CN109015690A - active interactive dialogue robot system and method - Google Patents

Abstract

A kind of active interactive dialogue robot system and method provided by the invention, module is handled including pre-sales consultation module and mid-sales, pre-sales consultation module includes words art management module, response means, satisfaction investigation module and line module, and it includes authentication module, session module and command execution module that module is handled in mid-sales；The system includes that can be verified before user's transacting business to user with upper module by setting, and whether identification user is to handle in person, improves safety, avoids bringing unnecessary trouble to user.

Description

Active interactive dialogue robot system and method

Technical field

The present invention relates to human-computer interaction technique field more particularly to a kind of active interactive dialogue robot systems and method.

Background technique

Existing customer service robot can carry out communication exchange in such a way that user puts question to user, and feature is triggering item The active that part relies on user is putd question to, and robot knowledge based library and support service type passive type are answered, and can not be provided for user It takes the initiative in offering a hand, Chinese patent CN107053208A provides a kind of active interactive dialogue robot system and the system is actively right The method of words, the system include words art management module, problem generation module, answer acquisition module, answer correction verification module and order Execution module, the robot system have active dialogue ability, extract art in scene simultaneously in practical dialog procedure It is configured in dialogue robot system, dialogue robot carries out the acquisition and verification of answer based on words art configuration, to realize Save a large amount of artificial purposes；Although the robot has active dialogue ability, but since the robot is to pass through question and answer Mode be user's transacting business, cannot be distinguished user whether be in propria persona to handle, lack safety, there are other people maliciously to emit With the risk of subscriber identity information transacting business, it is easy to bring unnecessary trouble to user.

Summary of the invention

In order to solve the above-mentioned technical problems, the present invention provides a kind of active interactive dialogue robot system and methods.

Specific technical solution of the present invention is as follows:

Module is handled in a kind of active interactive dialogue robot system, including pre-sales consultation module and mid-sales；

Pre-sales consultation module includes words art management module, response means, satisfaction investigation module and line module；

Words art management module: for creating industry words art library, customization meets the dialogue automatic question answering corpus of business；

Response means: for acquiring the problem of line module is sent, corresponding answer is matched in industry words art library and is incited somebody to action The answer is sent to line module；

Satisfaction investigation module: for when user seeks advice from or after consulting to the satisfaction to the secondary consulting of user It is investigated；

Line module: the answer for proposing problem for user, answering a question and receive response means transmission；

It includes authentication module, session module and command execution module that module is handled in mid-sales；

Authentication module: for verifying whether user is to handle in person；

Session module: for receiving the instruction of authentication module transmission, problem is proposed to user according to application scenarios, acquisition is used The answer at family, the legitimacy that verification user answers send to command execution module according to check results and instruct；

Command execution module: for executing corresponding business order or exiting；

Authentication module includes speech verification module；

Speech verification module: for sending speech verification information to user, judge whether it is and handle in person, if it is me It handles, then sends and instruct to session module, handled if not me, then send and instruct to command execution module, command content To exit.

Wherein, it is mentioned in the present invention execute corresponding business order and can be to place an order for client buy commodity or be visitor Buy the business such as train ticket in family；Speech verification is the prior art, can be using dialing voice call to user or other can be real The mode of existing verifying purpose, but it is not limited to a kind of this mode；The legitimacy that verification user mentioned in the present invention answers is conventional Technology, including all methods that can be realized the purpose, a kind of dialog interaction as disclosed in Chinese patent CN107053208A The method of robot system and the system active interlocution, this method is in answer correction verification module in the legitimacy that verification user answers Whether it is text, picture, sound that the dimension for illustrating verification judgement includes: the file format of answer, the content of answer whether include Whether particular text is specific enumerated value, and the legitimacy that verification user mentioned in the present invention answers includes above method, but It is not limited to a kind of this form；Speech verification module includes all devices that can be realized speech verification, such as Chinese patent A kind of cell phone payment system disclosed in CN104134140A, the speech verification module in the system includes microphone, loudspeaking Device and speech recognition program, speech verification module of the invention include apparatus above but are not limited to a kind of this form；In the present invention If art management module include all methods that can be realized the function, as Chinese patent CN107053208A provides one kind The method of active dialog interaction robot system and the system active interlocution, art manages mould in the dialogue robot system Block is mainly used for managing all problem during active interlocution, art state during oneself management active interlocution, this hair Art management module can be realized using same method in bright, but be not limited to a kind of this mode.

Further, dialogue robot system further includes Failure elimination module, and Failure elimination module includes:

Failure monitoring module: the working condition of moment acquisition dialogue robot system, and judge the work for talking with robot Whether state is abnormal, if abnormal, send and instructs to fault processing module, if normally, be not processed；

Fault processing module: for calling relevant staff, staff will be at first time be to failure Reason.

Further, failure monitoring module includes answering velocity estimated module, satisfaction investigation judgment module, speech verification Judgment module executes judgment module；

Answer velocity estimated module: for judging that line module sends the time T1 and dialogue robot system answer of problem The time T2 of problem, judges the size of time difference threshold value Tn and time difference T2-T1, if T2-T1 >=Tn, to troubleshooting Module sends instruction, if T2-T1 < Tn, sends and instructs to satisfaction investigation judgment module；

Satisfaction investigation judgment module: for judge user consulting after whether to user send satisfaction investigation note Or voice call sends to speech verification judgment module and instructs, if not sent, to fault processing module if sent Send instruction；

Speech verification judgment module: for judging whether to user's execution speech verification before executing related service, if It executes, then to judgment module transmission instruction is executed, if be not carried out, is instructed to fault processing module transmission；

Execute judgment module: for judge to talk with robot whether for user handle related industry after user selects to buy Business instructs if it is, being not processed if it is not, then sending to fault processing module.

Further, authentication module further includes image authentication module, and image authentication module includes:

First facial feature obtains module: for obtaining and identifying multiple facial characteristics of user, and will identify that more A facial characteristics is sent to first facial Characteristic Contrast module；

Identity information acquisition module: it for scanning the identity document of user, identifies the certificate photograph on identity document, will demonstrate,prove Part photo is sent to first facial Characteristic Contrast module；

First facial Characteristic Contrast module: the facial characteristics and identity that module is sent are obtained for receiving first facial feature The certificate photograph that data obtaining module is sent, and all facial characteristics are compared with certificate photograph, it is similar to calculate the two Tq is spent, and maximum similarity Tq is compared with similarity threshold Tp, as Tq >=Tp, is then proved to be successful, is sent out to session module Instruction is sent, as Tq < Tp, then authentication failed, sends to command execution module and instruct, and command content is to exit.

Wherein, it includes all devices that can be realized acquisition user's face feature that first facial feature, which obtains module, as in A kind of access control system of the control Image Acquisition of facial characteristics disclosed in state patent CN107749110A, the Image Acquisition in the patent Device includes camera and image processing unit, and camera is connected with image processing unit, is carried out for the image to monitoring area Acquisition, facial recognition modules of the invention include the image acquisition device in the patent, but are not limited to a kind of this form；By user face Portion's feature is compared as routine techniques with certificate photograph, including all modes that can be realized the effect, such as Chinese patent CN107742106A discloses a kind of facial match method and apparatus based on automatic driving vehicle, the facial match method are as follows: The face-image for taking the passenger of automatic driving vehicle is obtained, the face of passenger is determined based on human face recognition model trained in advance The similarity of image and the face-image of each known abducted personnel determines whether passenger is abducted personnel according to similarity, For another example Chinese patent CN103927747A discloses a kind of face matching space register method based on face biological characteristic, this method It include: the method for integrated structure light and binocular stereo vision acquisition patient's spatial surface three-dimensional point cloud in patient space, two It ties up and nose and eyes is identified by the ASM method based on statistical information on image, the medical image image of preoperative acquisition is carried out Resurfacing obtains number of people threedimensional model, extracts face by carrying out principal component analysis to local domain on the threedimensional model Feature goes out corresponding nose and eyes to the further clustering recognition of these features, obtains the corresponding points of nose and eyes, calculates people Initial transformation between head model and three-dimensional point cloud recycles closest approach algorithm (ICP) to complete number of people threedimensional model The registration of the point cloud in the model space and patient space, it is mentioned in the present invention that user's face feature and certificate photograph are subjected to matching packet Include the side using both the method for both comparison similarities and the calculating of the corresponding points by obtaining nose and eyes initialization matrix Method, but it is not limited to a kind of this form；

Further, authentication module further includes double authentication module, and double authentication module includes:

Double facial information identifies submodule: for obtaining two groups of facial characteristics of two users simultaneously, every group of face is special Sign includes multiple facial characteristics of one of user, and two groups of facial characteristics are sent to double facial characteristics contrast module；

Double identity information acquisition module: for scanning the identity document of two users and obtaining on two identity documents Certificate photograph, two certificate photographs that will acquire are sent to double facial characteristics contrast module；

Double facial characteristics contrast module: the two groups of facial characteristics sent for receiving double facial information identification submodule The certificate photograph sent with double identity information acquisition module, respectively compares two groups of facial characteristics with corresponding certificate photograph It is right, both calculate similarity Tm, and two maximum similarity Tm1 and Tm2 be compared with similarity threshold T3, when Tm1 >= T3 and Tm2 >=T3, then be proved to be successful, and sends and instructs to session module, as Tm1 < T3 or Tm2 < T3, then executes mould to order Block sends instruction, and command content is to exit.

Further, session module includes:

Problem memory module: the problem of for storing the lower enquirement in need of a certain scene, by enquirement in need ask Topic is arranged in order, and when receiving the instruction of problem generation module transmission, all the problem of being arranged in order is sent to and are asked Inscribe generation module；

Problem generation module: it for receiving the instruction of authentication module transmission, and sends and instructs to problem memory module, will connect The problem of arranged in sequence received, is sent in sequence to line module, and sends and instruct to answer acquisition module；

Answer acquisition module: for acquiring line module to the answer content of problem, and the answer content is sent to and is answered Case correction verification module；

Answer correction verification module: for receiving the answer content of answer acquisition module transmission, the legal of the answer content is verified Property, it to command execution module or repeats that module is putd question to send instruction according to check results；

It repeats to put question to module: for the answer of empty user input, and it is illegal to line module repetition transmission verification Problem；

Answer correction verification module includes:

Numeric class answer judgment module: for user's answer content to be converted to text, and judge user answer whether For number, if it is, identifying digital number Xn, and Xn is sent to digit length judgment module, if it is not, then to text Class answer judgment module sends instruction；

Digit length judgment module: for receive numeric class answer judgment module transmission digital number Xn, judge Xn with The size of digital number threshold X 1 sends to problem generation module and instructs if Xn=X1, sends to line module next Problem sends instruction to enquirement module is repeated if Xn ≠ X1；

Text class answer judgment module: for judging whether the answer of user is text class answer, if so, identification text In word content, and word content is sent to same text judgment module, if it is not, then sending to scanning judgment module Instruction；

Same text judgment module: for receiving the word content of text class answer judgment module transmission, judge the text Whether there is word identical with default answer in content, instructed if so, then being sent to problem generation module, is sent out to line module Next problem is sent, if it is not, to repeating that module is putd question to send instruction；

Scanning judgment module: for scanning the certificate of user's offer, and judging whether the certificate of the scanning is qualified, if closed Lattice then send to problem generation module and instruct, propose next problem, if unqualified, refer to enquirement module transmission is repeated It enables.

Further, scanning judgment module includes:

Content matching module: whether the word content on certificate for judging user's offer is consistent with problem, if phase Symbol sends to certificate matching module and instructs, if be not consistent, sends instruction to enquirement module is repeated；

Certificate matching module: the instruction sent for reception content matching module identifies and judges to affix one's seal on certificate interior Hold and whether be consistent with word content, if be consistent, sends and instruct to problem generation module, next problem is proposed, if not Be consistent, then determine it is unqualified, to repeat put question to module send instruction.

Further, problem generation module includes:

Number judgment module: for after sending line module for problem, judging that the number of certificate is handled in user's selection P sends instruction to an answer acquisition module, if P > 1, sends and instruct to answer-mode module if P=1；

Answer-mode module: for judge multiple users answer a certain problem answer it is whether identical, if identical, to One time answer acquisition module sends instruction, instructs if it is different, then sending to multiple answer acquisition module；

Answer acquisition module includes:

Answer acquisition module: it is sent out for disposably acquiring user to the answer content of problem, and by the answer content It send to answer correction verification module, examines the successfully next problem of rear line module transmission；

Multiple answer acquisition module: for multi collect user to the answer content of problem, the people of times of collection and user Number is identical, and all answer contents are sent to answer correction verification module according to acquisition order, and all answer contents verify success Rear line module sends next problem.

Further, dialogue robot system further includes review module, and review module includes:

List Generating Module: for after user answers all problems and verifies successfully, by all problems and corresponding The answer that user answers generates a list and is sent to line module；

Review submodule: for receiving the instruction of line module transmission and identifying the command content, if command content is It is not necessary to modify then send to secondary identity authentication module and instruct, if command content is to need to modify, send out to problem selecting module Send instruction；

Problem selecting module: the instruction sent for receiving review submodule, and transfer the problem of need to modifying and be sent to use Family module；

Secondary identity authentication module: for carrying out authentication again after user sends it is not necessary to modify instruction, and according to Verification result sends to command execution module and instructs；

Secondary identity authentication module includes:

Second facial characteristics obtains module: the face for acquiring user again after user sends it is not necessary to modify instruction is special Sign, and it is sent to the second facial characteristics contrast module；

Second facial characteristics contrast module: the facial characteristics and first of module acquisition is obtained for comparing the second facial characteristics Facial characteristics obtains the facial characteristics that module obtains, and judges whether to match, if it does, then referring to command execution module transmission It enables, command content is to execute corresponding business, is instructed if it does not match, sending to command execution module, and command content is to move back Out.

Further, satisfaction investigation module includes:

Mouth Expression Recognition module: the facial characteristics of several clients is acquired when for communicating with client, and identifies client Mouth expression, specifically: the number F2 of the number F1, smile that are in a pout, the number F3 for mouth of beeping, the number F4 that shows one's ivories；

Eye Expression Recognition module: the facial characteristics of several clients is acquired when for communicating with client, and identifies client Eye expression, specifically: the number G2 of the number G1, the supercilious look that stare；

Supercilium Expression Recognition module: the facial characteristics of several clients is acquired when for communicating with client, and identifies client Eye expression, specifically: the number H1 for choosing eyebrow, the number H2 to frown；

Headwork identification module: the facial characteristics of several clients is acquired when for communicating with client, and identifies client Eye expression, specifically: the number J1 to shake the head, the number J2 to nod；

Score computing module: for collecting mouth Expression Recognition module, eye Expression Recognition module, supercilium Expression Recognition mould Expression that block and headwork identification module recognize simultaneously calculates satisfaction Q, and Q is calculated by following methods；Q=-F1 × f1 + F21 × f2-F3 × f3+F4 × f4-G1 × g1-G2 × g2-H1 × h1-H2 × h2-J1 × j1+J2 × j2, wherein f1+f2 + f3+f4+g1+g2+h1+h2+j1+j2=100%.

Active interactive dialogue robot system provided by the invention, by the way that facial recognition modules, authentication module, dialogue are arranged Module, command execution module and line module can identify whether user is to handle in person, is mentioned before user handles certificate High security reduces other people and maliciously falsely uses the risk of subscriber identity information transacting business, avoids bringing unnecessary fiber crops to user It is tired.

Detailed description of the invention

The structural block diagram of 1 active interactive dialogue robot system of Fig. 1 embodiment；

The structural block diagram of 2 Failure elimination module of Fig. 2 embodiment；

The structural block diagram of 3 failure monitoring module of Fig. 3 embodiment；

The structural block diagram of 4 image authentication module of Fig. 4 embodiment；

The structural block diagram of the double authentication module of Fig. 5 embodiment 5；

The structural block diagram of 6 session module of Fig. 6 embodiment；

The structural block diagram of the scanning judgment module of Fig. 7 embodiment 7；

The structural block diagram of 8 problem generation module of Fig. 8 embodiment；

The structural block diagram of 9 review module of Fig. 9 embodiment；

The structural block diagram of 9 secondary identity authentication module of Figure 10 embodiment；

The structural block diagram of 10 satisfaction investigation module of Figure 11 embodiment.

Specific embodiment

The invention will be further described with reference to the accompanying drawings and examples, and the following example is for explaining only the invention Summary of the invention, be not intended to limit the scope of the invention.

Embodiment 1

The present invention provides a kind of active interactive dialogue robot systems, as shown in Figure 1, including pre-sales 1 He of consultation module Module 2 is handled in mid-sales；

Pre-sales consultation module 1 includes words art memory module 11, response means 12, satisfaction investigation module 13 and line module 14；

Words art memory module 11: for creating industry words art library, customization meets the dialogue automatic question answering corpus of business；

Response means 12: for acquiring the problem of line module 14 is sent, corresponding answer is matched in industry words art library And the answer is sent to line module 14；

Satisfaction investigation module 13: for when user seeks advice from or after consulting to the satisfaction to the secondary consulting of user Degree is investigated；

Line module 14: the answer for proposing problem for user, answering a question and receive the transmission of response means 12；

It includes authentication module 21, session module 22 and command execution module 23 that module 2 is handled in mid-sales；

Authentication module 21: for verifying whether user is to handle in person；

Session module 22: for receiving the instruction of the transmission of authentication module 21, problem is proposed to user according to application scenarios, is adopted Collect the answer of user, the legitimacy that verification user answers sends to command execution module 23 according to check results and instructs；

Command execution module 23: for executing corresponding business order or exiting；

Authentication module 21 includes speech verification module 211；

Speech verification module 211: for judging whether it is and handling in person to user's transmission speech verification information, if it is I handles, then sends and instruct to session module 22, handle if not me, then sends and instruct to command execution module 23, Command content is to exit.

Active interactive dialogue robot system provided in this embodiment, mainly with the above arrangement, can provide a user Consulting and purchase service, verify the identity of user before user buys commodity, identify whether to handle for me, if Authentication failed is then exited and is handled, if be proved to be successful, sends problem to user, is reduced other people and is maliciously falsely used user account purchase The risk for buying commodity avoids bringing unnecessary trouble to user.

Embodiment 2

The active interactive dialogue robot system of the present embodiment, the difference from embodiment 1 is that, as shown in Fig. 2, further It limits, dialogue robot system further includes Failure elimination module 3, and Failure elimination module 3 includes:

Failure monitoring module 31: the working condition of moment acquisition dialogue robot system, and judge the work for talking with robot Whether abnormal make state, if abnormal, send and instruct to fault processing module 32, if normally, be not processed；

Fault processing module 32: for calling relevant staff, staff will be carried out failure in first time Processing.

Active interactive dialogue robot system provided in this embodiment, by the way that Failure elimination module is arranged, when can be realized The working condition of monitoring dialogue robot is carved, once breaking down, relevant staff can called at the first time to event Barrier is handled, and the user experience of user is indirectly improved.

Embodiment 3

The present embodiment has carried out further restriction to failure monitoring module, as shown in figure 3, failure monitoring module 31 includes It answers velocity estimated module 311, satisfaction investigation judgment module 312, speech verification judgment module 313, execute judgment module 314；

Answer velocity estimated module 311: for judging that line module 14 sends time T1 and the dialogue system of robot of problem Unite the time T2 to answer a question, judges the size of time difference threshold value Tn and time difference T2-T1, if T2-T1 >=Tn, to event Hinder processing module 32 and send instruction, if T2-T1 < Tn, sends and instruct to satisfaction investigation judgment module 312；

Satisfaction investigation judgment module 312: for judge user consulting after whether to user send satisfaction investigation Short message or voice call send to speech verification judgment module 313 and instruct, if not sent, to failure if sent Processing module 32 sends instruction；

Speech verification judgment module 313: for judging whether to user's execution speech verification before executing related service, such as Fruit executes, then to the transmission instruction of judgment module 314 is executed, if be not carried out, instructs to the transmission of fault processing module 32；

Execute judgment module 314: for judge to talk with robot whether for user handle correlation after user selects to buy Business instructs if it is, being not processed if it is not, then sending to fault processing module 32.

Active interactive dialogue robot system provided in this embodiment, by specifically limiting failure monitoring module, various dimensions Judge robot system whether failure, to improve the accuracy of fault detection.

Embodiment 4

The present embodiment has carried out further restriction to authentication module, as shown in figure 4, authentication module 21 further includes that image is tested Module 212 is demonstrate,proved, image authentication module 212 includes:

First facial feature obtains module 2121: for obtaining and identifying multiple facial characteristics of user, and will identify that Multiple facial characteristics be sent to first facial Characteristic Contrast module 2123；

Identity information acquisition module 2122: for scanning the identity document of user, identifying the certificate photograph on identity document, Certificate photograph is sent to first facial Characteristic Contrast module 2123；

First facial Characteristic Contrast module 2123: the face that module 2121 is sent is obtained for receiving first facial feature The certificate photograph that feature and identity information acquisition module 2122 are sent, and all facial characteristics are compared with certificate photograph It is right, both calculate similarity Tq, and maximum similarity Tq be compared with similarity threshold Tp, as Tq >=Tp, then verify at Function sends to session module 22 and instructs, and as Tq < Tp, then authentication failed, sends to command execution module 23 and instruct, command content To exit.

This method is mainly for the business for needing to be handled with identity document, such as buys train ticket, applies for passport, visa； Active interactive dialogue robot system provided in this embodiment, it is main by obtaining user's face feature and user identity card respectively Certificate photograph on part, and the two is compared to the similarity for calculating the two, when similarity is higher than the threshold value of setting It is judged to handling in person, so that problem is sent to user, if similarity can determine that not it is to do in person lower than the threshold value of setting Reason, exits and handles, reduce other people and maliciously falsely use the risk of subscriber identity information transacting business, avoid bringing to user unnecessary Trouble.

Embodiment 5

The present embodiment further limits problem generation module, as shown in figure 5, authentication module 21 further includes double tests Module 213 is demonstrate,proved, double authentication module 213 includes:

Double facial information identifies submodule 2131: for obtaining two groups of facial characteristics of two users, every group of face simultaneously Portion's feature includes multiple facial characteristics of one of user, and two groups of facial characteristics are sent to double facial characteristics and compare mould Block 2133；

Double identity information acquisition module 2132: for scanning the identity document of two users and obtaining two identity documents On certificate photograph, two certificate photographs that will acquire are sent to double facial characteristics contrast module 2133；

Double facial characteristics contrast module 2133: two sent for receiving double facial information identification submodule 2131 Group facial characteristics and double identity information acquisition module 2132 send certificate photograph, respectively by two groups of facial characteristics with it is corresponding Certificate photograph is compared, and both calculates similarity Tm, and by two maximum similarity Tm1 and Tm2 and similarity threshold T3 into Row compares, and as Tm1 >=T3 and Tm2 >=T3, is then proved to be successful, and sends and instructs to session module 22, as Tm1 < T3 or Tm2 < T3 then sends to command execution module 23 and instructs, and command content is to exit.

Active interactive dialogue robot system provided in this embodiment can be realized auxiliary by the way that double authentication module is arranged It helps the function of handling, especially the elderly or is bad at the crowd using electronic product, can help to handle with through committing others, use When, the certificate photograph on the facial characteristics and corresponding identity document of personnel is handled by acquisition user and auxiliary and is carried out pair Than if the similarity of the two is on similarity threshold, it was demonstrated that one of them is auxiliary transactor person, can be sent out to user Problem is sent, if the similarity of wherein one or two is under similarity threshold, it was demonstrated that there is a people that the identity of oneself is not used Certificate will move out handle at this time, protects the legitimate rights and interests of user, user is avoided to have dust thrown into the eyes.

Embodiment 6

The present embodiment has carried out further restriction to session module, as shown in fig. 6, session module 22 includes:

Problem memory module 221: the problem of for storing the lower enquirement in need of a certain scene, by enquirement in need Problem is arranged in order, when receiving the instruction of the transmission of problem generation module 222, by all the problem of being arranged in order transmissions To problem generation module 222；

Problem generation module 222: refer to for receiving the instruction of the transmission of authentication module 2, and to the transmission of problem memory module 221 The problem of order, the arranged in sequence that will be received, is sent in sequence to line module 14, and sends and instruct to answer acquisition module 223；

Answer acquisition module 223: it is sent for acquiring line module 14 to the answer content of problem, and by the answer content To answer correction verification module 224；

Answer correction verification module 224: for receiving the answer content of the transmission of answer acquisition module 223, the answer content is verified Legitimacy, according to check results to command execution module 23 or repeat put question to module 225 send instruction；

It repeats to put question to module 225: for the answer of empty user input, and repeating to send verification not to line module 14 Legal problem；

Answer correction verification module 224 includes:

Numeric class answer judgment module 2241: for user's answer content to be converted to text, and judge the answer of user Whether it is number, if it is, identifying digital number Xn, and Xn is sent to digit length judgment module 2242, if not It is then to send and instruct to text class answer judgment module 2243；

Digit length judgment module 2242: for receiving the digital number Xn of the transmission of numeric class answer judgment module 2241, Judge that Xn and the size of digital number threshold X 1 send to problem generation module 222 and instruct, to user's mould if Xn=X1 Block 14 sends next problem, if Xn ≠ X1, sends instruction to enquirement module 225 is repeated；

Text class answer judgment module 2243: for judging whether the answer of user is text class answer, if so, identification Word content in text, and word content is sent to same text judgment module 2244, if it is not, then judging to scanning Module 2245 sends instruction；

Same text judgment module 2244: for receiving the word content of the transmission of text class answer judgment module 2243, sentence Whether have with default answer identical word, instruct if so, then sending to problem generation module 222 if breaking in the word content, Next problem is sent to line module 14, if it is not, to repeating that module 225 is putd question to send instruction；

Scanning judgment module 2245: for scanning the certificate of user's offer, and judge whether the certificate of the scanning is qualified, such as Fruit is qualified, then sends and instruct to problem generation module 222, propose next problem, if unqualified, puts question to module to repetition 225 send instruction.

Active interactive dialogue robot system provided in this embodiment, by specifically limiting answer check module, by more The legitimacy for judging user and answering of dimension, improves the accuracy of check results.

Embodiment 7

The present embodiment further limits scanning judgment module, as shown in fig. 7, scanning judgment module 2245 wraps It includes:

Content matching module 22451: whether the word content on certificate for judging user's offer is consistent with problem, such as Fruit is consistent, and sends and instructs to certificate matching module 22452, if be not consistent, sends instruction to enquirement module 225 is repeated；

Certificate matching module 22452: the instruction sent for reception content matching module 22451 identifies and judges certificate On the content affixed one's seal whether be consistent with word content, if be consistent, send and instruct to problem generation module 222, propose next A problem, if be not consistent, determine it is unqualified, to repeat put question to module 225 send instruction.

Scanning judgment module provided in this embodiment, the data that can be provided user is scanned, and judges whether to close Lattice, first determine whether the word content on the certificate of user's offer matches with problem, illustrate that user provides if mismatched Certificate and problem on require provide certificate be not inconsistent, put question to again at this time, user can provide the certificate being consistent with problem, If matching illustrate the certificate that user provides be correctly, the content that judges to affix one's seal on certificate at this time again whether with word content phase Symbol proves to affix one's seal correct, proposes next problem, if be not consistent, prove mistake of affixing one's seal, need to provide again if being consistent It affixes one's seal correct file, puts question to the problem to user again at this time, user can provide correct file of affixing one's seal.

Embodiment 8

The present embodiment has carried out further restriction to problem generation module, as shown in figure 8, number judgment module 2221: For after sending line module 14 for problem, judging that the number P of certificate is handled in user's selection, if P=1, to primary Answer acquisition module 2231 sends instruction, if P > 1, sends and instructs to answer-mode module 2222；

Answer-mode module 2222: for judge multiple users answer a certain problem answer it is whether identical, if identical, Instruction then is sent to an answer acquisition module 2231, is instructed if it is different, then being sent to multiple answer acquisition module 2232；

Answer acquisition module 223 includes:

Answer acquisition module 2231: for disposably acquiring user to the answer content of problem, and will be in the answer Appearance be sent to answer correction verification module 224, examine successfully rear line module 14 send next problem；

Multiple answer acquisition module 2232: for multi collect user to the answer content of problem, times of collection and user Number it is identical, and all answer contents are sent to answer correction verification module 224, all answer contents school according to acquisition order It tests successfully rear line module 14 and sends next problem.

The present embodiment further limits problem generation module, by judging that user's handles number, identifies One people handles or more people handle, and when number is 1 people, is then handled according to conventional program, when number is more people, then may be used It carries out more people while handling, judge whether the answer of a certain problem is identical at this time, if identical, only acquire returning for a user Answering can be verified, if it is not the same, then needing the answer of multi collect user, and verify the legitimacy repeatedly answered respectively, Such as: multiple students are based on going to school applying for passport together, and when asked questions are name, judging result is user to the problem Difference is answered, at this time the answer content of multi collect user, when asked questions are to go abroad reason, judging result is user to this The answer of problem is identical, only acquires the answer of a user at this time, saves the time and efforts of user, improves certificate acquiring Efficiency.

Embodiment 9

The present embodiment provides a kind of active interactive dialogue robot systems, as shown in Figure 9 and Figure 10, talk with robot system It further include review module 4, review module 4 includes:

List Generating Module 41: for after user answers all problems and verifies successfully, by all problems and correspondence User answer answer generate a list be sent to line module 14；

Review submodule 42: for receiving the instruction of the transmission of line module 14 and identifying the command content, if in instruction Hold for it is not necessary to modify then send to secondary identity authentication module 44 and instruct, if command content is to need to modify, select to problem Module 43 sends instruction；

Problem selecting module 43: the instruction sent for receiving review submodule 42, and transfer the problem of need to modifying transmission To line module 14；

Secondary identity authentication module 44: for carrying out authentication, and root again after user sends it is not necessary to modify instruction It sends and instructs to command execution module 23 according to verification result；

Secondary identity authentication module 44 includes:

Second facial characteristics obtains module 441: for acquiring the face of user again after user sends it is not necessary to modify instruction Portion's feature, and it is sent to the second facial characteristics contrast module 442；

Second facial characteristics contrast module 442: the face spy that module 441 acquires is obtained for comparing the second facial characteristics First facial of seeking peace feature obtains the facial characteristics that module 211 obtains, and judges whether to match, if it does, then executing to order Module 23 sends instruction, and command content is to execute corresponding business, if it does not match, referring to the transmission of command execution module 23 It enables, command content is to exit.

Active interactive dialogue robot system provided in this embodiment, by the way that review module is arranged, in all times of user It answers and verifies successfully one list of rear line transmission, judge whether answer information is all correct for user, if user awards It is not necessary to modify can handle certificate at this time for user, if necessary to modify, send to user and need the problem of modifying, at this time power User can modify to incorrect problem is answered, and user is avoided to occur caused by slip of the tongue or typing mistake during handling Certificate acquiring mistake；Secondary identity authentication module is further limited, by confirming answer content without repairing in user After changing, user is verified again, if the facial characteristics of user matches with the preceding facial characteristics acquired is handled at this time, Proof is still me at this time and is handling, and can handle corresponding certificate for user, if at this time the facial characteristics of user with do The facial characteristics acquired before reason mismatches, then proves that the user before handling personnel at this time and handling is not the same person, exit and do Reason, avoids causing unnecessary loss to user, further improves the safety of certificate acquiring.

Embodiment 10

The present embodiment has carried out further restriction to secondary identity authentication module, as shown in figure 11, satisfaction investigation mould Block 13 includes:

Mouth Expression Recognition module 131: the facial characteristics of several clients is acquired when for communicating with client, and identifies visitor The mouth expression at family, specifically: the number F2 of the number F1, smile that are in a pout, the number F3 for mouth of beeping, the number F4 that shows one's ivories；

Eye Expression Recognition module 132: the facial characteristics of several clients is acquired when for communicating with client, and identifies visitor The eye expression at family, specifically: the number G2 of the number G1, the supercilious look that stare；

Supercilium Expression Recognition module 133: the facial characteristics of several clients is acquired when for communicating with client, and identifies visitor The eye expression at family, specifically: the number H1 for choosing eyebrow, the number H2 to frown；

Headwork identification module 134: the facial characteristics of several clients is acquired when for communicating with client, and identifies visitor The eye expression at family, specifically: the number J1 to shake the head, the number J2 to nod；

Score computing module 135: for collecting mouth Expression Recognition module 131, eye Expression Recognition module 132, supercilium Expression that Expression Recognition module 133 and headwork identification module 134 recognize simultaneously calculates satisfaction Q, and Q is calculated by following methods It obtains；Q=-F1 × f1+F21 × f2-F3 × f3+F4 × f4-G1 × g1-G2 × g2-H1 × h1-H2 × h2-J1 × j1+J2 × j2, wherein f1+f2+f3+f4+g1+g2+h1+h2+j1+j2=100%.

The present embodiment further limits satisfaction investigation module, keeps calculated satisfaction more accurate.

Claims (10)

1. a kind of active interactive dialogue robot system, which is characterized in that handle module including pre-sales consultation module (1) and mid-sales (2)；

The pre-sales consultation module (1) include words art memory module (11), response means (12), satisfaction investigation module (13) and Line module (14)；

The words art memory module (11): for creating industry words art library, customization meets the dialogue automatic question answering corpus of business；

The response means (12): for acquiring the problem of line module (14) are sent, matching is corresponding in industry words art library The answer is simultaneously sent to line module (14) by answer；

The satisfaction investigation module (13): for expiring to the secondary consulting to user when user seeks advice from or after consulting Meaning degree is investigated；

The line module (14): for for user propose problem, answer a question and receive the response means (12) transmission Answer；

It includes authentication module (21), session module (22) and command execution module (23) that module (2) is handled in the mid-sales；

The authentication module (21): for verifying whether user is to handle in person；

The session module (22): it for receiving the instruction of the authentication module (21) transmission, is mentioned according to application scenarios to user It goes wrong, acquires the answer of user, the legitimacy that verification user answers, according to check results to the command execution module (23) Send instruction；

The command execution module (23): for executing corresponding business order or exiting；

The authentication module (21) includes speech verification module (211)；

The speech verification module (211): for judging whether it is and handling in person to user's transmission speech verification information, if It is to handle in person, then sends and instruct to session module (22), handled if not me, then is sent to command execution module (23) Instruction, command content are to exit.

2. active interactive dialogue robot system as described in claim 1, which is characterized in that the dialogue robot system is also Including Failure elimination module (3), the Failure elimination module (3) includes:

Failure monitoring module (31): the working condition of moment acquisition dialogue robot system, and judge the work for talking with robot Whether state is abnormal, if abnormal, send and instructs to fault processing module (32), if normally, be not processed；

Fault processing module (32): for calling relevant staff, staff will be at first time be to failure Reason.

3. active interactive dialogue robot system as described in claim 1, which is characterized in that the failure monitoring module (31) Including answering velocity estimated module (311), satisfaction investigation judgment module (312), speech verification judgment module (313), executing Judgment module (314)；

It is described to answer velocity estimated module (311): for judging that the line module (14) send time T1 and the dialogue of problem The time T2 that robot system is answered a question, judges the size of time difference threshold value Tn and time difference T2-T1, if T2-T1 >=Tn, It then sends and instructs to the fault processing module (32), if T2-T1 < Tn, to the satisfaction investigation judgment module (312) instruction is sent；

The satisfaction investigation judgment module (312): for judge user consulting after whether to user send satisfaction tune Short message or voice call are looked into, if sent, sends and instructs to the speech verification judgment module (313), if not sent, It sends and instructs to fault processing module (32)；

The speech verification judgment module (313): for whether judge before executing related service to user's execution speech verification, If executed, instruction is sent to judgment module (314) are executed, if be not carried out, sends and refers to fault processing module (32) It enables；

The execution judgment module (314): for judging to talk with robot whether for user handle phase after user select to buy Pass business instructs if it is, being not processed if it is not, then sending to fault processing module (32).

4. active interactive dialogue robot system as described in claim 1, which is characterized in that the authentication module (21) is also wrapped It includes image authentication module (212), described image authentication module (212) includes:

First facial feature obtains module (2121): for obtaining and identifying multiple facial characteristics of user, and will identify that Multiple facial characteristics are sent to first facial Characteristic Contrast module (2123)；

Identity information acquisition module (2122): it for scanning the identity document of user, identifies the certificate photograph on identity document, incites somebody to action The certificate photograph is sent to first facial Characteristic Contrast module (2123)；

First facial Characteristic Contrast module (2123): the face that module (2121) are sent is obtained for receiving the first facial feature The certificate photograph that portion's feature and the identity information acquisition module (2122) are sent, and by all facial characteristics and certificate photograph It is compared, calculates the two similarity Tq, and maximum similarity Tq is compared with similarity threshold Tp, as Tq >=Tp, then It is proved to be successful, sends and instruct to the session module (22), as Tq < Tp, then authentication failed, to the command execution module (23) instruction is sent, command content is to exit.

5. active interactive dialogue robot system according to claim 4, which is characterized in that the authentication module (21) is also Including double authentication module (213), the double authentication module (213) includes:

Double facial information identifies submodule (2131): for obtaining two groups of facial characteristics of two users, every group of face simultaneously Feature includes multiple facial characteristics of one of user, and two groups of facial characteristics is sent to double facial characteristics contrast module (2133)；

Double identity information acquisition module (2132): for scanning the identity document of two users and obtaining two identity documents Certificate photograph, two certificate photographs that will acquire are sent to double facial characteristics contrast module (2133)；

Double facial characteristics contrast module (2133): for receiving double facial information identification submodule (2131) transmission The certificate photograph that two groups of facial characteristics and the double identity information acquisition module (2132) are sent, respectively by two groups of facial characteristics It is compared with corresponding certificate photograph, both calculates similarity Tm, and by two maximum similarity Tm1 and Tm2 and similarity Threshold value T3 is compared, and as Tm1 >=T3 and Tm2 >=T3, is then proved to be successful, and is sent and is instructed to the session module (22), works as Tm1 < T3 or Tm2 < T3 then sends to the command execution module (23) and instructs, and command content is to exit.

6. active interactive dialogue robot system according to claim 1, which is characterized in that session module (22) packet It includes:

Problem memory module (221): the problem of for storing the lower enquirement in need of a certain scene, by enquirement in need ask Topic is arranged in order, when receiving the instruction of problem generation module (222) transmission, by all the problem of being arranged in order transmissions To problem generation module (222)；

Problem generation module (222): for receiving the instruction of the authentication module (2) transmission, and to described problem memory module (221) instruction is sent, the problem of arranged in sequence that will be received is sent in sequence to line module (14), and to answer acquisition module (223) instruction is sent；

Answer acquisition module (223): it is sent for acquiring line module (14) to the answer content of problem, and by the answer content To answer correction verification module (224)；

Answer correction verification module (224): the answer content sent for receiving the answer acquisition module (223) verifies the answer The legitimacy of content to command execution module (23) or repeats that module (225) is putd question to send instruction according to check results；

It repeats to put question to module (225): for the answer of empty user input, and repeating transmission verification to line module (14) and do not conform to The problem of method；

The answer correction verification module (224) includes:

Numeric class answer judgment module (2241): for user's answer content to be converted to text, and judge that the answer of user is It is no if it is, identifying digital number Xn, and Xn to be sent to digit length judgment module (2242) to be digital, if it is not, It then sends and instructs to text class answer judgment module (2243)；

Digit length judgment module (2242): the digital number sent for receiving the numeric class answer judgment module (2241) Xn judges that Xn and the size of digital number threshold X 1 send to described problem generation module (222) and instruct if Xn=X1, Next problem is sent to the line module (14), if Xn ≠ X1, sends instruction to enquirement module (225) is repeated；

Text class answer judgment module (2243): for judging whether the answer of user is text class answer, if so, identification text Word content in this, and word content is sent to same text judgment module (2244), if it is not, then judging to scanning Module (2245) sends instruction；

Same text judgment module (2244): for receiving in the text that the text class answer judgment module (2243) sends Hold, judges whether there is word identical with default answer in the word content, if so, then to described problem generation module (222) instruction is sent, sends next problem to the line module (14), if it is not, to repeating to put question to module (225) Send instruction；

Scanning judgment module (2245): for scanning the certificate of user's offer, and judging whether the certificate of the scanning is qualified, if Qualification then sends to described problem generation module (222) and instructs, proposes next problem, if unqualified, puts question to repetition Module (225) sends instruction.

7. active interactive dialogue robot system according to claim 6, which is characterized in that the scanning judgment module (2245) include:

Content matching module (22451): whether the word content on certificate for judging user's offer is consistent with problem, if It is consistent, sends and instruct to certificate matching module (22452), if be not consistent, sends instruction to enquirement module (225) is repeated；

Certificate matching module (22452): it for receiving the instruction of the content matching module (22451) transmission, identifies and judges Whether the content affixed one's seal on certificate is consistent with word content, if be consistent, sends and refers to described problem generation module (222) Enable, propose next problem, if be not consistent, determine it is unqualified, to repeat put question to module (225) send instruction.

8. active interactive dialogue robot system according to claim 6, which is characterized in that described problem generation module (222) include:

Number judgment module (2221): for after sending problem to line module (14), judging that certificate is handled in user's selection Number P instruction is sent to answer acquisition module (2231), if P > 1, to answer-mode module if P=1 (2222) instruction is sent；

Answer-mode module (2222): for judge multiple users answer a certain problem answer it is whether identical, if identical, Instruction is sent to an answer acquisition module (2231), is instructed if it is different, then being sent to multiple answer acquisition module (2232)；

The answer acquisition module (223) includes:

Answer acquisition module (2231): for disposably acquiring user to the answer content of problem, and by the answer content Answer correction verification module (224) are sent to, send next problem to the line module (14) after examining successfully；

Multiple answer acquisition module (2232): for multi collect user to the answer content of problem, times of collection and user's Number is identical, and all answer contents is sent to answer correction verification module (224) according to acquisition order, all answer contents school Next problem is sent to the line module (14) after testing successfully.

9. active interactive dialogue robot system according to claim 1, which is characterized in that the dialogue robot system It further include review module (4), the review module (4) includes: List Generating Module (41): for answering all ask in user After inscribing and verifying successfully, the answer that all problems and corresponding user are answered is generated into a list and is sent to line module (14)；

Review submodule (42): for receiving the instruction of line module (14) transmission and identifying the command content, if in instruction Hold for it is not necessary to modify then send to secondary identity authentication module (44) and instruct, if command content is to need to modify, select to problem It selects module (43) and sends instruction；

Problem selecting module (43): the instruction sent for receiving review submodule (42), and transfer the problem of need to modifying It is sent to line module (14)；

The secondary identity authentication module (44): for carrying out authentication again after user sends it is not necessary to modify instruction, and It is sent and is instructed to the command execution module (23) according to verification result；

The secondary identity authentication module (44) includes:

Second facial characteristics obtains module (441): for acquiring the face of user again after user sends it is not necessary to modify instruction Feature, and it is sent to the second facial characteristics contrast module (442)；

Second facial characteristics contrast module (442): the face of module (441) acquisition is obtained for second facial characteristics Feature and first facial feature obtain the facial characteristics that module (211) obtain, and judge whether to match, if it does, then to described Command execution module (23) sends instruction, and command content is to execute corresponding business, if it does not match, executing to the order Module (23) sends instruction, and command content is to exit.

10. active interactive dialogue robot system according to claim 1, which is characterized in that the satisfaction investigation mould Block (13) includes:

Mouth Expression Recognition module (131): the facial characteristics of several clients is acquired when for communicating with client, and identifies client Mouth expression, specifically: the number F2 of the number F1, smile that are in a pout, the number F3 for mouth of beeping, the number F4 that shows one's ivories；

Eye Expression Recognition module (132): the facial characteristics of several clients is acquired when for communicating with client, and identifies client Eye expression, specifically: the number G2 of the number G1, the supercilious look that stare；

Supercilium Expression Recognition module (133): the facial characteristics of several clients is acquired when for communicating with client, and identifies client Eye expression, specifically: the number H1 for choosing eyebrow, the number H2 to frown；

Headwork identification module (134): the facial characteristics of several clients is acquired when for communicating with client, and identifies client Eye expression, specifically: the number J1 to shake the head, the number J2 to nod；

It scores computing module (135): for collecting mouth Expression Recognition module (131), eye Expression Recognition module (132), eyebrow Expression that portion's Expression Recognition module (133) and headwork identification module (134) recognize simultaneously calculates satisfaction Q, and Q is by with lower section Method is calculated；Q=-F1 × f1+F21 × f2-F3 × f3+F4 × f4-G1 × g1-G2 × g2-H1 × h1-H2 × h2-J1 × J1+J2 × j2, wherein f1+f2+f3+f4+g1+g2+h1+h2+j1+j2=100%.