CN106407196A - Semantic analysis intelligent instruction robot applied to logistics management software - Google Patents
Semantic analysis intelligent instruction robot applied to logistics management software Download PDFInfo
- Publication number
- CN106407196A CN106407196A CN201510450049.1A CN201510450049A CN106407196A CN 106407196 A CN106407196 A CN 106407196A CN 201510450049 A CN201510450049 A CN 201510450049A CN 106407196 A CN106407196 A CN 106407196A
- Authority
- CN
- China
- Prior art keywords
- instruction
- resource
- parameter
- user
- node
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Pending
Links
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F16/00—Information retrieval; Database structures therefor; File system structures therefor
- G06F16/30—Information retrieval; Database structures therefor; File system structures therefor of unstructured textual data
- G06F16/33—Querying
- G06F16/332—Query formulation
- G06F16/3329—Natural language query formulation or dialogue systems
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F40/00—Handling natural language data
- G06F40/30—Semantic analysis
Abstract
The invention refers the conventional natural language analysis technology, and discloses a language analysis and instruction analysis algorithm based on large logistics management software. In one aspect, instruction information of an instruction of a natural language input by a user can be extracted through a semantic analysis engine, wherein the instruction information includes key information such as an instruction operation target, an operation command and parameters; in the other aspect, an execution demand of the user and an actual resource condition of a system is combined through an instruction analysis engine, and an executable scheme including an AJAX format command and description information is fed back to the user; and a semantic analysis intelligent instruction robot applied to logistics management software is established by combination of the two aspects, the user can directly input a natural language type instruction to directly operate the logistics management software, and the fuzziness and ambiguity of the natural language can be supported to some extent; the problem that the conventional management software is complex in interface, and an operator is difficult to operate the conventional management software can be solved; man-machine interaction is humanized and simple; and the operation efficiency is high.
Description
Technical field：
The present invention is based on logistics management software platform, has used for reference existing natural language analysis technology, and independent development is applied to logistics management software
A series of algorithm of linguistic analysiss and instruction analysis and flow process, can parse to the order of the natural language type of user input, and to a certain degree
The upper ambiguity supporting user input natural language and polysemy, construct the intelligent instruction type robot that can directly operate logistics management software.With
Family can directly operate logistics management software by input language.The technology platform that the present invention relies on is a comprehensive logistics management software, for
The difficult common fault of large-scale management software interface complexity, complex operation, user's left-hand seat, the intelligent instruction type robot being created using the present invention, Yong Huneng
Enough by simple, closer to ordinary language instruction Lai direct operating system, man-machine interaction more hommization, left-hand seat faster, and is conducive to raising
Operating efficiency.
Inside intelligent robot designed by the present invention, operator can directly input the order of natural language type：Such as：" create waybill
Delivery side is pepsi cola, and goods is 500 kilograms of colas " or " creating financial affairs receipt, object is that Zhang Lao is big, and receivable 5000 is fast " or " search
Outbound list of nearest one month Chengdu website of rope " etc..Automatically that user is defeated by the semantic analysis engine and instruction analysis engine of intelligent robot
The language entering, be parsed into system it will be appreciated that, the one or more of orders that can execute, and the corresponding descriptive comment of every order,
Feed back to client inside inputting interface, client can select command according to actual needs, click on execution, operation interface is as shown in accompanying drawing Figure 11.
One aspect of the present invention, has used for reference natural language analysis technology and algorithm, and the range based on itself：Logistics management software, independently opens
Send out a set of intelligent semantic parser, by the instruction resources bank providing based on physical management software platform, the read statement of user is analyzed to be needed
Instruction resource to be executed, including resources such as the order of execute instruction, target and parameters, and can support the ambiguity of natural language to a certain extent
And polysemy.On the other hand, system can analyze and a series of be available for customer selecting according to the demand of user input, according to the resource of system itself
Executable scheme, the description information including the instruction that can execute and execute instruction link.
Background technology：
The present invention based on background technology mainly have both direction：One is large-scale logistics management software platform, and two is existing natural language analysis technology.
The large-scale logistics management software that the present invention is based on, is the comprehension management software for logistics transportation industry, covers transportation management, storage
The many aspects of the loglstics enterprise operation workflow such as management, customer relationship, financial management, office automation, on the one hand have powerful advantage,
But also there is " common fault " of large-scale management software：Flow process is complicated：User left-hand seat is relatively difficult, the problem of resource searching inefficiency.This software
Technology platform, based on J2EE platform development, backstage core code resource is mainly made up of JAVA code, and this software desk Implementation B/S is (clear
Look at device/server) framework, user does not need specially to install client, but direct operating system inside browser.
On the other hand：Existing natural semantic analysis technology is gradually ripe, occurs in that many intelligent semantics based on semantic analysis technology analyze machine
People, such as Turing machine device people etc..Natural language processing can be based on, and achieve self-teaching and the extension of Custom Knowledge Base, end up till now,
Turing machine device people's platform has accumulated the basic data of nearly hundred million corpus and GB rank, measures through authoritative mechanism, and recognition correct rate has reached
To 85%.But the application based on semantic analysis technology in the market, and the expanded application based on semantic analysis machine people, also focus at language
Reason itself, such as translating robot, chat tool, the application of knowledge question instrument aspect, the effect of language intelligence analysis does not have really to play
Out.
Natural language analysis technology and logistics management software are combined by the present invention, construct the more efficient, operation interface of more hommization：I.e.
The semantic analysis that logistics management software can directly be operated intelligently instruct robot, and the Structure and Process of whole system is as shown in Figure 1.This intelligent dictating machine
Device people provides the simple interface the same similar to common chat software, and as shown in accompanying drawing Figure 11, user does not need inside complicated system interface
Operation is it is only necessary in the equally simple interface of chat tool, the instruction of input language type, as " created waybill XXX, inquiring customer information
XXX " etc..Intelligent robot is intended to according to the real resource situation of user input and system, the operation that after analysis, user is possible, and same
In individual inputting interface, various feasible executable schemes are fed back to client.Every suit can perform scheme and carries complete description information, carries simultaneously
Have and with the link of execute instruction, and series of parameters can be carried.User is held according to the information of system feedback, the operation scheme selecting oneself to need
OK.
Operator instruct the man-machine interface of robot manipulation's logistics management software using intelligence, are simple interface similar to chat software,
As shown in accompanying drawing Figure 11.In such interface, user's picture system as daily chat proposes order, obtains the feedback of system and executes order.
The present invention, by complicated for original flow process, complex operation management software interface, simplifies as a simple chat interface, the operation of user is also by traditional machine
The mode of operation of tool formula, becomes closer to the operating language in human language so that man-machine interaction is simpler, more humane.
The operating process of intelligent robot is as follows：As shown in accompanying drawing Figure 11：User clicks on opening operation interface, input order：Such as：" create
Waybill delivery side is pepsi cola, and goods is 500 kilograms of colas ".Robot according to the instruction of user input, according to the real resource situation of system,
Judge that user may need the sentence executing, and feed back to user, a plurality of executable scheme is shown on the interface of chat tool：Such as：" 1. create
Build waybill, delivery side is corporate client：【Pepsi cola Chengdu branch company】, Description of Goods is：【Laughable】, weight is【500】Kg " and " 2.
Create waybill, delivery side is corporate client：【Pepsi cola Xi Pu factory】, Description of Goods is：【Laughable】, weight is【500】Kg " etc.,
And have the link of clicking operation below.
User according to actual needs, chooses a certain bar instruction, after click, is directly entered waybill and creates the page, and the page is according to user instruction
The information comprising has set parameter, and such as delivery side is set to " pepsi cola Chengdu branch company ", and Description of Goods is " laughable ", weight 500kg,
As shown in accompanying drawing Figure 12.
As can be seen here：The semantic analysis machine people of logistics management software, has carried out semantic parsing and instruction parsing so that user language to user instruction
The instruction of type can directly operating system so that logistics software using more hommization, more press close to the routine use custom of people it is easier on
Handss.The present invention optimal use scene：One is to systemic-function and the unfamiliar user of resource distribution, can help them to system quickly
Introduction, can also more effectively find the resource of needs, and two is the comparatively laborious function of operation, and such as waybill creates, and needs setting delivery side, receives
The information such as cargo interests, Description of Goods, weight, need to switch to and fro in several pages just can complete, can be disposably by institute by intelligent robot
The resource distribution needing is good, improves the efficiency of execution.
Ultimate principle, process and difficult point that the technology of system is realized：
The command operation mode of user input natural language property, compares with mode of operation inside traditional interface, difficulty increased a lot：
First difficult point shows in the ambiguity and inaccuracy of input instruction：This ambiguity one is to be determined by the property of user language itself,
Routine use language itself is not 100% accurate, and the language of user language input can not possibly be with the instruction resource 100% that can run in system
Join, though user in the case that system resource is very familiar with it is ensured that the sound instruction directivity of input clearly, the finger that language and system can execute
Order also has certain deviation, let alone in the case of common, user initiates order fully according to the daily custom spoken to system, and fuzziness is bigger.
As chapters and sections introduction above, in actual mechanical process, we can be according to the input of user language instruction, according to certain fuzzy matching degree, instead
The multiple scheme that can execute of the user that feeds, the instruction of the wish input of user installation oneself, system often feeds back the finger that many bars can execute
Make scheme, select execution for user.
Next to that the inaccuracy that semantic machine people operation brings, inside traditional system operatio interface, during submiting command, first, order itself
With accurate parameter, and in the interface of intelligent robot, the instruction of user language input can not possibly be with 100% completely accurate parameter；Than
In traditional interface, when choosing certain waybill to enter edlin or check, inside the hiding parameter that the page is submitted to backstage, contain this waybill
UUID, usually the uniqueness coding of 16 or 32, can uniquely determine the documents such as waybill.And in the interface of intelligent robot, use
Family is only possible to input close to fuzzy, the inaccurate parameter of ordinary language custom it is impossible to input is similar to the equally accurate parameter of UUID.
Second, during traditional operation interface operation, the different pages represents different business scenarios, the contextual information that the page itself is also concealed with,
Inside the page of such as waybill, all of operation is both for what waybill was carried out, is therefore operated inside the waybill page, and system can automatic mapping
To in this operation target resource of system background " waybill ".
Further say：The button of each operation or action inside traditional interface, are also completely one-to-one with backstage order, such as
A certain bar waybill is entered with edlin, execution is backstage " editor " order, the button of click " waybill is newly-built ", execution is that backstage is " new
Build " order.And instruct inside the interface of robot in general intelligence, these contextual informations being brought by the UI element such as traditional page and button
All without unique information source is the analysis to user language.
The process of inaccurate user instruction：One is the built-in algorithm supporting fuzziness of existing semantic analysis engine, can according to each resource
On node, the matching degree of setting, such as matching degree 80%, carry out keyword match, and also pre-set some ambiguities, many in semantic analysis engine
The parameter of justice, eventually matches in multiple accurate system command resources.Two is that system instruction analysis engine below (may be referred to Fig. 1：System
System Structure and Process), the instruction for ambiguity has carried out related process：It is provided with the instruction resource of ambiguity, these resources in instruction resources bank
Including fuzzy performance objective, order and parameter resource, the instruction resource of ambiguity is not the resource that may finally execute, and needs according to certain calculation
Method, matches on data resource that is accurate, can finally executing further.Such as：" delivery side " is the instruction of an ambiguity inside system
Parameter resource, when user input " delivery side " information, eventually according to certain algorithm, matches order parameter " the individual Crinis Carbonisatus of real execution
Cargo interests " and " delivery side of enterprise " are above.
Second difficult point be how by the order of natural language property of user input, construct the instruction that final system can execute and accurately
Description information.
The instruction of user input natural language property, and finally need to feed back to a plurality of feasible the carrying into execution a plan of user, including referring to can click on execution
The page link of order, and detailed description information, build the instruction that can execute and include building the order that can execute and relevant parameter, and every
The accurate description information carrying into execution a plan.
The present invention based on logistics management software be B/S (browser/server) framework, final order execution is by submitting AJAX lattice to backstage
Realizing, the content that AJAX instruction is submitted to includes order execution address and command parameter for the instruction of formula.
AJAX instructs the composition of resource address：Previous section is the general address of server and engineering, and remainder can be by instruction target and order
Two partly built-up, and when such as waybill creates, the command address that needs are submitted to is：http：//127.0.0.1：8080/torstein/bookingNote/create.html.Wherein,http：//127.0.0.1：8080/torsteinPart be server and
The public address of engineering project, the part of all of resource-sharing of system." bookingNote " is the resource address of instruction target, represents present instruction
The target of execution is " waybill (bookingNote) "；" create " is the resource address of designated command sentence, represents that the order of execution is and " creates
(create)”.Obtained by analysis above, the address that instruction execution is submitted to can be combined by performance objective and exectorial resource address.
Meanwhile, the execution of instruction also needs to certain parameter information, including parameter name and parameter value.As in the example that above waybill creates,
When waybill being created by AJAX order in system, if specifying delivery side to be the existing corporate client of system【Pepsi cola Chengdu branch company】, hold
During row, the parameter information of needs submission is：senderCorporate：" pepsi cola Chengdu branch company ", that is, parameter name is
" senderCorporate ", corresponding parameter value is " pepsi cola Chengdu branch company ".
Analysis above is as can be seen that construct the order that system can execute and parameter it is necessary to by the linguistic analysiss of user, obtain three necessity
Factor：1. operate target, 2. operational order, 3 command parameters.The instruction resources bank of system, these three factors, as instruction resource, are pressed
Stored according to tree structure and managed, the instruction resource of storage inside system command resources bank, it is referred to as " secondary response instruction resource ", such
Resource, by the instruction analysis engine control based on physical management software, during instruction analysis, carries out the coupling of resource and maps again." secondary
The structural representation of level response instruction resource " refer to Fig. 2.The root of tree structure is operation target resource, and leaf node is executable command
Parameter, and every first-level instruction resource all saves the corresponding address element of this grade of resource.Instruction analysis engine is finding the " secondary of final coupling
After response instruction resource ", constructed entirely to the order of leaf node (command parameter) according to address element according to from root node (operation target)
The address resource of manipulable order.And the description information of executable instruction, the process of structure and order build process ratio relatively similar, be also by
Operation target, order, order parameter are constituted successively, and " secondary response instruction resource " not only contains the address factor of executable command, also comprises
Description information, after instruction analysis engine completes to instruct the coupling of resource, according to from root node (operation target) to leaf node (command parameter)
Sequentially, building can be with the description information of execute instruction.
3rd difficult point is the semantic analysis technology of natural language：How to extract basic operation target, order from user input natural language
With key messages such as parameters.As described above, instruction analysis engine needs, from user language, to extract target, order and parameter information,
Respectively as input signal, then could carry out coupling in instruction analysis engine neutralized system resource and analyze again.
How this problem in the course of the study, is especially realized in natural language analysis technology, experienced more difficulty, during research, asks
Topic is concentrated mainly on the direction of semantic analysis technology realization：One is based on existing intelligent semantic analysis robot, and such as Turing machine device people realizes language
Speech analytical technology；Two is independent development linguistic analysiss algorithm, is specifically designed for this logistics management software and uses.
Finally we analyze the advantage shortcoming in two kinds of directions, combine the application scenarios of itself, have used for reference existing semantic analysis technology, self-developing
Based on the semantic analysis engine of instruction resources bank, and it is exclusively used in this logistics management software.
So select, as follows the reason concrete：
1. semantic analysis range according to the present invention and vocabulary quantity are limited：It is limited only to logistics management software to use, be operable to system
Instruction analysis, rather than common chat software, the lexicon ratio of semantic analysis is relatively limited, can be by manually being safeguarded it is not necessary to be utilized
The magnanimity lexicon of existing intelligent semantic analysis robot.
2. the linguistic context processing is fairly simple, and the disposable sentence length needing to analyze is limited：The length of metalanguage is defined to a sentence, only
Need to analyze limited verb and noun it is not necessary to analysis auxiliary word, adjective etc., it is not necessary to linguistic context before and after considering, therefore, do not have complexity
Language environment need to process it is not necessary to process in natural language complicated polysemy and ambiguousness using complicated semantic analysis algorithm, and permissible
The polysemy of language and the process of ambiguity are realized by the instruction analysis engine of designed, designed semantic analysis engine and rear end.If running into lexicon
The word that can not parse, directly can be processed as illegal command.
3. existing semantic analysis machine people, the interface function providing at present is single：Turing machine device people's interface of once investigating as early stage it is impossible to
The support of variable is provided, leads to not realize the transmission of order parameter inside logistics management software scene it is impossible to realize supporting instruction analysis stream below
Journey.
Content of the invention：
In the present invention, overall process that user command is processed, actually by the instruction of the natural language of user input, a series of analyses of process
After process, produce the process of the scheme that can execute.
Inside this logistic management system, executable scheme by the order of AJAX form, parameter (the AJAX instruction link that can execute), and
Descriptive language composition.
User command processes whole flow process as shown in figure 1, three phases can be divided into：Semantic analysis, instruction analysis and instruction combination of resources.
Wherein the semantic analysis stage is to be executed by semantic analysis engine, by the natural language of user, according to system command resource, be parsed into most basic target,
Order, parameter information etc. carry " the primary response instruction " of structural information.Instruction analysis are to be executed by instruction analysis engine.By " primary response refers to
Make ", analyze and change into the address of AJAX order that really can execute and parameter sets further.
Semantic analysis phase flow：The natural language class order of analysis user input, extracts the key message of three aspects：Instruction target, refer to
The command action of order and parameter information.
The process that semantic analysis engine executes semantic analysis is as follows：
The first step：Engine load instruction resources bank first, the data structure of instruction resources bank is as shown in Figure 2：Data structure presents tree structure,
And it has been divided into " primary resource response part " and " secondary resource response " part.Wherein " secondary resource response " is partly the main body of resource,
Contain from root node " instruction target ", " order ", the node to leaf node " parameter " Deng You the superior and the subordinate structural relation, these are " secondary
The node of resource response part ", is referred to as " resource node ".
In every one-level " resource node ", all hang with " primary responsive node ", for during semantic analysis, according to setting matching degree and
Matching content, and the input language of user mated.From fig. 2 it can be seen that all " resource node ", " primary sound all can be included
Answer node ", " primary responsive node ", in instruction resources bank, is collectively referred to as " primary resource response part ".
Second step：Key of Implementation word mates：According to the matching content (content) of " primary resource response " at different levels node setting, and matching degree
(matchDegree) the character string forward direction matching algorithm, providing according to engine, mates to the sentence of user input, successively when certain keyword
When being made into work(, successful " primary resource response " is added in return parameterss, finally returns that parameter can be by a series of " primary resource responses "
The list of node instance composition, referred to as " primary response instruction ", and be input in instruction analysis engine below.
The order of semantic matches to be mated according to the sequencing of target resource, order resource and parameter resource：First attempt to mate target resource,
After coupling target resource success, according to the tree of resource response, attempt the child node of target resource command below resource is mated,
As shown in step 1, step 2 in Fig. 3.Legal " primary response instruction " is it is necessary to successfully match instruction target letter from the language of user
Breath and command information.If it is impossible to successfully match target resource and order resource in the read statement of user, it fails to match it is believed that using
Family input is illegal command statement.
The coupling flow process of target resource and order resource：The coupling flow process of both resources is relatively simple：We are by user input sentence
Find after observation, statistics, the target resource structure typically all single with order, and be all to show in plain text in user instruction.
Inside instruction resources bank, hang over " primary resource response " node below instruction target resource node and order resource node, in semantic analysis
During, directly according to matching content (content), matching degree (matchDegree), the character string forward direction matching algorithm being provided using engine, right
The sentence of user input is mated, once finding and keyword match success, just corresponding this two " primary resource response " nodes is added to
In return parameterss " primary response instruction ".
Fig. 3 gives a coupling flow process to target resource and order resource.When user input sentence is：" creating waybill delivery side can for Pepsi
Happy, goods is 500 kilograms of colas " when, coupling flow process is as follows：
Step 1：Mate target resource first.According to instruction resources bank in, target resource：" the primary sound that " waybill BookingNote " hangs below
Answer resource " matching content of node includes the keyword of " waybill ", then and the match is successful for target resource, is added to return parameterss：" primary response
In instruction ".In the same manner, step 2, in target resource target resource：In " waybill BookingNote " command below resource " newly-built Create "
The matching content of administrative " primary resource response " node includes the keyword of " establishment ", then order resource matched success, is added to return ginseng
Number：In " primary response instruction ".
Process to target resource, order resource semantic matches polysemy：If according to the matching degree of system setting and matching content, when there being many entries
When mark resource and order resource are capable of coupling, generate a plurality of executable scheme the most at last, select for user.
The coupling flow process of parameter resource：
Parameter resource, the structure of itself is more complicated, leads to matching logic and matching process more complicated.The data structure of parameter resource, not only wraps
Include parameter name, also comprise parameter value, some parameters present composite structured：Except parameter value itself, also carry unit and operative symbol (such as
It is more than, is less than etc.).In the sound instruction of user input, (current system only supports single order for target resource and order resource only one of which
Sentence, and single operation and single target can only be executed), and show in plain text in sentence, can be realized by direct keyword match.
And in the read statement of user, the parameter of input can be multiple, each parameter not necessarily can be mated with obvious keyword, and each parameter
Or with complicated parameter value, unit and operative symbol information, between different parameters, also have the contact of correlation.
In this case, semantic analysis engine is divided into several situations to the extracting mode of parameter：The first situation：The parameter of single structure is permissible
Parameter value is mated out, such as " delivery side by way of keyword directly mates：The parameter of pepsi cola " so type, using similar finger
Make target and the mode of order, using keyword match, extracting directly, can accurately obtain：Parameter name：" delivery side ", parameter value：
" pepsi cola ".
Second situation, parameter contains composite structured：We are found by statistics：With composite structured parameter, mainly carry unit
With operator both additional structures.For the parameter of single structure, such as " delivery side " " recipient " such parameter.In resource data knot
In structure, parameter resource node is leaf node in itself, does not have child node further below.For the parameter resource with unit and operator, instruction money
In the storehouse of source, below Parameter nodes, also have two child nodes：" Parameter units node " and " parameter operator node ".As shown in figure 4, this two
Also " primary responsive node " is hung with respectively, for semantic analysis coupling above adnexa node.
Such as：" goods weight "：This carries the parameter resource of Parameter units, and the data structure in resources bank is as shown in Figure 5.Parameter resource
Node, entitled " goods weight ", hang with the Parameter units node of multiple correlations below, such as：" kilogram ", " gram ", " pausing ".Often
Individual Parameter units node all includes " primary responsive node ", " primary responsive node " that such as " kilogram " this unit-node is hung, includes
Matching content " kilogram, KG, kg, kilogram ", when comprising such keyword in the language of user input, it is possible to match " kilogram ", this is heavy
On amount unit-node.
The third situation, parameter itself not only itself contains labyrinth, further comprises other subdivision parameters, such " parameter " substantially
It is the general designation being made up of several related parameters.Such as, in the example above：" goods：500 kilograms of colas ", " goods " this parameter,
The set that actually a series of subparameters related with goods information are constituted, it actually contains " goods weight ", " measurement of cargo ", " goods
The hiding parameter of the more subdivision such as title ".If parameter-the goods obtaining according to direct matching process：" 500 kilograms of colas ", is can not be direct
Match on " Description of Goods ", nor match on " goods weight ".Need through Secondary Match, according to certain logic, for the first time
The parameter value matching, further participle, mates, the parameter of the subdivision really being played a role further：" Description of Goods：Cola, goods
Weight：500 kilograms ".
How to realize this subdivision, the logic of Secondary Match parameter：On " primary resource response " node, by arranging field：
NextMappingProxyClass (Secondary Match proxy class), realizes the logic of Secondary Match, obtains " the primary response that next step really plays a role
Resource " node.When, on node, such is set to space-time, then do not need Secondary Match, after the match is successful, directly using the example of this node as
Join successful " primary resource response " node to return；If such setting is not empty, after this node matching success, semantic analysis engine needs reality
The example of this nextMappingProxyClass of exampleization (Secondary Match proxy class), and method mapToNextCommand being provided with such, real
The logic of existing Secondary Match, returns new return value：The example lists being made up of some " primary resource response " node.
The above example of " goods weight ", is also to be realized by the process of this " Secondary Match ".When user input " goods：500 kilograms can
Happy " when, do not mention the keyword of " goods weight " or " weight " in the read statement of user, according to " goods weight " shown in Fig. 5
Parameter resource structure, " goods weight " this Parameter nodes itself, cannot realize mating, and need according to following Parameter units node " kilogram "
There is provided keyword realize coupling, will user input word " kilogram ", match on Parameter units node, then restart this unit-node
" Secondary Match proxy class ", as shown in the example code of Fig. 6, the logic that provided according to this proxy class, reverse of " 500 " this keyword
Be fitted on " goods weight " this Parameter nodes, 500 as " goods weight " this parameter parameter value.
Fig. 6 gives the example of this Secondary Match proxy class, and such is defined on " kilogram " node, i.e. " goods weight " Parameter nodes, under
On unit of weight " kilogram " node in face, core logic is shown in the code in red frame：After when this node matching success, that is, in user input
Be found that in sentence can with the content of the node matching of unit of weight " kilogram ", then this Secondary Match proxy class can by semantic analysis engine start,
Enter the method：mapToNextCommand.
The core logic of the method：First, the product goods template class being provided using semantic analysis engine framework
LanDefProductTemplateProxy, returns the unit of goods weight, and is written to return value " primary resource response "：
List<LanuageFirstCommand>Inside result.In addition, the sentence being extracted according to front nodal " goods cargoWeightValue "
SubRawSentence is if it is possible to according to the extraction pattern pre-setting from product goods template class, extract the keyword of goods weight
The key value cargoName of cargoName and goods name, and depend on lanuageResourceNodeContext to instruct what resources bank exposed
Context interface, from goods weight " kilogram " node, extract respectively for the node of goods weight and the node of goods name, and phase
The value closed is given to their " primary responsive node " and is added in return value result.
To in input instruction, the coupling of useless, meaningless character and processing method：
In the example of above semantic analysis, it may be seen that some play the word of interconnection function, in such as " goods is cola " " for ", also
There are all of space, punctuation mark etc., these do not have the vocabulary of any practical function to command operating, need to filter out in semantic analysis, with
When current vocabulary be also important participle mark, represent some keywords, the mark that such as keyword match such as parameter, order terminates.Right in the system
These are meaningless, and the identification of useless vocabulary equally employs " primary responsive node " and mated, according to being defined on these " primary responsive node "
On matching content and matching degree carry out keyword match.But except for the difference that：" the primary responsive node " of nonsense words is not tied to certain " money
On source node ", but present " free state ", semantic analysis engine is during processing, once characters matching is to " free state "
" primary responsive node " when, just think the corresponding sentence of this node be nonsense words, need to filter out.
In sum, semantic analysis engine, by the resource of analysis instruction resources bank, is mainly based upon " primary resource response " node and is mated,
Finally obtain a series of list of " primary resource response " node instance composition of successful match, referred to as " primary response instruction ", give below
Instruction analysis engine, enters the instruction analysis of next step.
Instruction analysis engine, " the primary response instruction " of list structure is parsed into further the secondary response instruction of composite construction, and secondary response refers to
Make comprised information, on the one hand will meet AJAX execution form, meet again simultaneously and feed back enough, readability information to user, sieved
Choosing.
The secondary response instruction employing composite model structure contains more abundant information, and as shown in accompanying drawing Fig. 2, secondary response instruction resource is in
Reveal the tree structure with " target (target) resource " as root node, destination node is presented herein below the child node of order (command), illustrates one
Below target resource, all of executable command that may comprise, and instruction here and target are composite structured, all contain resource number id,
Display text label and instruction link linkURL.
Instruction analysis process：
The data structure of instruction resources bank, as shown in Fig. 2 passing through tree structure, defines all instruction resources that can execute, model inside system
Root node be " target resource ".The logistic management system that instruction analysis engine is based on, each commerce model that underlying platform is adopted, just
It is the target resource of.Such as newly-built waybill, searches for waybill, or newly-built financial account.Here business prototype：" waybill " or " wealth
Business account ", is the commerce model of logistic management system, is also the target resource of instruction." order resource " is " target resource " child node below,
One target resource can comprise multiple order resources, and such as " waybill " just contains multiple order resources below：As newly-built, editor, examination ＆ verification
Etc..
The process of instruction analytics engine execute instruction parsing, contains mapping process twice：Be for the first time mapping " secondary response instruction " target and
Order resource, is properly termed as " command mapping " again.According to the instruction target node in primary response instruction and command node, according in instruction resource
The structure in storehouse, directly returns corresponding secondary response instruction node.
Map for second, i.e. " parameter mapping ", the mapping for the first time relatively of its process is more complicated, and the flow process of mapping is by the control on parameter resource
Field：" mapping decisions class " determines, such illustrates the mapping logic of each |input paramete, if this parameter needs correlation in mapping process
Logical orientation in other parameters when, need to define this mapping decisions class.
When " mapping decisions class " is set to space-time, then by the way of " directly map ", that is, directly using the current parameter resource obtaining as next
The |input paramete of phase instruction combination of resources；If " mapping decisions class " setting is not empty, needs the definition according to Decision Classes, be mapped to new
One, even in multiple parameters, if parameter current needs to be mapped in multiple parameters, can form multiple secondary instruction responses, after eventually passing through
After the instruction combination of resources in face, a plurality of instruction that can execute of client can be returned to.If " mapping decisions class " is space-time, employ " direct
Join " mapping mode, directly carry out by the title and all parameters mating below the order resource obtaining for the first time of parameter in the instruction of first secondary response
Coupling, after the match is successful, be directly sent to the phase III carry out instruct resource combination, during Fig. 8 describes instruction analysis, instruct resource response
The control flow mapping twice.
In actual moving process, " mapping decisions class " that instruction analysis engine is arranged according to parameter resource, carry out D ynamic instantiation, and reflected
Penetrate logic control.
Illustrate the flow process that " mapping decisions class " controls mapping：Assume when executing the order of " waybill establishment ", primary response instruction is used
Family input parameter be：Delivery side (sender) is " pepsi cola "：And in background system, the corresponding real resource in delivery side is " individual client "
" corporate client " two kinds of customer resources, when the order of user input cannot judge specific customer resources type, matches for the first time
Secondary resource response is：" delivery side (sender) " of universal class, and this general " delivery side (sender) " be not final can be straight
Connect the parameter of execution, need according to certain logical mappings to real " individual client " that can execute and " corporate client " up.
The parameter resource being somebody's turn to do " delivery side " is provided with " mapping decisions class " in instruction resources bank, and example code is as shown in Figure 7.In this Decision Classes
Logic substantially：First according to this parameter：" pepsi cola " in the input value of " delivery side ", such as this example, goes " individual client " respectively
Inquired about in the database table of " corporate client ", if in customer resources table, having inquired the data record of coupling, then will be based on this
The customer data of type, to create waybill as delivery side, without the data record inquiring coupling, then " the delivery side " based on this parameter
Input value, as zero when property client, and as delivery side, to implement to create waybill, and during waybill creates, simultaneously need to being created that
Client when zero.
Assume, in real system runs, when user input delivery side information is " pepsi cola ", there is following several scene：
Scene 1：" mapping decisions class " through logical judgment, finds to have had in backstage " corporate client " database table and is named as " pepsi cola "
Enterprise-level customer resources, then parameter resource need to be redirected to the ginseng of delivery side of standard enterprise from general delivery side's parameter " delivery side (sender) "
Number resources " senderCorporate ", and to arrange the value of this parameter be " pepsi cola ", then can realize that " Pepsi can based on existing enterprise's resource
Happy ", as delivery side, implement waybill and create.
Scene 2：Assume, in actual motion, to find there is, in backstage " individual client " database table, the resource being named as " pepsi cola ",
Then need to be defined into the parameter resource " senderIndividual " of individual client, and to arrange the value of this parameter be " pepsi cola ".
Scene 3：Assume, in actual motion, all to have matched resource in " individual client " and " corporate client " resource simultaneously, then needed to reset
To on two parameter resource " corporate client delivery side senderCorporate " and " individual client delivery side senderIndividual ", and
And generate two secondary response instructions.Finally feed back on the interface of user it may appear that two executable instructions：Article one, it is based on the existing enterprise of system
Industry client " pepsi cola ", as delivery side, sets up waybill, and one is to have individual client's " pepsi cola " for delivery side based on system, builds
Vertical waybill.Client can be selected according to the wish of oneself.
Scene 4：During the hypothesis time is run, all cannot find resource in " individual client " and " corporate client " resource simultaneously, and " Pepsi can
Happy " input value realize coupling, according to the logic of " mapping decisions class ", need to be redirected to following two parameter simultaneously and " set up corporate client to send out
On cargo interests tmpSenderCorporate " and " setting up individual client delivery side tmpSenderCorporate " two parameters, that is, creating waybill
While, the customer information of delivery side of enterprise " pepsi cola " when simultaneously creating zero, or while creating waybill, create Crinis Carbonisatus when zero simultaneously
The customer information of cargo interests " pepsi cola ", user can select to need the sentence of execution according to the wish of oneself.
As can be seen here：It is defined on " the mapping decisions class " on parameter resource, there is provided to user input and background logic, backstage resource, handed over
Mutual chance.Analysis engine can determine whether the intention with analog subscriber, and feed back according to the practical situation of the input of user and system resource
Selected to client, so that the execution of client is more accurate to really can perform scheme.On the other hand, it is defined on as " delivery side ", " receive
Cargo interests " so has certain fuzziness, polysemy, " the mapping decisions class " on non-immediate execution resource node it is achieved that to user input ambiguity,
The support of polysemy parameter.
The process of instruction combination of resources：
According to the description of analysis process above, after instruction analysis, the resource that system obtains through instruction analysis is with " secondary response instruction "
The data of so composite construction exists, and cannot be fed directly to user, also cannot directly execute.So needing the process of a combination.
On the one hand, current system framework using AJAX format order by the way of seek unity of action the instruction of user, the final implementing result meeting of instruction
Unification is reoriented to client to be needed on the page of execution.Such as previous section for " establishment waybill " example in, the instruction feeding back to user is final
Result be to be reoriented to inside the page of waybill establishment, and inside the page, the input according to user has set parameter, in such as the example above
Face, the delivery side of waybill has been configured to " pepsi cola ".The form of instruction execution is unified AJAX form, can direct execute instruction, needs
Two Fundamentals：The address of execute instruction and the parameter of instruction.
The composition of execution address：Can be come by the link resources (linkURL) that " target " and " order " resource in " secondary response instruction " comprises
Combination producing, and parameter to be generated by instructing the final parameter resource after parsing.In logistic management system based on JAVA platform, AJAX form
Order, execution link generate form be：<Engineering chained address>/<Object linking address>/<Chaining command address>.html.Such as above create
In the example of waybill, whole address can be：【localhost：8080/projectName/bookingNote/create.html】.And AJAX
The parameter of instruction, can be directly by the name value preset of all command parameters that may finally execute, and the key-value of composition is to forming.
On the other hand：System also need to instruction describe out and feed back to client according to accurate, readability language, have been provided to client make do into
The judgement of one step, selection：In the same resources bank in instruction, each node resource：Order resource, target resource and parameter resource, have correspondence
Field " label (display text) " be used for indicating the descriptive information of this resource, instruction analysis engine can be according to certain rule to each resource
Display text is combined, and display is on a user interface.
In the example that waybill before such as creates, target resource corresponding display text is " waybill ", and order resource corresponding display text is " wound
Build ", delivery side " pepsi cola " is to be mapped to " corporate client delivery side senderCorporate " and " individual client delivery side simultaneously
During two parameters of senderIndividual ", their corresponding display texts are respectively【Delivery side is corporate client " %S "】With【Delivery side is
Individual client " %S "】, wherein " %S " is placeholder, is filled by parameter value during operation, and end-user interface can show two simultaneously can be executed
Order, and text is：
1. create waybill, delivery side is corporate client " pepsi cola ".
2. create waybill, delivery side is individual client's " pepsi cola ".
If delivery side " pepsi cola " does not find customer resources coupling inside system, two parameters that factor has arrived next stage " set up enterprise
When client delivery side tmpSenderCorporate " and " setting up individual client delivery side tmpSenderCorporate ", their corresponding displays
Text is respectively【Delivery side when creating zero：Corporate client " %S "】With【Delivery side when creating zero：Individual client " %S "】.User interface shows
The text showing becomes：
1. create waybill, delivery side when creating zero：Corporate client " pepsi cola ".
2. create waybill, delivery side when creating zero：Individual client's " pepsi cola ".
Brief description：
Fig. 1 illustrates the whole flow process of intelligent semantic analysis instruction robot work.
Fig. 2 illustrates the data model structure of instruction resources bank.
Fig. 3 illustrates in semantic analysis process by example, the coupling quick-reading flow sheets of target resource and order resource.
Fig. 4 illustrates the data model structure of compound parameter instruction resource.
Fig. 5, with " goods weight " as example, illustrates the data model structure of compound argument structure.
Fig. 6 illustrates in semantic analysis process, for goods weight unit " kilogram " " Secondary Match proxy class " realize code.
During Fig. 7 illustrates instruction analysis, for control this ambiguity parameter of waybill " delivery side " " control decision class " realize code.
During Fig. 8 illustrates instruction analysis, the control flow of instruction resource response mapping.
Fig. 9 illustrates in logistic management system, instructs the interface of resource for setting instruction resources bank.
Figure 10 illustrates in logistic management system, for the interface of setting instruction resources bank " primary resource response ".
Figure 11 illustrates in logistic management system, and operation intelligence instructs the interface of robot.
Figure 12 illustrates in logistic management system, and operation intelligence instructs robot, and clicks on the result after executable command.
Specific embodiment：
The specific implementation process of system includes the content of several aspects：Developer's how extended instruction resources bank under the framework of instruction analysis engine；
Include how the primary response instruction resource of extended instruction resources bank, to realize carrying out semantic analysis to the order of user input natural language, second is
The how secondary response instruction resource of extended instruction resources bank, to realize the mapping that instructs and combination finally feeds back to client only carries out AJAX order
With explanation text.Finally introduce in application, how operator operate intelligence instruction robot and operating effect inside system.
How to extend extended instruction information bank：
When system has new function, and intelligence is needed to instruct robot enforcement this New function of control, developer needs New function, as system
Instruction resource, is registered to inside the instruction data storehouse under instruction analysis engine.
Registration process is：Operator initial landing system command resource management interface, by navigation bar【System administration】-【Instruction resource management】,
Enter system command target resource searched page.User can enumerate and search system all instructions target resource.The data structure of instruction resources bank is such as
As shown in Fig. 2 to manage according to " target "-" order "-" parameter " such tree structure, target resource is root, parameter resource
It is slaves to order resource, and order resource is slaves to target resource.Therefore want parameter resource or order resource are operated it is necessary to
Enter from related upper level resource.
Edit operation to target resource：If necessary to newly-increased target resource, click on " increasing newly " inside target resource searched page, entry instruction
The editing interface of resource, edits existing target resource if necessary, chooses the target resource needing editor in lists, enters editor circle of resource
Face.Inside target resource editing interface, user needs to specify numbering ID, the link of target resource, and display text, in the combination of resource
This information is needed in journey.
" waybill " target resource of such as system, resource number ID is " BookingNote ", is linked as " bookingNote ", and display text is " fortune
Single ".The editing interface of target resource simultaneously, shows all order resources belonging to this target resource, the such as resource edit of " waybill " below simultaneously
Below the page, include " newly-built ", " editor ", " search ", the elementary instruction resource such as " checking ".
Edit operation to order resource：If user needs newly-increased order resource, or edits existing order resource, need to first confirm that one
Level instruction target resource, in target resource list of commands below, click on " increasing newly " button or.In the same manner, certain life is set if necessary
The lower parameter of order, needs to initially enter the edit page changing order resource, then following parameter setting partly in, increase newly or editing parameter provide
Source.
If this parameter carries transport symbol or unit, need, inside editing interface, to click on newly-increased " Parameter units " button and " parameter computing
Symbol " button is realizing newly-increased unit and parameter operator resource.
If this parameter can be directly as the parameter submitted to during execution order it is only necessary to name and the display text of arrange parameter be it is not necessary to setting is reflected
Penetrate Decision Classes, if the logic that the presence of this parameter maps further, need to be further mapped in the parameter that next stage directly operates in resolving
When, need to define mapping ruler class, realize mapping ruler, and the full name of mapping decisions class is arranged the field of parameter resource：Mapping decisions class
Above.
Flow process realized by the code of mapping ruler class：Need to realize " command mappings Decision Classes " standard interface：
【ILanuageComParaMappingDesicion】, and realize core methed：【executeMapping】.The example of mapping decisions class refer to figure
7.This example realizes code to the mapping decisions class of parameter " delivery side (sender) ", and such core methed is：By primary response instruction bag
The delivery side's parameter value containing, and the coupling of the corporate client in system, individual client's resource, according to different situations, are divided into 4 kinds of scenes, successively
Current parameter is mapped to the parameter of next stage：Personal delivery side, delivery side of enterprise, personal delivery side and delivery side of enterprise, and create personal
In the different next stage parameter such as delivery side and delivery side of establishment enterprise, to enter the combination of the resources such as line parameter and order, formation is finally able to carry out
Parameter.
The how primary resource response part of extended instruction resources bank：
The data structure of instruction resources bank as shown in Fig. 2 core, i.e. secondary resource response part, according to " target "-" order "-
Managing, every one-level " resource node " all comprises " primary resource response " node to " parameter " such tree structure, for semantic analysis
Engine carries out semantic analysis and coupling.In current system framework, one " resource node " all comprises " primary resource response " node, with
After can take into account and expand to multiple " primary resource response " node.
Edit operation to " primary resource response " node：Do not need newly-increased operation, after entering every one-level instruction target resource edit interface, can be certainly
Dynamic " primary resource response " node section, after directly selecting, clicks on Edit button, enters editing interface.
Wherein, field " matching content "：It is used for editing the keyword of semantic analysis engine coupling, during if there are multiple keyword, with comma separately,
Operation interface is as shown in figure 9, contain 4 coupling keywords inside goods weight unit " kilogram "：" kilogram, kilogram, kg, KG " is defeated as user
Enter to comprise in sentence during this four keywords it is believed that successful match is on the node of goods unit of weight " kilogram ".
Field " matching degree " is used for representing match-percentage during keyword match, and the matching degree of acquiescence is 80%：I.e. when the keyword of coupling has 10
During individual word, 8 words then think that the match is successful can to contact that the match is successful.
If necessary to define " Secondary Match logic ", such as from goods weight unit " kilogram " through Secondary Match, match goods weight and goods
Name deserves to be called face, then need developer's self-defined " Secondary Match proxy class nextMappingProxyClass ", and is arranged on this " primary response
On resource " node, as shown in Figure 9.Code realize " Secondary Match proxy class nextMappingProxyClass " when it is necessary first to inherit standard
Interface【ILanuageNextCommandMappingProxy】, and realize Secondary Match inside core methed mapToNextCommand
Logic.
User is in the process of system operatio semantic machine people：
First log into system, by navigation bar【Customer center】The user job center page can be entered.By clicking on button, start semantic
Analysis robot, pop-up dialogue box interface, is that the intelligent robot of logistic management system starts the page as shown in figure 11.User can be in this page
The instruction of middle input language type, and parameter can be added.As illustrated, working as user input：" create waybill, delivery side：" pepsi cola ",
Goods is 500 kilograms of colas " when, after systematic analysiss, according to the real data situation of system, two executable schemes of feedback user, it is " wound respectively
Build waybill, delivery side is corporate client【Pepsi cola Chengdu branch company】, Description of Goods is【Laughable】, weight is【500】Kg " and " establishment
Waybill, delivery side is corporate client【Pepsi cola Chengdu Xi Pu factory】, Description of Goods is【Laughable】, weight is【500】kg”.When user's choosing
Select first to carry into execution a plan, and after clicking on execution link, the page jump of system arrive waybill and creates the page, and automatic by delivery side of enterprise information,
The information of goods, is filled into inside waybill including Description of Goods and goods weight etc., has saved the time of operator, and has been conducive to system not
Familiar operator are faster familiar with the function of system.
Claims (7)
1. the present invention passes through to have invented semantic analysis engine and the instruction analysis engine based on physical management software platform, constructs the intelligence of logistics management software
Language manipulation interface, user can be inside general interface, by inputting the instruction direct operation management software of natural language type.
2. the intelligent language operation interface based on claim one description, instruction analysis engine can input according to the language form of user, after analysis,
It is inferred to that user is all possible to execute intention, and the real resource situation of coupling system, generate one or more executable scheme, including
The link of execution AJAX format command can be clicked on, with related parameter, and the corresponding description information of each executable scheme, feedback
On general operation interface, select for user and execute.
3., based on the instruction analysis engine described by claim one, all manipulable instruction resources of system, according to " instruction target-＞ operational order-＞
The tree structure data model storage of order parameter " and management are in the instruction resources bank of logistics management software, and the section of each tree structure
The information that information and executable instruction for semantic analysis coupling build is contained in point.
4., based on the instruction resource model described by claim three, any first-level instruction resource can support the instruction of the ambiguity of user input, polysemy
Resource, including fuzzy operation target, operational order, command parameter, developer passes through the " mapping decisions on User Defined Resource model node
Class " realizes fuzzy command resource to the mapping of one or more precise instructions resource and reorientation.
5. based on the instruction resource model described by claim three, below " command parameter " resource node, by the resource node of more lower：" parameter
Operator " and " Parameter units " resource node, to realize the coupling to compound parameter resource.
6. based on the instruction resource model described by claim three：The back end at different levels of " instruction target-＞ operational order-＞ order parameter " tree structure
On all contain and build the required address information of AJAX format order and description information, these addresss of node are believed by instruction analysis engine
Breath and description information, are combined into the final execution address of the AJAX order need in executable scheme and final description information according to certain way.
7., based on the instruction resource model described by claim three, in each data model node, include semantic analysis, required for Keywords matching
Essential information：By the setting of matching degree, for controlling the percentage ratio of Keywords matching, one or more is arranged by matching content
Keyword, realizes the semantic matching relationship of complexity by Secondary Match proxy class.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201510450049.1A CN106407196A (en) | 2015-07-29 | 2015-07-29 | Semantic analysis intelligent instruction robot applied to logistics management software |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201510450049.1A CN106407196A (en) | 2015-07-29 | 2015-07-29 | Semantic analysis intelligent instruction robot applied to logistics management software |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN106407196A true CN106407196A (en) | 2017-02-15 |
Family
ID=58008945
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201510450049.1A Pending CN106407196A (en) | 2015-07-29 | 2015-07-29 | Semantic analysis intelligent instruction robot applied to logistics management software |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN106407196A (en) |
Cited By (9)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN107423363A (en) * | 2017-06-22 | 2017-12-01 | 百度在线网络技术（北京）有限公司 | Art generation method, device, equipment and storage medium based on artificial intelligence |
| CN107861879A (en) * | 2017-11-28 | 2018-03-30 | 成都视达科信息技术有限公司 | A kind of method and system for realizing software automated testing |
| CN108170679A (en) * | 2017-12-28 | 2018-06-15 | 中国联合网络通信集团有限公司 | It can recognize that the semantic matching method and system of natural language description based on computer |
| CN110188983A (en) * | 2019-04-18 | 2019-08-30 | 定制加(深圳)科技有限公司 | A kind of business automation collaboration software robot device and method based on integrated architecture |
| CN112201252A (en) * | 2020-10-10 | 2021-01-08 | 南京机电职业技术学院 | Voice interaction learning and application system of express robot |
| CN115220400A (en) * | 2022-03-04 | 2022-10-21 | 弥费实业(上海)有限公司 | Wafer transfer-based supervisory control method, system, computer equipment and storage medium |
| CN115269159A (en) * | 2022-09-27 | 2022-11-01 | 苏州美集供应链管理股份有限公司 | Scheduling system and method based on artificial intelligence and edge calculation support |
| CN116011460A (en) * | 2023-02-13 | 2023-04-25 | 安徽龙鼎信息科技有限公司 | Logistics capacity matching method and system based on natural language processing |
| CN116303697A (en) * | 2023-05-18 | 2023-06-23 | 深圳鹏锐信息技术股份有限公司 | Model display system based on artificial intelligence |
Citations (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US5555169A (en) * | 1992-05-20 | 1996-09-10 | Hitachi, Ltd. | Computer system and method for converting a conversational statement to computer command language |
| CN1680937A (en) * | 2004-04-05 | 2005-10-12 | 华为技术有限公司 | Man-machine language generating system and method thereof |
| CN101178705A (en) * | 2007-12-13 | 2008-05-14 | 中国电信股份有限公司 | Free-running speech comprehend method and man-machine interactive intelligent system |
-
2015
- 2015-07-29 CN CN201510450049.1A patent/CN106407196A/en active Pending
Patent Citations (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US5555169A (en) * | 1992-05-20 | 1996-09-10 | Hitachi, Ltd. | Computer system and method for converting a conversational statement to computer command language |
| CN1680937A (en) * | 2004-04-05 | 2005-10-12 | 华为技术有限公司 | Man-machine language generating system and method thereof |
| CN101178705A (en) * | 2007-12-13 | 2008-05-14 | 中国电信股份有限公司 | Free-running speech comprehend method and man-machine interactive intelligent system |
Cited By (13)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN107423363A (en) * | 2017-06-22 | 2017-12-01 | 百度在线网络技术（北京）有限公司 | Art generation method, device, equipment and storage medium based on artificial intelligence |
| US10755148B2 (en) | 2017-06-22 | 2020-08-25 | Baidu Online Network Technology (Beijing) Co., Ltd. | Method and apparatus for generating a response based on artificial intelligence, and storage medium |
| CN107861879A (en) * | 2017-11-28 | 2018-03-30 | 成都视达科信息技术有限公司 | A kind of method and system for realizing software automated testing |
| CN108170679B (en) * | 2017-12-28 | 2021-09-03 | 中国联合网络通信集团有限公司 | Semantic matching method and system based on computer recognizable natural language description |
| CN108170679A (en) * | 2017-12-28 | 2018-06-15 | 中国联合网络通信集团有限公司 | It can recognize that the semantic matching method and system of natural language description based on computer |
| CN110188983A (en) * | 2019-04-18 | 2019-08-30 | 定制加(深圳)科技有限公司 | A kind of business automation collaboration software robot device and method based on integrated architecture |
| CN112201252A (en) * | 2020-10-10 | 2021-01-08 | 南京机电职业技术学院 | Voice interaction learning and application system of express robot |
| CN115220400A (en) * | 2022-03-04 | 2022-10-21 | 弥费实业(上海)有限公司 | Wafer transfer-based supervisory control method, system, computer equipment and storage medium |
| WO2023165189A1 (en) * | 2022-03-04 | 2023-09-07 | 弥费科技(上海)股份有限公司 | Wafer transfer supervision control method and system, computer device and storage medium |
| CN115269159A (en) * | 2022-09-27 | 2022-11-01 | 苏州美集供应链管理股份有限公司 | Scheduling system and method based on artificial intelligence and edge calculation support |
| CN116011460A (en) * | 2023-02-13 | 2023-04-25 | 安徽龙鼎信息科技有限公司 | Logistics capacity matching method and system based on natural language processing |
| CN116303697A (en) * | 2023-05-18 | 2023-06-23 | 深圳鹏锐信息技术股份有限公司 | Model display system based on artificial intelligence |
| CN116303697B (en) * | 2023-05-18 | 2023-08-08 | 深圳鹏锐信息技术股份有限公司 | Model display system based on artificial intelligence |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN106407196A (en) | Semantic analysis intelligent instruction robot applied to logistics management software | |
| US10262080B2 (en) | Enhanced search suggestion for personal information services | |
| EP1672537B1 (en) | Data semanticizer | |
| US10824798B2 (en) | Data collection for a new conversational dialogue system | |
| US8566699B2 (en) | Intent-based information processing and updates | |
| US10109297B2 (en) | Context-based virtual assistant conversations | |
| US7685116B2 (en) | Transparent search query processing | |
| US7672908B2 (en) | Intent-based information processing and updates in association with a service agent | |
| JP5040925B2 (en) | Information extraction rule creation support system, information extraction rule creation support method, and information extraction rule creation support program | |
| US20140156282A1 (en) | Method and system for controlling target applications based upon a natural language command string | |
| US20070203869A1 (en) | Adaptive semantic platform architecture | |
| US20150286943A1 (en) | Decision Making and Planning/Prediction System for Human Intention Resolution | |
| US20070106497A1 (en) | Natural language interface for driving adaptive scenarios | |
| WO2005010727A2 (en) | Extracting data from semi-structured text documents | |
| López et al. | From process models to chatbots | |
| CN101178705A (en) | Free-running speech comprehend method and man-machine interactive intelligent system | |
| AU2018299826B2 (en) | System and method for integrating message content into a target data processing device | |
| CN101490668A (en) | Reuse of available source data and localizations | |
| CN110399470A (en) | Conversation message processing | |
| JP2020135135A (en) | Dialog content creation assisting method and system | |
| D’Abrera et al. | A formally verified cut-elimination procedure for linear nested sequents for tense logic | |
| CN108615529A (en) | A kind of control method, device and domestic robot based on semantics recognition | |
| CN109891410B (en) | Data collection for new session dialog systems | |
| US20240126813A1 (en) | Global Data Management Framework | |
| CN116956828A (en) | Document text input method and device, storage medium and electronic equipment |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| C06 | Publication | ||
| PB01 | Publication | ||
| C10 | Entry into substantive examination | ||
| SE01 | Entry into force of request for substantive examination | ||
| WD01 | Invention patent application deemed withdrawn after publication | ||
| WD01 | Invention patent application deemed withdrawn after publication |
Application publication date: 20170215 |