CN107784033A - A kind of dialogue-based method and apparatus recommended - Google Patents

Abstract

It is an object of the invention to provide a kind of dialogue-based method and apparatus for carrying out real-time recommendation, based on the current sessions between active user and its existing customer, according to the attribute information of the existing customer, matching acquisition is corresponding with the current problem in the current sessions from knowledge base recommends answer；The recommendation answer is supplied to the active user；Wherein, the structure of the knowledge in the knowledge base includes<Client properties, problem, answer>；Compared with prior art, the present invention efficiently utilizes existing knowledge, facilitates active user to answer the inquiry of existing customer, improves the usage experience of user.

Description

A kind of dialogue-based method and apparatus recommended

Technical field

The present invention relates to field of computer technology, more particularly to a kind of dialogue-based technology recommended.

Background technology

Department or company for needing a large amount of voice calls, it have accumulated substantial amounts of phonetic material, these phonetic materials In include many experience and knowledges.But due to lacking effective management, the office worker newly entered is difficult to benefit from it.One master Will the problem of be, by manually needing substantial amounts of manpower and materials to carry out voiced translation, to allow new office worker expertly to grasp extraction The knowledge gone out also not a duck soup.

Existing scheme mainly manually carries out Knowledge Extraction, goes to remember and applies by user oneself, however it is existing this Following drawback be present in kind mode：1) artificial extract is difficult to ensure that the comprehensive of knowledge, fully relies on the experience of individual, may hold Knowledge be have ignored, or bad Knowledge Extraction is come out, it is difficult to ensure optimal answer；2) memory of knowledge needs the time, increases Add the work load of user, differ surely to be in due course and remembered correct knowledge；3) knowledge can not be optimized, it is good Knowledge can not more be exposed, bad knowledge will not also reduce exposure frequency.

Therefore, how one of dialogue-based progress answer recommendation, the problem of turning into those skilled in the art's urgent need to resolve.

The content of the invention

It is an object of the invention to provide a kind of dialogue-based method and apparatus recommended.

According to an aspect of the invention, there is provided a kind of dialogue-based method recommended, wherein, this method includes Following steps：

Based on the current sessions between active user and its existing customer, according to the attribute information of the existing customer, from Matching obtains recommendation answer corresponding with the current problem in the current sessions in knowledge base；

The recommendation answer is supplied to the active user；

Wherein, the structure of the knowledge in the knowledge base includes<Client properties, problem, answer>.

Preferably, the knowledge base is established or updated using following processes：

Obtain the historical session record between each user and its client；

The historical session is recorded and carries out Knowledge Extraction, knowledge corresponding to acquisition, wherein, the structure of the knowledge includes <Client properties, problem, answer>；

According to the knowledge, establish or update the knowledge base.

Preferably, described recorded to the historical session carries out Knowledge Extraction, includes corresponding to acquisition the step of knowledge：

Historical session record is converted into corresponding session text message；

Knowledge Extraction is carried out to the session text message, obtains the knowledge, wherein, the structure of the knowledge includes< Client properties, problem, answer>.

It is highly preferred that described carry out Knowledge Extraction to the session text message, include corresponding to acquisition the step of knowledge：

The session text message is abstracted as the binary structure of knowledge, wherein, the binary structure of knowledge includes<User or Client, what is said or talked about>；

The binary structure of knowledge is reassembled as the ternary structure of knowledge, wherein, the ternary structure of knowledge includes<Client, Problem, answer>；

Each element in the ternary structure of knowledge is clustered respectively, obtains the knowledge, wherein, the knowledge Structure include<Client properties, problem, answer>.

Preferably, this method also includes：

According to the occurrence number of the recommendation answer and its corresponding current problem in the knowledge base, it is determined that described push away Recommend the priority of answer；

Wherein, the described the step of recommendation answer is supplied into the active user, includes：

According to the priority, the recommendation answer is supplied to the active user.

Preferably, this method also includes：

The actual answer of the current problem is directed in the current sessions according to the active user, is pushed away described in adjustment Recommend the priority of answer.

Preferably, this method also includes：

It is current described in real time modifying according to the active user and question and answer of the existing customer in the current sessions The attribute information of client.

Preferably, this method also includes：

The current sessions between the active user and the existing customer are obtained, are recorded as historical session, with Knowledge Extraction corresponding to progress.

According to another aspect of the present invention, a kind of dialogue-based recommendation apparatus recommended is additionally provided, wherein, should Recommendation apparatus includes：

For based on the current sessions between active user and its existing customer, being believed according to the attribute of the existing customer Breath, the matching acquisition device for recommending answer corresponding with the current problem in the current sessions from knowledge base；

For the recommendation answer to be supplied to the device of the active user；

Wherein, the structure of the knowledge in the knowledge base includes<Client properties, problem, answer>.

Preferably, the recommendation apparatus also includes：

For obtaining the device of the record of the historical session between each user and its client；

Knowledge Extraction is carried out for being recorded to the historical session, the device of knowledge corresponding to acquisition, wherein, the knowledge Structure include<Client properties, problem, answer>；

For according to the knowledge, establishing or updating the device of the knowledge base；

Preferably, it is described to be used to record the historical session progress Knowledge Extraction, the device bag of knowledge corresponding to acquisition Include：

For historical session record to be converted to the unit of corresponding session text message；

For carrying out Knowledge Extraction to the session text message, the unit of the knowledge is obtained, wherein, the knowledge Structure includes<Client properties, problem, answer>.

It is highly preferred that described be used to carry out Knowledge Extraction to the session text message, the unit for obtaining the knowledge is used In：

The session text message is abstracted as the binary structure of knowledge, wherein, the binary structure of knowledge includes<User or Client, what is said or talked about>；

The binary structure of knowledge is reassembled as the ternary structure of knowledge, wherein, the ternary structure of knowledge includes<Client, Problem, answer>；

Each element in the ternary structure of knowledge is clustered respectively, obtains the knowledge, wherein, the knowledge Structure include<Client properties, problem, answer>.

Preferably, the recommendation apparatus also includes：

For the occurrence number according to the recommendation answer and its corresponding current problem in the knowledge base, institute is determined State the device for the priority for recommending answer；

Wherein, it is described to be used to be supplied to the device of the active user to be used for the recommendation answer：

According to the priority, the recommendation answer is supplied to the active user.

Preferably, the recommendation apparatus also includes：

For being directed to the actual answer of the current problem in the current sessions according to the active user, institute is adjusted State the device for the priority for recommending answer.

Preferably, the recommendation apparatus also includes：

For according to the active user and question and answer of the existing customer in the current sessions, described in real time modifying The device of the attribute information of existing customer.

Preferably, the recommendation apparatus also includes：

For obtaining the current sessions between the active user and the existing customer, remember as historical session Record, with the device of Knowledge Extraction corresponding to progress.

According to a further aspect of the invention, a kind of computer equipment is additionally provided, the computer equipment includes：

One or more processors；

Memory, for storing one or more computer programs；

When one or more of computer programs are by one or more of computing devices so that it is one or Multiple processors realize method as described above.

Compared with prior art, the present invention is based on the current sessions between active user and its existing customer, according to described The attribute information of existing customer, matching acquisition is corresponding with the current problem in the current sessions from knowledge base recommends to answer Case, the recommendation answer is supplied to the active user, wherein, the structure of the knowledge in the knowledge base includes<Client belongs to Property, problem, answer>, existing knowledge is efficiently utilized, facilitates active user to answer the inquiry of existing customer, improves use The usage experience at family.

Further, the present invention records according to the historical session between each user and its client, therefrom extracts knowledge, will The knowledge with<Client properties, problem, answer>Mode stored, so, when it is follow-up there is user to be conversed with client when, According to the attribute information of the client, therefrom matching recommends answer to be answered to user, on the one hand realizes the management to knowledge, On the other hand based on context scene carries out knowledge recommendation in real time, improves system effectiveness, improves the usage experience of user.

Further, the present invention can also go out occurrence according to recommendation answer and its corresponding current problem in knowledge base Number, its priority is determined, and according to priority answer will be recommended to be supplied to the active user, further improve the use of user Experience.Further, whether the present invention can also be according to the actual answer of active user with recommending answer consistent, to adjust this Recommend the priority of answer, be effectively utilized the actual feedback of user, further improve the usage experience of user.

Further, the present invention is according to the question and answer of existing customer, its attribute information of real time modifying so that follow-up matching behaviour It is more accurate to make, and further improves the usage experience of user.

Brief description of the drawings

By reading the detailed description made to non-limiting example made with reference to the following drawings, of the invention is other Feature, objects and advantages will become more apparent upon：

Fig. 1 shows the structural representation for the dialogue-based device recommended according to one aspect of the invention；

Fig. 2 shows the structural representation according to an embodiment of the invention for the dialogue-based device recommended；

Fig. 3 shows the schematic flow sheet for the dialogue-based method recommended according to a further aspect of the present invention；

Fig. 4 shows that the flow in accordance with another embodiment of the present invention for the dialogue-based method recommended is illustrated Figure.

Same or analogous reference represents same or analogous part in accompanying drawing.

Embodiment

It should be mentioned that some exemplary embodiments are described as before exemplary embodiment is discussed in greater detail The processing described as flow chart or method.Although operations are described as the processing of order by flow chart, therein to be permitted Multioperation can be implemented concurrently, concomitantly or simultaneously.In addition, the order of operations can be rearranged.When it The processing can be terminated when operation is completed, it is also possible to the additional step being not included in accompanying drawing.The processing It can correspond to method, function, code, subroutine, subprogram etc..

Alleged within a context " computer equipment ", also referred to as " computer ", referring to can be by running preset program or referring to Order performs the intelligent electronic device of the predetermined process process such as numerical computations and/or logical calculated, its can include processor with Memory, the survival that is prestored in memory by computing device are instructed to perform predetermined process process, or by ASIC, The hardware such as FPGA, DSP perform predetermined process process, or are realized by said two devices combination.Computer equipment includes but unlimited In server, PC, notebook computer, tablet personal computer etc..

The computer equipment includes user equipment and the network equipment.Wherein, the user equipment includes but is not limited to individual People's computer, notebook computer, tablet personal computer etc.；The network equipment includes but is not limited to single network server, multiple networks Server group into server group or based on cloud computing (Cloud Computing) by a large amount of computers or webserver structure Into cloud, wherein, cloud computing is one kind of Distributed Calculation, a super void being made up of the computer collection of a group loose couplings Intend computer.Wherein, the computer equipment can isolated operation realize the present invention, also can access network and by with network The interactive operations of other computer equipments realize the present invention.Wherein, the network residing for the computer equipment is included but not It is limited to internet, wide area network, Metropolitan Area Network (MAN), LAN, VPN etc..

It should be noted that the user equipment, the network equipment and network etc. are only for example, other are existing or from now on may be used The computer equipment or network that can occur such as are applicable to the present invention, should also be included within the scope of the present invention, and to draw It is incorporated herein with mode.

Method (some of them are illustrated by flow) discussed hereafter can be by hardware, software, firmware, centre Part, microcode, hardware description language or its any combination are implemented.Implement when with software, firmware, middleware or microcode When, to implement the program code of necessary task or code segment can be stored in machine or computer-readable medium and (for example deposit Storage media) in.(one or more) processor can implement necessary task.

Concrete structure and function detail disclosed herein are only representational, and are for describing showing for the present invention The purpose of example property embodiment.But the present invention can be implemented by many alternative forms, and it is not interpreted as It is limited only by the embodiments set forth herein.

Although it should be appreciated that may have been used term " first ", " second " etc. herein to describe unit, But these units should not be limited by these terms.It is used for the purpose of using these terms by a unit and another unit Make a distinction.For example, in the case of the scope without departing substantially from exemplary embodiment, it is single that first module can be referred to as second Member, and similarly second unit can be referred to as first module.Term "and/or" used herein above include one of them or Any and all combination of more listed associated items.

It should be appreciated that when a unit is referred to as " connecting " or during " coupled " to another unit, it can directly connect Connect or be coupled to another unit, or there may be temporary location.On the other hand, when a unit is referred to as " directly connecting Connect " or " direct-coupling " when arriving another unit, then in the absence of temporary location.It should in a comparable manner explain and be used to retouch State the relation between unit other words (such as " between being in ... " compared to " between being directly in ... ", " and with ... it is adjacent Closely " compared to " with ... be directly adjacent to " etc.).

Term used herein above is not intended to limit exemplary embodiment just for the sake of description specific embodiment.Unless Context clearly refers else, otherwise singulative used herein above "one", " one " also attempt to include plural number.Should also When understanding, term " comprising " and/or "comprising" used herein above provide stated feature, integer, step, operation, The presence of unit and/or component, and do not preclude the presence or addition of other one or more features, integer, step, operation, unit, Component and/or its combination.

It should further be mentioned that in some replaces realization modes, the function/action being previously mentioned can be according to different from attached The order indicated in figure occurs.For example, depending on involved function/action, the two width figures shown in succession actually may be used Substantially simultaneously to perform or can perform in a reverse order sometimes.

The present invention is described in further detail below in conjunction with the accompanying drawings.

Fig. 1 shows the structural representation for the dialogue-based device recommended according to one aspect of the invention.

For example in computer equipment, the recommendation apparatus 1 is included for current with it based on active user recommendation apparatus 1 Current sessions between client, according to the attribute information of the existing customer, matching acquisition and the current meeting from knowledge base Recommend the device of answer, hereinafter referred to as coalignment 101 corresponding to current problem in words；And for by the recommendation answer The device of the active user is supplied to, device 102 is hereinafter referred to as provided, wherein, the structure bag of the knowledge in the knowledge base Include<Client properties, problem, answer>.

Wherein, coalignment 101 is based on the current sessions between active user and its existing customer, according to the current visitor The attribute information at family, matching acquisition is corresponding with the current problem in the current sessions from the knowledge base recommends answer, Wherein, the structure of the knowledge in the knowledge base includes<Client properties, problem, answer>.

Specifically, all kinds of knowledge are stored with knowledge base, for example, these knowledge are with structure<Client properties, problem, answer Case>It is stored in the knowledge base, these knowledge are all according to the session experience between a large number of users and its client, are therefrom extracted Come.For example, for the client of a certain attribute, when it inquires some problem, what given answer is, by these Problem or the statistics of answer, arrange, and be stored in the knowledge base, can facilitate how other users answer client's to refer to Problem.The knowledge base can store all kinds of knowledge with speech form, can also store all kinds of knowledge, the knowledge base with written form It can be located in the recommendation apparatus 1, can also be in the third party device being connected with the recommendation apparatus 1 by network.

When active user and its existing customer carry out current sessions, for example, it carries out voice call, coalignment 101 according to the attribute information of the existing customer, from the knowledge base matching obtain and currently asking in the current sessions Recommend answer corresponding to topic, for example, coalignment 101 is first according to the attribute information of the existing customer, such as the row of existing customer The attribute informations such as industry, region, scale, matching obtains the client properties corresponding with the attribute information in knowledge base and its institute is right All knowledge answered, the current problem asked further according to the existing customer in current sessions, match and obtain in these knowledge It is corresponding with the problem to recommend answer.

Here, the attribute information of the existing customer is, for example, known to the active user, for example, user is carried out in this prior It is just known before the current sessions, or or by the real-time attribute analyzed after drawing or being adjusted of the recommendation apparatus 1 Information.

Here, the coalignment 101 can also be changed the current problem in the current sessions first by speech recognition For text information, further according to the text information, matched and searched is carried out in knowledge base.

Those skilled in the art will be understood that above-mentioned matching recommends the mode of answer to be only for example, and other are existing or modern The matching being likely to occur afterwards recommends the mode of answer to be such as applicable to the present invention, should also be included within the scope of the present invention, And it is incorporated herein by reference.

Device 102 is provided the recommendation answer is supplied to the active user.

Specifically, obtained recommendation answer is matched for coalignment 101, there is provided device 102 for example passes through voice, text The presentation mode of word or other agreements, the recommendation answer is supplied to the active user, selected for the active user. For example, when the active user carries out current sessions using Wearable with the existing customer, there is provided device 102 and the wearing Formula equipment is interacted, and the recommendation answer that the coalignment 101 matches to obtain is supplied to via the Wearable and deserved Preceding user.Preferably, for the current problem in current sessions, if coalignment 101 matches multiple recommendation answers, provide The plurality of recommendation answer can be supplied to the user by device 102 randomly or according to certain order.

Those skilled in the art will be understood that above-mentioned offer recommends the mode of answer to be only for example, and other are existing or modern The offer being likely to occur afterwards recommends the mode of answer to be such as applicable to the present invention, should also be included within the scope of the present invention, And it is incorporated herein by reference.

Here, recommendation apparatus 1 is based on the current sessions between active user and its existing customer, according to the existing customer Attribute information, from knowledge base matching obtain it is corresponding with the current problem in the current sessions recommends answer, described in general Answer is recommended to be supplied to the active user, wherein, the structure of the knowledge in the knowledge base includes<Client properties, problem, answer Case>, existing knowledge is efficiently utilized, facilitates active user to answer the inquiry of existing customer, improve user uses body Test.

Fig. 2 shows the structural representation according to an embodiment of the invention for the dialogue-based device recommended.

Recommendation apparatus 1 also includes being used to obtain the device of the historical session record between each user and its client, below Referred to as the first acquisition device 203；Knowledge Extraction is carried out for being recorded to the historical session, the device of knowledge corresponding to acquisition, Wherein, the structure of the knowledge includes<Client properties, problem, answer>, hereinafter referred to as Knowledge Extraction device 204；And it is used for According to the knowledge, the device of the knowledge base, hereinafter referred to as updating device 205 are established or updated.Wherein, coalignment 201, It is identical or essentially identical with corresponding intrument shown in Fig. 1 to provide device 202, therefore here is omitted, and wrap by reference It is contained in this.

Wherein, the first acquisition device 203 obtains the historical session record between each user and its client.

Specifically, when user is exchanged with client, the content record that can be exchanged gets off, and is used for example, working as When family is made a phone call with client, recorded by voice call device used in the user, by its voice call content Record, first acquisition device 203 either periodically or in real-time obtains each for example by being interacted with each voice call device Multiple historical sessions record between individual user and its client.

Those skilled in the art will be understood that the mode of above-mentioned acquisition historical session record is only for example, and other are existing Or the mode for the acquisition historical session record being likely to occur from now on is such as applicable to the present invention, present invention protection model should be also included in Within enclosing, and it is incorporated herein by reference.

Knowledge Extraction device 204 records to the historical session carries out Knowledge Extraction, knowledge corresponding to acquisition, wherein, institute Stating the structure of knowledge includes<Client properties, problem, answer>.

Specifically, recorded for the historical session acquired in the first acquisition device 203, Knowledge Extraction device 204 is gone through to this History conversation recording carries out Knowledge Extraction, for example, the historical session record of the speech form is converted into going through for written form first History conversation recording, then knowledge is extracted from the historical session record of the written form, the structure of the knowledge includes<Client belongs to Property, problem, answer>, for example, the client properties according to corresponding to different clients, the problem of client is corresponded to or answer are entered The problem of going and classify, then belonging in a session and answer are so as to obtain structure as a knowledge<Client properties, Problem, answer>Knowledge.

Preferably, the Knowledge Extraction device 204 includes converting unit (not shown) and extracting unit (not shown).This turn Change unit and historical session record is converted into corresponding session text message；Extracting unit enters to the session text message Row Knowledge Extraction, the knowledge is obtained, wherein, the structure of the knowledge includes<Client properties, problem, answer>.

Specifically, recorded for the historical session of the speech form acquired in the first acquisition device 203, converting unit is for example By technologies such as speech recognition, text conversions, the historical session record of the speech form is converted into corresponding session text envelope Breath；Further, converting unit can also be located in advance to the historical session record acquired in first acquisition device 203 first Reason, such as judge the family of languages of historical session record, determine the front and rear nasal sound of historical session record, to obtain pretreatment letter Breath；Recorded further according to the historical session, or further combined with the user related information and the pretreatment information, it is determined that corresponding meeting Talk about text message.

Then, the session text message that extracting unit is converted to the converting unit carries out Knowledge Extraction, obtains institute Knowledge is stated, for example, the client properties according to corresponding to different clients, the problem of client is corresponded to or answer are classified, The problem of belonging to again in a session and answer are so as to obtain structure as a knowledge<Client properties, problem, answer Case>Knowledge.

It is highly preferred that the session text message is abstracted as the binary structure of knowledge by the extracting unit, wherein, described two Meta-knoeledge structure includes<User or client, what is said or talked about>；The binary structure of knowledge is reassembled as the ternary structure of knowledge, its In, the ternary structure of knowledge includes<Client, problem, answer>；Each element in the ternary structure of knowledge is entered respectively Row cluster, obtains the knowledge, wherein, the structure of the knowledge includes<Client properties, problem, answer>.

Specifically, after historical session record is converted into session text message by converting unit, extracting unit is for example logical The mode sorted out is crossed, the session text message is abstracted as the binary structure of knowledge, wherein, the binary structure of knowledge includes<With Family or client, what is said or talked about>, here, it can be problem or answer that what is said or talked about, that is, extracting unit is literary by session This information categorization is<User, problem>、<User, answer>、<Client, problem>、<Client, answer>；Then, the extracting unit is again The binary structure of knowledge is reassembled as the ternary structure of knowledge, wherein, the ternary structure of knowledge includes<Client, problem, answer>, For example, will be mutually corresponding in the same session for same client the problem of and answer are as a ternary structure of knowledge； Then, extracting unit clusters respectively to each element in the ternary structure of knowledge, obtains the knowledge, wherein, institute Stating the structure of knowledge includes<Client properties, problem, answer>, for example, extracting unit is by visitors such as the industry of client, region, scales Family attribute abstraction comes out, and based on these client properties, each element in the ternary structure of knowledge is clustered respectively, obtained Corresponding cluster result, if three kinds of elements in the cluster result, i.e. client, problem, answer, appear in same session, then Using client properties, problem and corresponding answer corresponding to the client as a knowledge, it is so as to obtain the structure of knowledge<Client Attribute, problem, answer>Knowledge.

Those skilled in the art will be understood that the mode of above-mentioned extraction knowledge is only for example, and other are existing or from now on may be used The mode for the extraction knowledge that can occur such as is applicable to the present invention, should also be included within the scope of the present invention, and to quote Mode is incorporated herein.Those skilled in the art should also be understood that above-mentioned client properties are only for example, and other are existing or from now on may be used The client properties that can occur such as are applicable to the present invention, should also be included within the scope of the present invention, and wrap by reference It is contained in this.

Updating device 205 is according to the knowledge, knowledge base corresponding to foundation or renewal.

Specifically, the knowledge that updating device 205 is extracted the Knowledge Extraction device 204, knowledge corresponding to deposit In storehouse, to realize foundation or renewal to the knowledge base, the knowledge is with structure<Client properties, problem, answer>This is stored in know Know in storehouse.The knowledge base can be common database, it is preferable that can be lucene text index storehouse.

Here, the knowledge base can be located in the recommendation apparatus 1, it can also be located at and pass through network phase with the recommendation apparatus 1 In the third party device of connection, updating device 205 is interacted by network with the knowledge base, when there is new knowledge to be extracted, It is stored in by network in the knowledge base.

Here, recommendation apparatus 1 records according to the historical session between each user and its client, knowledge is therefrom extracted, will The knowledge with<Client properties, problem, answer>Mode stored, so, when it is follow-up there is user to be conversed with client when, According to the attribute information of the client, therefrom matching recommends answer to be answered to user, on the one hand realizes the management to knowledge, On the other hand based on context scene carries out knowledge recommendation in real time, improves system effectiveness, improves the usage experience of user.

Preferably (referring to Fig. 1), the recommendation apparatus 1 also includes being used for according to the recommendation answer and its corresponding currently asking The occurrence number in the knowledge base is inscribed, determines the device of the priority of the recommendation answer, hereinafter referred to as priority device (not shown)；Wherein, the recommendation answer is supplied to the active user by the offer device 102 according to the priority.

Specifically, answer is recommended to have certain priority, for example, each problem or answer are equal in knowledge base It is likely to occur more than once, priority device matches obtained recommendation answer and its corresponding current according to coalignment 101 Problem, counts the recommendation answer and its number that corresponding current problem occurs in knowledge base, determines the excellent of the recommendation answer First level, for example, occurrence number is more, then priority is higher, and occurrence number is fewer, then priority is lower, here, the priority fills The number that can occur respectively according to the number or current problem of recommending answer appearance is put, can also to determine the priority Both are considered to determine the priority；Then, there is provided device 102 will recommend answer to provide according to the priority of the determination The active user is given, for example, the arrangement according to priority from high to low, the active user is supplied to by the recommendation answer.

Preferably, storage problem or answer can go out simultaneously in whole knowledge base when stored knowledge in the knowledge base Existing number, for example, by knowledge with structure<Client properties, problem, problem occurrence number, answer, answer occurrence number>Storage In the knowledge base.

Here, according to recommending occurrence number in knowledge base of answer and its corresponding current problem, recommendation apparatus 1 can be with Its priority is determined, and according to priority answer will be recommended to be supplied to the active user, further improve user uses body Test.

More preferably (referring to Fig. 1), the recommendation apparatus also includes being used for according to the active user in the current sessions In be directed to the actual answer of the current problem, adjust the device of the priority for recommending answer, hereinafter referred to as adjusting apparatus (not shown).

Specifically, there is provided device 102 is used after coalignment 101 is matched obtained recommendation answer be supplied to user Family can select according to or refer to the recommendation answer to answer the current problem of existing customer, can also select not push away according to this Answer is recommended to answer, therefore, adjusting apparatus can obtain the reality that the active user is directed to the current problem in current sessions Answer, for example, adjusting apparatus by with active user used in voice call device interact, obtain the current use in real time Family is directed to the actual answer of the current problem in current sessions；Then, the adjusting apparatus adjusts this and pushed away according to the actual answer Recommend the priority of answer, for example, adjusting apparatus by the actual answer got compared with recommending answer, if unanimously, table Show that the active user employs the recommendation answer, adjusting apparatus can improve the priority of the recommendation answer, if inconsistent, table Show that the active user does not use the recommendation answer, adjusting apparatus can reduce the priority of the recommendation answer.

Preferably, adjusting apparatus can pass through the statistics to a large number of users for the utilization rate of recommendation answer, then row adjustment The priority of the recommendation answer, for example, a utilization rate threshold value can be set, when the utilization rate of the recommendation answer is less than the use During rate threshold value, the priority of the recommendation answer is reduced.Here, the utilization rate threshold value is, for example, system intialization, can also basis Actual conditions are adjusted by user.

Here, actual answer or the recommendation answer can be written form or speech form, if the reality It is all speech form to answer and recommend answer, then adjusting apparatus can first pass through the modes such as speech recognition, by the actual answer Text information is converted to recommendation answer, then the text information is compared.

Here, it is not to represent the rwo " completely the same, tally in every detail " that " actual to answer " is " consistent " with " recommendation answer ", and It is as long as that similarity reaches predetermined threshold, you can think that the actual answer is consistent with recommending answer, for example, for having been converted into The actual answer of text information and recommend answer, if the similarity of the rwo reaches 90%, it is considered that the actual answer and Recommend answer consistent.Here, the predetermined threshold is, for example, system intialization, can also be adjusted according to actual conditions by user It is whole.

Here, whether recommendation apparatus 1 can also be according to the actual answer of active user with recommending answer consistent, to adjust this Recommend the priority of answer, be effectively utilized the actual feedback of user, further improve the usage experience of user.

Preferably (referring to Fig. 1), the recommendation apparatus 1 also includes being used to be existed according to the active user and the existing customer Question and answer in the current sessions, the device of the attribute information of existing customer described in real time modifying, hereinafter referred to as change device (not Show).

Specifically, active user and existing customer can have contact question and answer several times, existing customer in current sessions It can be quizmaster or answerer, in this prior among the enquirement or answer of client, the current visitor may be implied The attribute information at family, for example, implying the attribute informations such as the industry of the existing customer, region, scale, these existing customers are being worked as The attribute information for revealing out in preceding session may have with attribute information known to prestore in system or active user Enter, therefore, modification device can obtain the active user and question and answer of the existing customer in current sessions, for example, by with language The interaction of sound communicator, the question and answer are obtained, and the attribute information of the existing customer is extracted from the question and answer, if the attribute information The attribute information of the existing customer from being prestored in system is different, then changes the attribute letter that device changes the existing customer in real time Breath.

Based on the amended attribute information, coalignment 101 can be obtained and deserved for existing customer matching again Recommend answer corresponding to current problem in preceding session, or, coalignment 101 can be in the next round for the existing customer During the matching of problem, matching is corresponding with the amended attribute information to recommend answer；Or if the existing customer is over The current sessions, then recommend when the amended attribute information can be used for client's session next time in this prior corresponding to matching Answer.

Here, question and answer of the recommendation apparatus 1 according to existing customer, its attribute information of real time modifying so that follow-up matching behaviour It is more accurate to make, and further improves the usage experience of user.

Preferably, the recommendation apparatus 1 also includes described between the active user and the existing customer for obtaining Current sessions, recorded as historical session, with the device of Knowledge Extraction corresponding to progress, hereinafter referred to as the second acquisition device (not Show).

Specifically, for the current sessions between active user and existing customer, it can also be used as historical session to record, For carrying out Knowledge Extraction, the second acquisition device for example by being interacted with voice call device, obtains the current meeting in real time Words, for example, whenever the active user or existing customer are finished in short, the second acquisition device just comes the word acquisition, or Person, after the active user and existing customer all complete question and answer all in the current sessions, the second acquisition device just will All voices in the current sessions are obtained, and are recorded these voices as historical session, so as to know corresponding to carrying out Know and extract, for example, transferring to Knowledge Extraction device to carry out Knowledge Extraction.

Fig. 3 shows the schematic flow sheet for the dialogue-based method recommended according to a further aspect of the present invention.

In step S301, recommendation apparatus 1 is based on the current sessions between active user and its existing customer, according to described The attribute information of existing customer, matching acquisition is corresponding with the current problem in the current sessions from the knowledge base recommends Answer, wherein, the structure of the knowledge in the knowledge base includes<Client properties, problem, answer>.

Specifically, all kinds of knowledge are stored with knowledge base, for example, these knowledge are with structure<Client properties, problem, answer Case>It is stored in the knowledge base, these knowledge are all according to the session experience between a large number of users and its client, are therefrom extracted Come.For example, for the client of a certain attribute, when it inquires some problem, what given answer is, by these Problem or the statistics of answer, arrange, and be stored in the knowledge base, can facilitate how other users answer client's to refer to Problem.The knowledge base can store all kinds of knowledge with speech form, can also store all kinds of knowledge, the knowledge base with written form It can be located in the recommendation apparatus 1, can also be in the third party device being connected with the recommendation apparatus 1 by network.

When active user and its existing customer carry out current sessions, for example, it carries out voice call, in step In S301, recommendation apparatus 1 is according to the attribute information of the existing customer, matching acquisition and the current meeting from the knowledge base Recommend answer corresponding to current problem in words, for example, in step S301, recommendation apparatus 1 is first according to the existing customer The industry of attribute information, such as existing customer, region, scale attribute information, match and obtain and the attribute information in knowledge base Corresponding client properties and its corresponding all knowledge, currently asked further according to what the existing customer was asked in current sessions Topic, in these knowledge matching obtain corresponding with the problem recommending answer.

Here, the attribute information of the existing customer is, for example, known to the active user, for example, user is carried out in this prior It is just known before the current sessions, or or by the real-time attribute analyzed after drawing or being adjusted of the recommendation apparatus 1 Information.

Here, in step S301, recommendation apparatus 1 can also be first by speech recognition, by working as in the current sessions Preceding problem is converted to text information, and further according to the text information, matched and searched is carried out in knowledge base.

Those skilled in the art will be understood that above-mentioned matching recommends the mode of answer to be only for example, and other are existing or modern The matching being likely to occur afterwards recommends the mode of answer to be such as applicable to the present invention, should also be included within the scope of the present invention, And it is incorporated herein by reference.

In step s 302, the recommendation answer is supplied to the active user by recommendation apparatus 1.

Specifically, in step S301, recommendation apparatus 1 matches obtained recommendation answer, in step s 302, pushes away Device 1 is recommended for example by the presentation mode of voice, word or other agreements, the recommendation answer is supplied to the current use Family, selected for the active user.For example, when the active user with the existing customer currently can using Wearable During words, in step s 302, recommendation apparatus 1 is interacted with the Wearable, and this is matched to obtain in step S301 Recommendation answer be supplied to the active user via the Wearable.Preferably, for the current problem in current sessions, if In step S301, recommendation apparatus 1 matches multiple recommendation answers, then in step s 302, recommendation apparatus 1 can randomly or According to certain order, the plurality of recommendation answer is supplied to the user.

Those skilled in the art will be understood that above-mentioned offer recommends the mode of answer to be only for example, and other are existing or modern The offer being likely to occur afterwards recommends the mode of answer to be such as applicable to the present invention, should also be included within the scope of the present invention, And it is incorporated herein by reference.

Here, recommendation apparatus 1 is based on the current sessions between active user and its existing customer, according to the existing customer Attribute information, from knowledge base matching obtain it is corresponding with the current problem in the current sessions recommends answer, described in general Answer is recommended to be supplied to the active user, wherein, the structure of the knowledge in the knowledge base includes<Client properties, problem, answer Case>, existing knowledge is efficiently utilized, facilitates active user to answer the inquiry of existing customer, improve user uses body Test.

Fig. 4 shows that the flow in accordance with another embodiment of the present invention for the dialogue-based method recommended is illustrated Figure.

This method also includes step S403：Obtain the historical session record between each user and its client；Step S404： The historical session is recorded and carries out Knowledge Extraction, knowledge corresponding to acquisition, wherein, the structure of the knowledge includes<Client belongs to Property, problem, answer>；And step S405：According to the knowledge, establish or update the knowledge base.Wherein, step S401, S402 is identical with corresponding to step shown in Fig. 3 or essentially identical, therefore here is omitted, and is incorporated herein by reference.

In step S403, recommendation apparatus 1 obtains the historical session record between each user and its client.

Specifically, when user is exchanged with client, the content record that can be exchanged gets off, and is used for example, working as When family is made a phone call with client, recorded by voice call device used in the user, by its voice call content Record, in step S403, recommendation apparatus 1 with each voice call device for example by interacting, regularly or in real time Multiple historical sessions that ground is obtained between each user and its client record.

Those skilled in the art will be understood that the mode of above-mentioned acquisition historical session record is only for example, and other are existing Or the mode for the acquisition historical session record being likely to occur from now on is such as applicable to the present invention, present invention protection model should be also included in Within enclosing, and it is incorporated herein by reference.

In step s 404, recommendation apparatus 1 records to the historical session carries out Knowledge Extraction, knowledge corresponding to acquisition, Wherein, the structure of the knowledge includes<Client properties, problem, answer>.

Specifically, recorded for acquired historical session in step S403, in step s 404, recommendation apparatus 1 is right Historical session record carries out Knowledge Extraction, for example, the historical session record of the speech form is converted into written form first Historical session record, then extract knowledge from the historical session of written form record, the structure of the knowledge includes<Client Attribute, problem, answer>, for example, the client properties according to corresponding to different clients, the problem of client is corresponded to or answer The problem of being classified, then being belonged in a session and answer are so as to obtain structure as a knowledge<Client belongs to Property, problem, answer>Knowledge.

Preferably, the step S404 includes sub-step S404a (not shown) and sub-step S404b (not shown).In son In step S404a, historical session record is converted to corresponding session text message by recommendation apparatus 1；In sub-step S404b In, recommendation apparatus 1 carries out Knowledge Extraction to the session text message, obtains the knowledge, wherein, the structure bag of the knowledge Include<Client properties, problem, answer>.

Specifically, recorded for the historical session in the speech form acquired in step S403, in sub-step S404a, The technology such as by speech recognition, text conversion of recommendation apparatus 1, the historical session record of the speech form is converted to correspondingly Session text message；Further, in sub-step S404a, recommendation apparatus 1 can also be first to being obtained in step S403 Take historical session record pre-processed, such as judge the historical session record the family of languages, determine the historical session record before Nasal sound etc. afterwards, to obtain pretreatment information；Recorded further according to the historical session, or further combined with the user related information with being somebody's turn to do Information is pre-processed, it is determined that corresponding session text message.

Then, in sub-step S404b, recommendation apparatus 1 is to the session text envelope that is converted in sub-step S404a Breath carries out Knowledge Extraction, obtains the knowledge, for example, the client properties according to corresponding to different clients, corresponding by the client The problem of or answer classified, then the problem of will belong in a session and answer is as a knowledge, so as to be tied Structure is<Client properties, problem, answer>Knowledge.

It is highly preferred that in sub-step S404b, the session text message is abstracted as two meta-knoeledge knots by recommendation apparatus 1 Structure, wherein, the binary structure of knowledge includes<User or client, what is said or talked about>；The binary structure of knowledge is reassembled as three Meta-knoeledge structure, wherein, the ternary structure of knowledge includes<Client, problem, answer>；To every in the ternary structure of knowledge One element is clustered respectively, obtains the knowledge, wherein, the structure of the knowledge includes<Client properties, problem, answer>.

Specifically, in sub-step S404a, after historical session record is converted into session text message by recommendation apparatus 1, In sub-step S404b, recommendation apparatus 1 is abstracted as two meta-knoeledge knots for example by way of classification, by the session text message Structure, wherein, the binary structure of knowledge includes<User or client, what is said or talked about>, here, it can be problem that what is said or talked about, Can be answer, that is, in sub-step S404b, session text message is classified as by recommendation apparatus 1<User, problem>、<With Family, answer>、<Client, problem>、<Client, answer>；Then, the binary structure of knowledge is reassembled as ternary again and known by recommendation apparatus 1 Know structure, wherein, the ternary structure of knowledge includes<Client, problem, answer>, for example, by for the same of same client The problem of mutually corresponding in individual session and answer are as a ternary structure of knowledge；Then, recommendation apparatus 1 is to three meta-knoeledge Each element in structure is clustered respectively, obtains the knowledge, wherein, the structure of the knowledge includes<Client properties, ask Topic, answer>, for example, in sub-step S404b, recommendation apparatus 1 takes out the client properties such as the industry of client, region, scale Come, based on these client properties, each element in the ternary structure of knowledge is clustered respectively, obtain corresponding cluster knot Fruit, if three kinds of elements in the cluster result, i.e. client, problem, answer, appear in same session, then it is the client is corresponding Client properties, problem and corresponding answer as a knowledge, be so as to obtain the structure of knowledge<Client properties, problem, answer Case>Knowledge.

Those skilled in the art will be understood that the mode of above-mentioned extraction knowledge is only for example, and other are existing or from now on may be used The mode for the extraction knowledge that can occur such as is applicable to the present invention, should also be included within the scope of the present invention, and to quote Mode is incorporated herein.Those skilled in the art should also be understood that above-mentioned client properties are only for example, and other are existing or from now on may be used The client properties that can occur such as are applicable to the present invention, should also be included within the scope of the present invention, and wrap by reference It is contained in this.

In step S405, recommendation apparatus 1 is according to the knowledge, knowledge base corresponding to foundation or renewal.

Specifically, in step S405, knowledge that recommendation apparatus 1 will be extracted in step s 404, deposit is corresponding Knowledge base in, to realize foundation or renewal to the knowledge base, the knowledge is with structure<Client properties, problem, answer>Storage In the knowledge base.The knowledge base can be common database, it is preferable that can be lucene text index storehouse.

Here, the knowledge base can be located in the recommendation apparatus 1, it can also be located at and pass through network phase with the recommendation apparatus 1 In the third party device of connection, in step S405, recommendation apparatus 1 is interacted by network with the knowledge base, when there is new knowledge When being extracted, it is stored in by network in the knowledge base.

Here, recommendation apparatus 1 records according to the historical session between each user and its client, knowledge is therefrom extracted, will The knowledge with<Client properties, problem, answer>Mode stored, so, when it is follow-up there is user to be conversed with client when, According to the attribute information of the client, therefrom matching recommends answer to be answered to user, on the one hand realizes the management to knowledge, On the other hand based on context scene carries out knowledge recommendation in real time, improves system effectiveness, improves the usage experience of user.

Preferably (referring to Fig. 3), this method also includes step S306 (not shown)：According to the recommendation answer and its correspondingly Occurrence number of the current problem in the knowledge base, determine the priority of the recommendation answer；Wherein, in step S302 In, the recommendation answer is supplied to the active user by recommendation apparatus 1 according to the priority.

Specifically, answer is recommended to have certain priority, for example, each problem or answer are equal in knowledge base Be likely to occur more than once, in step S306, recommendation apparatus 1 according to the recommendation answer for matching to obtain in step S301 and Its corresponding current problem, counts the recommendation answer and its number that corresponding current problem occurs in knowledge base, it is determined that should Recommend the priority of answer, for example, occurrence number is more, then priority is higher, and occurrence number is fewer, then priority is lower, This, in step S306, recommendation apparatus 1 can occur secondary according to the number or current problem of recommending answer appearance respectively Number, to determine the priority, can also consider both to determine the priority；Then, in step s 302, recommendation apparatus 1 according to the priority of the determination, and answer will be recommended to be supplied to the active user, for example, the arrangement according to priority from high to low, The recommendation answer is supplied to the active user.

Preferably, storage problem or answer can go out simultaneously in whole knowledge base when stored knowledge in the knowledge base Existing number, for example, by knowledge with structure<Client properties, problem, problem occurrence number, answer, answer occurrence number>Storage In the knowledge base.

Here, according to recommending occurrence number in knowledge base of answer and its corresponding current problem, recommendation apparatus 1 can be with Its priority is determined, and according to priority answer will be recommended to be supplied to the active user, further improve user uses body Test.

More preferably (referring to Fig. 3), this method also includes step S307 (not shown)：According to the active user described The actual answer of the current problem is directed in current sessions, adjusts the priority for recommending answer.

Specifically, in step s 302, recommendation apparatus 1 is supplied to by the recommendation for matching to obtain in step S301 answer After user, user can select according to or refer to the recommendation answer to answer the current problem of existing customer, can also select Select and do not answered according to the recommendation answer, therefore, in step S307, recommendation apparatus 1 can obtain the active user current The actual answer of the current problem is directed in session, for example, in step S307, recommendation apparatus 1 with active user by being made The interaction of voice call device, reality of the active user for the current problem in current sessions is obtained in real time and is returned Answer；Then, in step S307, recommendation apparatus 1 is according to the actual priority answered, adjust the recommendation answer, for example, in step In rapid S307, recommendation apparatus 1 by the actual answer got compared with recommending answer, if unanimously, then it represents that the current use Family employs the recommendation answer, and recommendation apparatus 1 can improve the priority of the recommendation answer, if inconsistent, then it represents that this is current User does not use the recommendation answer, and recommendation apparatus 1 can reduce the priority of the recommendation answer.

Preferably, in step S307, recommendation apparatus 1 can recommend the utilization rate of answer by being directed to a large number of users Statistics, then row adjust the priority of the recommendation answer, for example, a utilization rate threshold value can be set, when making for the recommendation answer When being less than the utilization rate threshold value with rate, the priority of the recommendation answer is reduced.Here, the utilization rate threshold value is, for example, system intialization , it can also be adjusted according to actual conditions by user.

Here, actual answer or the recommendation answer can be written form or speech form, if the reality It is all speech form to answer and recommend answer, then in step S307, recommendation apparatus 1 can first pass through the side such as speech recognition Formula, actual answer and the recommendation answer is converted into text information, then the text information is compared.

Here, it is not to represent the rwo " completely the same, tally in every detail " that " actual to answer " is " consistent " with " recommendation answer ", and It is as long as that similarity reaches predetermined threshold, you can think that the actual answer is consistent with recommending answer, for example, for having been converted into The actual answer of text information and recommend answer, if the similarity of the rwo reaches 90%, it is considered that the actual answer and Recommend answer consistent.Here, the predetermined threshold is, for example, system intialization, can also be adjusted according to actual conditions by user It is whole.

Here, whether recommendation apparatus 1 can also be according to the actual answer of active user with recommending answer consistent, to adjust this Recommend the priority of answer, be effectively utilized the actual feedback of user, further improve the usage experience of user.

Preferably (referring to Fig. 3), this method also includes step S308 (not shown)：Worked as according to the active user with described Question and answer of the preceding client in the current sessions, the attribute information of existing customer described in real time modifying.

Specifically, active user and existing customer can have contact question and answer several times, existing customer in current sessions It can be quizmaster or answerer, in this prior among the enquirement or answer of client, the current visitor may be implied The attribute information at family, for example, implying the attribute informations such as the industry of the existing customer, region, scale, these existing customers are being worked as The attribute information for revealing out in preceding session may have with attribute information known to prestore in system or active user Enter, therefore, in step S308, recommendation apparatus 1 can obtain the active user and question and answer of the existing customer in current sessions, For example, by being interacted with voice call device, the question and answer are obtained, and the attribute letter of the existing customer is extracted from the question and answer Breath, if the attribute information of the existing customer of the attribute information from being prestored in system is different, in step S308, recommendation apparatus 1 changes the attribute information of the existing customer in real time.

Based on the amended attribute information, in step S301, recommendation apparatus 1 can match for the existing customer again Obtain it is corresponding with the current problem in the current sessions recommend answer, or, in step S301, recommendation apparatus 1 can be For the next round problem of the existing customer matching when, matching recommendation answer corresponding with the amended attribute information；Or Person, if the existing customer is over the current sessions, the amended attribute information can be used under client in this prior Recommend answer corresponding to being matched during session.

Here, question and answer of the recommendation apparatus 1 according to existing customer, its attribute information of real time modifying so that follow-up matching behaviour It is more accurate to make, and further improves the usage experience of user.

Preferably (referring to Fig. 3), this method also includes step S309 (not shown)：The active user is obtained with described to work as The current sessions between preceding client, recorded as historical session, to carry out corresponding Knowledge Extraction.

Specifically, for the current sessions between active user and existing customer, it can also be used as historical session to record, For carrying out Knowledge Extraction, in step S309, recommendation apparatus 1 for example by being interacted with voice call device, obtains in real time The current sessions are taken, for example, whenever the active user or existing customer are finished in short, in step S309, recommendation apparatus 1 is just The word acquisition is come, or, when question and answer all in the current sessions are all completed it by the active user and existing customer Afterwards, in step S309, the ability of recommendation apparatus 1 is obtained all voices in the current sessions, and using these voices as Historical session records, so as to carry out corresponding Knowledge Extraction, for example, transferring to recommendation apparatus 1 to carry out Knowledge Extraction.

It should be noted that the present invention can be carried out in the assembly of software and/or software and hardware, for example, this hair Bright each device can using application specific integrated circuit (ASIC) or any other realized similar to hardware device.In one embodiment In, software program of the invention can realize steps described above or function by computing device.Similarly, it is of the invention Software program (including related data structure) can be stored in computer readable recording medium storing program for performing, for example, RAM memory, Magnetically or optically driver or floppy disc and similar devices.In addition, some steps or function of the present invention can employ hardware to realize, example Such as, coordinate as with processor so as to perform the circuit of each step or function.

It is obvious to a person skilled in the art that the invention is not restricted to the details of above-mentioned one exemplary embodiment, Er Qie In the case of without departing substantially from spirit or essential attributes of the invention, the present invention can be realized in other specific forms.Therefore, no matter From the point of view of which point, embodiment all should be regarded as exemplary, and be nonrestrictive, the scope of the present invention is by appended power Profit requires rather than described above limits, it is intended that all in the implication and scope of the equivalency of claim by falling Change is included in the present invention.Any reference in claim should not be considered as to the involved claim of limitation.This Outside, it is clear that the word of " comprising " one is not excluded for other units or step, and odd number is not excluded for plural number.That is stated in system claims is multiple Unit or device can also be realized by a unit or device by software or hardware.The first, the second grade word is used for table Show title, and be not offered as any specific order.

Claims (17)

1. a kind of dialogue-based method recommended, wherein, this method comprises the following steps：

Based on the current sessions between active user and its existing customer, according to the attribute information of the existing customer, from knowledge Matching obtains recommendation answer corresponding with the current problem in the current sessions in storehouse；

The recommendation answer is supplied to the active user；

Wherein, the structure of the knowledge in the knowledge base includes<Client properties, problem, answer>.

2. according to the method for claim 1, wherein, the knowledge base is established or updated by following processes：

Obtain the historical session record between each user and its client；

The historical session is recorded and carries out Knowledge Extraction, knowledge corresponding to acquisition, wherein, the structure of the knowledge includes<Visitor Family attribute, problem, answer>；

According to the knowledge, establish or update the knowledge base.

3. according to the method for claim 2, wherein, described recorded to the historical session carries out Knowledge Extraction, acquisition pair The step of knowledge answered, includes：

Historical session record is converted into corresponding session text message；

Knowledge Extraction is carried out to the session text message, obtains the knowledge, wherein, the structure of the knowledge includes<Client Attribute, problem, answer>.

4. the method according to claim 11, wherein, it is described that Knowledge Extraction, acquisition pair are carried out to the session text message The step of knowledge answered, includes：

The session text message is abstracted as the binary structure of knowledge, wherein, the binary structure of knowledge includes<User or visitor Family, what is said or talked about>；

The binary structure of knowledge is reassembled as the ternary structure of knowledge, wherein, the ternary structure of knowledge includes<Client, problem, Answer>；

Each element in the ternary structure of knowledge is clustered respectively, obtains the knowledge, wherein, the knot of the knowledge Structure includes ＜ client properties, problem, answer>.

5. method according to any one of claim 1 to 4, wherein, this method also includes：

According to the occurrence number of the recommendation answer and its corresponding current problem in the knowledge base, determine that the recommendation is answered The priority of case；

Wherein, the described the step of recommendation answer is supplied into the active user, includes：

According to the priority, the recommendation answer is supplied to the active user.

6. according to the method for claim 5, wherein, this method also includes：

The actual answer of the current problem is directed in the current sessions according to the active user, the recommendation is adjusted and answers The priority of case.

7. method according to any one of claim 1 to 6, wherein, this method also includes：

According to the active user and question and answer of the existing customer in the current sessions, existing customer described in real time modifying Attribute information.

8. method according to any one of claim 1 to 7, wherein, this method also includes：

The current sessions between the active user and the existing customer are obtained, are recorded as historical session, to carry out Corresponding Knowledge Extraction.

9. a kind of dialogue-based recommendation apparatus recommended, wherein, the recommendation apparatus includes：

For based on the current sessions between active user and its existing customer, according to the attribute information of the existing customer, from Matching obtains the device for recommending answer corresponding with the current problem in the current sessions in knowledge base；

For the recommendation answer to be supplied to the device of the active user；

Wherein, the structure of the knowledge in the knowledge base includes<Client properties, problem, answer>.

10. recommendation apparatus according to claim 9, wherein, the recommendation apparatus also includes：

For obtaining the device of the record of the historical session between each user and its client；

Knowledge Extraction is carried out for being recorded to the historical session, the device of knowledge corresponding to acquisition, wherein, the knot of the knowledge Structure includes<Client properties, problem, answer>；

For according to the knowledge, establishing or updating the device of the knowledge base.

11. recommendation apparatus according to claim 10, wherein, it is described to be used to take out historical session record progress knowledge Take, the device of knowledge corresponding to acquisition includes：

For historical session record to be converted to the unit of corresponding session text message；

For carrying out Knowledge Extraction to the session text message, the unit of the knowledge is obtained, wherein, the structure of the knowledge Including<Client properties, problem, answer>.

12. recommendation apparatus according to claim 11, wherein, it is described to be used to take out session text message progress knowledge Take, the unit for obtaining the knowledge is used for：

The session text message is abstracted as the binary structure of knowledge, wherein, the binary structure of knowledge includes<User or visitor Family, what is said or talked about>；

The binary structure of knowledge is reassembled as the ternary structure of knowledge, wherein, the ternary structure of knowledge includes<Client, problem, Answer>；

Each element in the ternary structure of knowledge is clustered respectively, obtains the knowledge, wherein, the knot of the knowledge Structure includes<Client properties, problem, answer>.

13. the recommendation apparatus according to any one of claim 9 to 12, wherein, the recommendation apparatus also includes：

For the occurrence number according to the recommendation answer and its corresponding current problem in the knowledge base, it is determined that described push away Recommend the device of the priority of answer；

Wherein, it is described to be used to be supplied to the device of the active user to be used for the recommendation answer：

According to the priority, the recommendation answer is supplied to the active user.

14. recommendation apparatus according to claim 13, wherein, the recommendation apparatus also includes：

For being directed to the actual answer of the current problem in the current sessions according to the active user, pushed away described in adjustment Recommend the device of the priority of answer.

15. the recommendation apparatus according to any one of claim 9 to 14, wherein, the recommendation apparatus also includes：

It is current described in real time modifying for according to the active user and question and answer of the existing customer in the current sessions The device of the attribute information of client.

16. the recommendation apparatus according to any one of claim 9 to 15, wherein, the recommendation apparatus also includes：

For obtaining the current sessions between the active user and the existing customer, recorded as historical session, with The device of Knowledge Extraction corresponding to progress.

17. a kind of computer equipment, the computer equipment includes：

One or more processors；

Memory, for storing one or more computer programs；

When one or more of computer programs are by one or more of computing devices so that one or more of Processor realizes the method as any one of claim 1 to 8.