a kind of account display methods and server, based on to dispatch account belonging to dispatch account The determination of classification realizes the display control to dispatch account, and then convenient for optimization network environment, brings better net for user Network experience.
Dispatch account identification device provided in an embodiment of the present invention is introduced below, dispatch account described below is known Other device is regarded as, the server dispatch account recognition methods that embodiment provides to realize the present invention, the program of required setting Module.Dispatch account identification device content described below, can be mutually right with above-described dispatch account recognition methods content It should refer to.
Figure 12 is a kind of structural block diagram of account identification device of sending the documents provided by the embodiments of the present application, which can be applied to Server, referring to Fig.1 2, which may include:
Dispatch account determination unit 121 to be identified, for determining dispatch account to be identified；
First information acquiring unit 122, for obtaining the first information relevant to the dispatch dispatch behavior of account to be identified；
Recognition result determination unit 123 is obtained for being identified using the first information to dispatch account to be identified wait know Not Fa Wen account recognition result.
Optionally, recognition result determination unit, comprising:
Second information acquisition unit, for obtaining second information relevant to the dispatch account attribute of account to be identified；
Recognition result determines subelement, for being known using the first information and the second information to dispatch account to be identified Not, the recognition result of dispatch account to be identified is obtained.
In the embodiment of the present application, it is preferred that the first information includes the article sequence that dispatch account to be identified is delivered, identification As a result subelement is determined, comprising:
Word sequence generation unit, for the sequence according to the article in the article sequence in the first information, sequence obtains every The each word successively to sort in piece article, generates the word sequence of article sequence；
Term vector sequence generating unit obtains and word sequence pair for each word in word sequence to be converted into term vector The term vector sequence answered；
First identification information generation unit, for term vector sequence inputting to dispatch account identification model, to be based on dispatch Account identification model handles term vector sequence, obtains the first identification information；
Second identification information generation unit obtains second for handling based on the second information the first identification information Identification information.
In the embodiment of the present application, it is preferred that the second identification information generation unit, be specifically used for by the first identification information with Second information input to attention Mechanism Model, based on attention Mechanism Model to the first identification information and the second information at Reason, obtains the second identification information.
In the embodiment of the present application, it is preferred that the first identification information generation unit, specifically for by the word of article sequence to Sequence inputting is measured to dispatch account identification model, term vector sequence is handled based on dispatch account identification model, output is made For the expression of the article sequence of the first identification information.
Further, in the embodiment of the present application, it is preferred that the second identification information generation unit is specifically used for according to the Two information handle the expression of article sequence, obtain the scoring of every article in article sequence；Based on every identified The scoring of article handles the expression of article sequence, obtains the final expression of the article sequence as the second identification information.
In the embodiment of the present application, it is preferred that the second information acquisition unit, comprising:
Attribute determining unit, for determining at least one account attribute and at least one account of dispatch account to be identified The attribute value of each account attribute in attribute；
Property parameters determination unit, for determining that attribute corresponding with account attribute is joined according to the attribute value of account attribute Number；
Second information generating unit, it is corresponding for each account attribute at least one account attribute based on determined by Property parameters generate second information relevant to dispatch account to be identified, each account attribute pair at least one account attribute The property parameters answered are an element in the second information.
It further, further include that identification model generates in a kind of dispatch account identification device provided by the embodiments of the present application The detailed construction schematic diagram of unit, identification model generation unit asks Parameter Map 13.
As shown in figure 13, identification model generation unit, comprising:
Training sample determination unit 131, for determining at least one dispatch account training sample；
Training samples information acquiring unit 132, for obtaining relevant to the dispatch behavior of account training sample of sending the documents One information and the second information relevant to the dispatch account attribute of account training sample；
The first identification information of training sample generation unit 133, it is true for the first information based on dispatch account training sample Surely the term vector sequence of dispatch account training sample, by term vector sequence inputting to Recognition with Recurrent Neural Network model, based on circulation mind Term vector sequence is handled through network model, obtains the first identification information；
The second identification information of training sample generation unit 134, for extremely infusing the first identification information and the second information input Meaning power mechanism module, handles the first identification information and the second information based on attention mechanism module, obtains the second identification Information；
Predicting unit 135 of classifying obtains classification prediction result for carrying out classification prediction to the second identification information；
It is reversed to adjust unit 136, for based on classification prediction result reversely adjust in Recognition with Recurrent Neural Network model for pair It is used in the parameter and attention mechanism module that term vector sequence is handled to the first identification information and the second information The parameter of reason, with generate corresponding with Recognition with Recurrent Neural Network model dispatch account identification model and with attention mechanism module pair The attention Mechanism Model answered.
In the embodiment of the present application, it is preferred that classification predicting unit is specifically used for: the second identification information is input to Softmax function carries out classification prediction to the second identification information, obtains classification prediction result.
In the embodiment of the present application, it is preferred that reversely adjust unit, be specifically used for: being utilized and divided based on gradient descent algorithm Class prediction result reversely adjusts parameter and attention machine in Recognition with Recurrent Neural Network model for being handled term vector sequence Parameter in molding block for being handled the first identification information and the second information.
Dispatch account display device provided in an embodiment of the present invention is introduced below, dispatch account described below is aobvious Showing device is regarded as, the terminal dispatch account display methods that embodiment provides to realize the present invention, the program mould of required setting Block.Dispatch account display device context described below can be corresponded to each other with above-described dispatch account display methods content Reference.
Figure 14 is a kind of structural block diagram of account display device of sending the documents provided by the embodiments of the present application, which can be applied to Terminal, referring to Fig.1 4, which may include:
Account of sending the documents shows request reception unit 141, for receiving dispatch account display request；
Dispatch account acquiring unit 142, for obtaining each dispatch account of dispatch account display request instruction；
Dispatch account number classification determination unit 143, for determining dispatch account number classification belonging to each dispatch account, account of sending the documents Number classification is that rubbish sends the documents account number classification or non-junk is sent the documents account number classification；
Dispatch account display unit 144, for regular, the display based on dispatch account display relevant to dispatch account number classification Each dispatch account.
In the embodiment of the present application, it is preferred that dispatch account shows that rule is used to indicate shielding and belongs to rubbish dispatch account The dispatch account of classification, alternatively, the display priority for being used to indicate the dispatch account for belonging to non-junk dispatch account number classification is higher than Belong to the display priority of the dispatch account of rubbish dispatch account number classification.
Further, the embodiment of the present invention also provides a kind of storage medium, which realizes that account is known for storing The program of other method；Or the program of account display methods is realized in storage.
The detailed description of program in relation to storing in storage medium provided by the embodiments of the present application can refer to above-described embodiment, This will not be repeated here.
The embodiment of the present application provides a kind of identification of account, display device and storage medium, passes through determination dispatch account to be identified Number；Obtain the first information relevant to the dispatch dispatch behavior of account to be identified；Using the first information to dispatch account to be identified It is identified, obtains the mode of the recognition result of dispatch account to be identified, so that based on the dispatch row with dispatch account to be identified The identification to dispatch account to be identified is realized for the relevant first information, reduces the prior art according to individually to hair to be identified It is caused to dispatch account to be identified when the recognition result of every article realizes the identification to dispatch account to be identified in literary account Recognition result inaccuracy, cannot effectively identify rubbish send the documents account the problem of.
Each embodiment in this specification is described in a progressive manner, the highlights of each of the examples are with other The difference of embodiment, the same or similar parts in each embodiment may refer to each other.For device disclosed in embodiment For, since it is corresponded to the methods disclosed in the examples, so being described relatively simple, related place is said referring to method part It is bright.
Professional further appreciates that, unit described in conjunction with the examples disclosed in the embodiments of the present disclosure And algorithm steps, can be realized with electronic hardware, computer software, or a combination of the two, in order to clearly demonstrate hardware and The interchangeability of software generally describes each exemplary composition and step according to function in the above description.These Function is implemented in hardware or software actually, the specific application and design constraint depending on technical solution.Profession Technical staff can use different methods to achieve the described function each specific application, but this realization is not answered Think beyond the scope of this invention.
The step of method described in conjunction with the examples disclosed in this document or algorithm, can directly be held with hardware, processor The combination of capable software module or the two is implemented.Software module can be placed in random access memory (RAM), memory, read-only deposit Reservoir (ROM), electrically programmable ROM, electrically erasable ROM, register, hard disk, moveable magnetic disc, CD-ROM or technology In any other form of storage medium well known in field.
The foregoing description of the disclosed embodiments enables those skilled in the art to implement or use the present invention. Various modifications to these embodiments will be readily apparent to those skilled in the art, as defined herein General Principle can be realized in other embodiments in the case where not departing from core of the invention thought or scope.Therefore, originally Invention is not intended to be limited to the embodiments shown herein, and is to fit to and the principles and novel features disclosed herein Consistent widest scope.

Claims (15)

1. a kind of account recognition methods characterized by comprising
Determine dispatch account to be identified；
Obtain the first information relevant to the dispatch dispatch behavior of account to be identified；
The dispatch account to be identified is identified using the first information, obtains the identification of the dispatch account to be identified As a result.
2. the method according to claim 1, wherein described utilize the first information to the dispatch to be identified Account is identified, the recognition result of the dispatch account to be identified is obtained, comprising:
Obtain second information relevant to the dispatch account attribute of account to be identified；
The dispatch account to be identified is identified using the first information and the second information, obtains the dispatch to be identified The recognition result of account.
3. according to the method described in claim 2, it is characterized in that, the first information includes the dispatch account hair to be identified The article sequence of table, it is described that the dispatch account to be identified is identified using the first information and the second information, it obtains The recognition result of the dispatch account to be identified, comprising:
According to the sequence of the article in the article sequence in the first information, sequentially obtains and successively sort in every article Each word, generate the word sequence of the article sequence；
Each word in the word sequence is converted into term vector, obtains term vector sequence corresponding with the word sequence；
By the term vector sequence inputting to dispatch account identification model, based on the dispatch account identification model to institute's predicate to Amount sequence is handled, and the first identification information is obtained；
First identification information is handled based on second information, obtains the second identification information.
4. according to the method described in claim 3, it is characterized in that, described believe first identification based on second information Breath is handled, and the second identification information is obtained, comprising:
By first identification information and second information input to attention Mechanism Model, it is based on the attention mechanism mould Type handles first identification information and the second information, obtains the second identification information.
5. according to the method described in claim 3, it is characterized in that, described know the term vector sequence inputting to dispatch account Other model handles the term vector sequence based on the dispatch account identification model, obtains the first identification information, packet It includes:
By the term vector sequence inputting of the article sequence to dispatch account identification model, identified based on the dispatch account Model handles the term vector sequence, exports the expression of the article sequence as the first identification information.
6. according to the method described in claim 5, it is characterized in that, described believe first identification based on second information Breath is handled, and the second identification information is obtained, comprising:
The expression of the article sequence is handled according to second information, obtains every text in the article sequence The scoring of chapter；
The expression of the article sequence is handled based on the scoring of identified every article, is obtained as the second identification letter The final expression of the article sequence of breath.
7. according to the method described in claim 2, it is characterized in that, the account category of the acquisition and the dispatch account to be identified Relevant second information of property, comprising:
Determine each institute at least one account attribute and at least one described account attribute of the dispatch account to be identified State the attribute value of account attribute；
Property parameters corresponding with the account attribute are determined according to the attribute value of the account attribute；
Based on the corresponding property parameters of the account attribute each at least one identified described account attribute, second is generated Information, the corresponding property parameters of each account attribute are one in second information at least one described account attribute A element.
8. according to the method described in claim 2, it is characterized by further comprising:
Determine at least one dispatch account training sample；
It obtains the first information relevant to the dispatch dispatch behavior of account training sample and trains sample with the dispatch account This relevant second information of account attribute；
The term vector sequence that the dispatch account training sample is determined based on the first information of the dispatch account training sample, will The term vector sequence inputting is to Recognition with Recurrent Neural Network model, based on the Recognition with Recurrent Neural Network model to the term vector sequence It is handled, obtains the first identification information；
By first identification information and the second information input to attention mechanism module, it is based on the attention mechanism module pair First identification information and the second information are handled, and the second identification information is obtained；
Classification prediction is carried out to second identification information, obtains classification prediction result；
Based on the classification prediction result reversely adjust in the Recognition with Recurrent Neural Network model for the term vector sequence into It is used to handle first identification information and the second information in the parameter of row processing and the attention mechanism module Parameter, with generate corresponding with Recognition with Recurrent Neural Network model dispatch account identification model and with the attention mechanism mould The corresponding attention Mechanism Model of block.
9. according to the method described in claim 8, it is characterized in that, described carry out classification prediction to second identification information, Obtain classification prediction result, comprising:
Second identification information is input to softmax function, classification prediction is carried out to second identification information, is divided Class prediction result.
10. a kind of account display methods characterized by comprising
Receive dispatch account display request；
It obtains the dispatch account display request and indicates each dispatch account；
Determine that dispatch account number classification belonging to each dispatch account, the dispatch account number classification are rubbish dispatch account number classification Or non-junk dispatch account number classification；
Rule is shown based on dispatch account relevant to the dispatch account number classification, shows each dispatch account.
11. a kind of account identification device characterized by comprising
Dispatch account determination unit to be identified, for determining dispatch account to be identified；
First information acquiring unit, for obtaining the first information relevant to the dispatch dispatch behavior of account to be identified；
Recognition result determination unit obtains institute for identifying using the first information to the dispatch account to be identified State the recognition result of dispatch account to be identified.
12. a kind of server characterized by comprising at least one processor and at least one processor；The memory is deposited Program is contained, the processor calls the program of the memory storage, and described program is any for realizing such as claim 1-9 Account recognition methods described in one.
13. a kind of account display device characterized by comprising
Account of sending the documents shows request reception unit, for receiving dispatch account display request；
Dispatch account acquiring unit, for obtaining each dispatch account of the dispatch account display request instruction；
Dispatch account number classification determination unit, for determining dispatch account number classification belonging to each dispatch account, the dispatch Account number classification is rubbish dispatch account number classification or non-junk dispatch account number classification；
Dispatch account display unit, for showing each based on dispatch account relevant to dispatch account number classification display rule A dispatch account.
14. a kind of terminal characterized by comprising processor and memory, the processor and memory pass through communication Bus is connected；Wherein, the processor, for calling and executing the program stored in the memory；The memory, is used for Program is stored, described program is for realizing account display methods as claimed in claim 10.
15. a kind of storage medium, which is characterized in that realize that account as described in any one of claims 1-9 is known for storing The program of other method；Or the program of account display methods as claimed in claim 10 is realized in storage.
CN201810288695.6A 2018-03-30 2018-03-30 Account identification and display method and device, server, terminal and storage medium Active CN110321546B (en)

Priority Applications (1)

Application Number Priority Date Filing Date Title
CN201810288695.6A CN110321546B (en) 2018-03-30 2018-03-30 Account identification and display method and device, server, terminal and storage medium

Applications Claiming Priority (1)

Application Number Priority Date Filing Date Title
CN201810288695.6A CN110321546B (en) 2018-03-30 2018-03-30 Account identification and display method and device, server, terminal and storage medium

Publications (2)

Publication Number Publication Date
CN110321546A true CN110321546A (en) 2019-10-11
CN110321546B CN110321546B (en) 2023-05-02

Family

ID=68112160

Family Applications (1)

Application Number Title Priority Date Filing Date
CN201810288695.6A Active CN110321546B (en) 2018-03-30 2018-03-30 Account identification and display method and device, server, terminal and storage medium

Country Status (1)

Country Link
CN (1) CN110321546B (en)

Cited By (4)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN110765973A (en) * 2019-10-31 2020-02-07 上海掌门科技有限公司 Account type identification method and device
CN111353019A (en) * 2020-02-25 2020-06-30 上海昌投网络科技有限公司 WeChat public number topic classification method and device
CN112487176A (en) * 2020-11-26 2021-03-12 北京智源人工智能研究院 Social robot detection method, system, storage medium and electronic device
CN114385903A (en) * 2020-10-22 2022-04-22 腾讯科技（深圳）有限公司 Application account identification method and device, electronic equipment and readable storage medium

Citations (3)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
US20050268212A1 (en) * 2004-05-28 2005-12-01 Michael Dagel System, apparatus, and method for desktop-based creation and publication of a periodic community newsletter
US20140068459A1 (en) * 2012-08-28 2014-03-06 Facebook, Inc. Providing a locality viewport through a social networking system
CN106886518A (en) * 2015-12-15 2017-06-23 国家计算机网络与信息安全管理中心 A kind of method of microblog account classification

Patent Citations (3)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
US20050268212A1 (en) * 2004-05-28 2005-12-01 Michael Dagel System, apparatus, and method for desktop-based creation and publication of a periodic community newsletter
US20140068459A1 (en) * 2012-08-28 2014-03-06 Facebook, Inc. Providing a locality viewport through a social networking system
CN106886518A (en) * 2015-12-15 2017-06-23 国家计算机网络与信息安全管理中心 A kind of method of microblog account classification

Non-Patent Citations (1)

* Cited by examiner, † Cited by third party
Title
张宇翔 等: "新浪微博反垃圾中特征选择的重要性分析", 《通信学报》 *

Cited By (6)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN110765973A (en) * 2019-10-31 2020-02-07 上海掌门科技有限公司 Account type identification method and device
CN110765973B (en) * 2019-10-31 2023-07-04 上海掌门科技有限公司 Account type identification method and device
CN111353019A (en) * 2020-02-25 2020-06-30 上海昌投网络科技有限公司 WeChat public number topic classification method and device
CN114385903A (en) * 2020-10-22 2022-04-22 腾讯科技（深圳）有限公司 Application account identification method and device, electronic equipment and readable storage medium
CN114385903B (en) * 2020-10-22 2024-02-06 腾讯科技（深圳）有限公司 Application account identification method and device, electronic equipment and readable storage medium
CN112487176A (en) * 2020-11-26 2021-03-12 北京智源人工智能研究院 Social robot detection method, system, storage medium and electronic device

Also Published As

Publication number Publication date
CN110321546B (en) 2023-05-02

Similar Documents

Publication Publication Date Title
CN110032632A (en) Intelligent customer service answering method, device and storage medium based on text similarity
Sun et al. Near real-time twitter spam detection with machine learning techniques
CN110321546A (en) Account identification, display methods, device, server, terminal and storage medium
CN105808590B (en) Search engine implementation method, searching method and device
CN109299344A (en) The generation method of order models, the sort method of search result, device and equipment
CN107832432A (en) A kind of search result ordering method, device, server and storage medium
CN106600052A (en) User attribute and social network detection system based on space-time locus
CN109388675A (en) Data analysing method, device, computer equipment and storage medium
CN112380859A (en) Public opinion information recommendation method and device, electronic equipment and computer storage medium
WO2018118601A1 (en) Conversion of static images into interactive maps
CN112667877A (en) Scenic spot recommendation method and equipment based on tourist knowledge map
CN111538844A (en) Target field knowledge base generation and problem solution method and device
CN109766557A (en) A kind of sentiment analysis method, apparatus, storage medium and terminal device
CN108255706A (en) Edit methods, device, terminal device and the storage medium of automatic test script
CN106844407A (en) Label network production method and system based on data set correlation
US20210110277A1 (en) Textual entailment
WO2021151305A1 (en) Sample analysis method, apparatus, electronic device, and medium based on missing data
CN111563158A (en) Text sorting method, sorting device, server and computer-readable storage medium
CN111538852B (en) Multimedia resource processing method, device, storage medium and equipment
CN110210038A (en) Kernel entity determines method and its system, server and computer-readable medium
CN104408144B (en) The detection method and device of web search keyword
CN110245310B (en) Object behavior analysis method, device and storage medium
CN111930897B (en) Patent retrieval method, device, electronic equipment and computer-readable storage medium
Chipofya et al. A metaheuristic approach for efficient and effective sketch-to-metric map alignment
CN110069686A (en) User behavior analysis method, apparatus, computer installation 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