CN109325106A - A kind of U.S. chat robots intension recognizing method of doctor and device - Google Patents
A kind of U.S. chat robots intension recognizing method of doctor and device Download PDFInfo
- Publication number
- CN109325106A CN109325106A CN201810856548.4A CN201810856548A CN109325106A CN 109325106 A CN109325106 A CN 109325106A CN 201810856548 A CN201810856548 A CN 201810856548A CN 109325106 A CN109325106 A CN 109325106A
- Authority
- CN
- China
- Prior art keywords
- intention
- text data
- intention assessment
- result
- classifier
- 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
- 238000000034 method Methods 0.000 title claims abstract description 48
- 241001378783 Myrmecocichla nigra Species 0.000 title claims abstract description 21
- 238000012549 training Methods 0.000 claims abstract description 55
- 230000011218 segmentation Effects 0.000 claims description 13
- 238000012545 processing Methods 0.000 claims description 12
- 238000012360 testing method Methods 0.000 claims description 12
- 230000000694 effects Effects 0.000 claims description 11
- 239000000463 material Substances 0.000 claims description 10
- 238000007637 random forest analysis Methods 0.000 claims description 9
- 238000003066 decision tree Methods 0.000 claims description 7
- 235000013399 edible fruits Nutrition 0.000 claims description 5
- 238000011156 evaluation Methods 0.000 claims description 4
- 239000000284 extract Substances 0.000 claims description 4
- 238000013507 mapping Methods 0.000 claims description 4
- 208000024891 symptom Diseases 0.000 claims description 3
- 230000008859 change Effects 0.000 claims description 2
- 238000007781 pre-processing Methods 0.000 claims 1
- 230000008569 process Effects 0.000 description 8
- 238000005516 engineering process Methods 0.000 description 4
- 238000013473 artificial intelligence Methods 0.000 description 3
- 210000000744 eyelid Anatomy 0.000 description 3
- 238000012986 modification Methods 0.000 description 3
- 230000004048 modification Effects 0.000 description 3
- 230000003466 anti-cipated effect Effects 0.000 description 2
- 238000012544 monitoring process Methods 0.000 description 2
- 230000004044 response Effects 0.000 description 2
- 241001269238 Data Species 0.000 description 1
- 230000001133 acceleration Effects 0.000 description 1
- 230000032683 aging Effects 0.000 description 1
- 230000003796 beauty Effects 0.000 description 1
- 230000009286 beneficial effect Effects 0.000 description 1
- 238000004422 calculation algorithm Methods 0.000 description 1
- 238000006243 chemical reaction Methods 0.000 description 1
- 238000010586 diagram Methods 0.000 description 1
- 230000007717 exclusion Effects 0.000 description 1
- 230000036541 health Effects 0.000 description 1
- 230000000750 progressive effect Effects 0.000 description 1
- 230000001737 promoting effect Effects 0.000 description 1
- 238000007493 shaping process Methods 0.000 description 1
- 231100000033 toxigenic Toxicity 0.000 description 1
- 230000001551 toxigenic effect Effects 0.000 description 1
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F40/00—Handling natural language data
- G06F40/20—Natural language analysis
- G06F40/205—Parsing
- G06F40/216—Parsing using statistical methods
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F18/00—Pattern recognition
- G06F18/20—Analysing
- G06F18/24—Classification techniques
- G06F18/243—Classification techniques relating to the number of classes
- G06F18/24323—Tree-organised classifiers
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F40/00—Handling natural language data
- G06F40/20—Natural language analysis
- G06F40/279—Recognition of textual entities
- G06F40/289—Phrasal analysis, e.g. finite state techniques or chunking
-
- G—PHYSICS
- G10—MUSICAL INSTRUMENTS; ACOUSTICS
- G10L—SPEECH ANALYSIS TECHNIQUES OR SPEECH SYNTHESIS; SPEECH RECOGNITION; SPEECH OR VOICE PROCESSING TECHNIQUES; SPEECH OR AUDIO CODING OR DECODING
- G10L15/00—Speech recognition
- G10L15/06—Creation of reference templates; Training of speech recognition systems, e.g. adaptation to the characteristics of the speaker's voice
- G10L15/063—Training
-
- G—PHYSICS
- G10—MUSICAL INSTRUMENTS; ACOUSTICS
- G10L—SPEECH ANALYSIS TECHNIQUES OR SPEECH SYNTHESIS; SPEECH RECOGNITION; SPEECH OR VOICE PROCESSING TECHNIQUES; SPEECH OR AUDIO CODING OR DECODING
- G10L15/00—Speech recognition
- G10L15/08—Speech classification or search
-
- G—PHYSICS
- G10—MUSICAL INSTRUMENTS; ACOUSTICS
- G10L—SPEECH ANALYSIS TECHNIQUES OR SPEECH SYNTHESIS; SPEECH RECOGNITION; SPEECH OR VOICE PROCESSING TECHNIQUES; SPEECH OR AUDIO CODING OR DECODING
- G10L15/00—Speech recognition
- G10L15/26—Speech to text systems
Landscapes
- Engineering & Computer Science (AREA)
- Physics & Mathematics (AREA)
- Theoretical Computer Science (AREA)
- Computational Linguistics (AREA)
- Health & Medical Sciences (AREA)
- Audiology, Speech & Language Pathology (AREA)
- Artificial Intelligence (AREA)
- General Physics & Mathematics (AREA)
- Human Computer Interaction (AREA)
- Acoustics & Sound (AREA)
- Multimedia (AREA)
- General Engineering & Computer Science (AREA)
- General Health & Medical Sciences (AREA)
- Data Mining & Analysis (AREA)
- Probability & Statistics with Applications (AREA)
- Life Sciences & Earth Sciences (AREA)
- Bioinformatics & Cheminformatics (AREA)
- Bioinformatics & Computational Biology (AREA)
- Computer Vision & Pattern Recognition (AREA)
- Evolutionary Biology (AREA)
- Evolutionary Computation (AREA)
- Management, Administration, Business Operations System, And Electronic Commerce (AREA)
Abstract
The invention discloses a kind of U.S. chat robots intension recognizing method of doctor and devices, by obtaining training corpus, according in training corpus text data and corresponding intention labels be trained to obtain intention assessment classifier；When prediction, current text data to be identified is pre-processed and is inputted in trained intention assessment classifier, the intention assessment classifier returns to intention assessment result；And further judge whether the intention assessment result is effective as a result, if so, exporting the current corresponding intention assessment result of text data；If it is not, the corresponding effective intention assessment result of the text data for then exporting one group；To judge visitor's intent features by information above, so that intention assessment result is more acurrate, is made with auxiliary robot and correctly respond decision.
Description
Technical field
The present invention relates to field of artificial intelligence, the U.S. chat robots intension recognizing method of especially a kind of doctor and its answer
With the device of this method, suitable for medical advice and the consulting of shaping and beauty etc..
Background technique
The rise of artificial intelligence technology in recent years, so that medical industries are merged with the acceleration that starts of artificial intelligence.Another party
Face, with the awakening of social progress and people's health consciousness, the continuous aggravation of the problem of an aging population, people are for promoting medical treatment
Technology extends human longevity, enhances healthy demand also more urgently.And in practicing there is medical resource distribution it is uneven with
And medical worker's toxigenic capacity it is excessively high the problems such as.
First generation robot carries out template matching based on traditional rule system, and brought problem is on the one hand to need
All possible rule template is summarized, once there is the content that rule template is not covered, medical robot just can not be according to right
The information just sent makes corresponding response；On the other hand, if robot has included a large amount of rule template, although matching rule
Range solve to a certain extent, but efficiency will because of template number increase and have a greatly reduced quality.
Second generation robot introduces FAQ and semantic similarity concept, compares first generation robot, solving needs to arrange
A large amount of rule template improves accuracy and matching efficiency problem.Although the solution obtained in semantic matches, context are asked
Topic is not handled, such as " apple how much ", may there are two types of the meanings without reference to contextual information: first is that apple
(fruit) how much, second is that apple (mobile phone) how much；So that robot can not provide correct response.
Summary of the invention
The present invention is logical to solve the above problems, provide a kind of U.S. chat robots intension recognizing method of doctor and device
Information above is crossed to judge visitor's intent features, so that intention assessment result is more acurrate, is made with auxiliary robot and correctly being returned
Answer decision.
To achieve the above object, the technical solution adopted by the present invention are as follows:
A kind of U.S. chat robots intension recognizing method of doctor comprising following steps:
A. obtain training corpus, according in training corpus text data and corresponding intention labels be trained and anticipated
Figure recognition classifier；
B. current text data to be identified is pre-processed and is inputted in trained intention assessment classifier, institute
It states intention assessment classifier and returns to intention assessment result；
C. judge whether the intention assessment result is effective as a result, if so, the current text data of output is corresponding
Intention assessment result；If it is not, the corresponding effective intention assessment result of the text data for then exporting one group.
Preferably, when being trained in the step a to the text data in the training corpus or the step
When carrying out Forecasting recognition to the current text data or upper one group of text data in rapid b, further to described
Text data carries out word segmentation processing and vectorization handles to obtain term vector, and the term vector is inputted the intention assessment classifier
In be trained or Forecasting recognition.
Further, the word segmentation processing is using in IKAnalyzer segmenter, ICTCLAS segmentation methods, Ansj
The method of text participle or stammerer participle is segmented；The vectorization processing is using word2vec model that participle is good
Text data is converted into term vector.
Preferably, in the step a, the training of the intention assessment classifier further comprises:
A1. training corpus described in cutting extracts therein 10% and is used as testing material, and residue 90% is used as training corpus；
A2. it will be trained in described 90% training corpus input Random Forest model, the Random Forest model packet
Include multiple decision trees, voted by each decision tree classification results, and according to voting results export intention assessment as a result,
Obtain intention assessment classifier；
A3. the intention assessment classifier is tested using described 10% testing material, and by extensive index
The intention assessment classifier is assessed；
A4. output meets the intention assessment classifier of evaluation index.
Preferably, in the step a, in the training corpus text data and corresponding intention labels instruct
Before white silk, digital mapping first is carried out to the intention labels, the intention labels in the training corpus are mapped to unique correspondence
Number.
Preferably, in the step b, before being pre-processed to current text data to be identified, first to described
Text data carries out word counting, judges the text quantity in the text data whether less than 3, if so, directly returning to nothing
The intention assessment result of effect；Otherwise, word segmentation processing further carried out to the text data and vectorization handle to obtain word to
Amount, and the term vector is inputted in the intention assessment classifier and carries out Forecasting recognition.
It preferably, further include step d, according to the effective intention assessment of output as a result, corresponding answer is returned to visit
Visitor.
Preferably, the effective intention assessment is as a result, refer to intention corresponding to the current text data
Label；The invalid intention assessment is as a result, refer to without other identification knots other than output result or the intention labels
Fruit.
Preferably, the intention labels include: description symptom, consulting price, consulting address, determine subscription time, consult
Inquiry project, consulting course for the treatment of curative effect, the preferential activity of consulting, consulting side effect, consulting materials instrument treatment method, consulting correspondent party
Formula.
Corresponding, the present invention also provides a kind of U.S. chat robots intention assessment devices of doctor comprising:
It is intended to training module, for obtaining training corpus, according to the text data and corresponding intention mark in training corpus
Label are trained to obtain intention assessment classifier；
Intention Anticipation module, for being pre-processed to current text data to be identified and inputting trained intention
In recognition classifier, the intention assessment classifier returns to intention assessment result；
Recognition result judgment module, for judging whether the intention assessment result is effective as a result, if so, output is worked as
The corresponding intention assessment result of preceding text data；If it is not, the corresponding effective intention of the text data for then exporting one group is known
Other result.
The beneficial effects of the present invention are:
(1) present invention is by by the effective of the intention assessment result of current text data and upper one group of text data
Intention assessment result is associated, visitor's intent features is judged by information above, so that intention assessment result is more acurrate, with auxiliary
It helps robot to make and correctly responds decision；
(2) intention labels are mapped as unique corresponding number by the present invention, consequently facilitating computer is in training and predicts
Accurate quickly identification, improves the accuracy rate and computational efficiency of algorithm in journey；
(3) before the present invention pre-processes current text data to be identified, first the text data is carried out
Word counting, if the text quantity in the text data less than 3, directly returns to invalid intention assessment as a result, to letter
Change calculating process, improves computational efficiency.
Detailed description of the invention
The drawings described herein are used to provide a further understanding of the present invention, constitutes a part of the invention, this hair
Bright illustrative embodiments and their description are used to explain the present invention, and are not constituted improper limitations of the present invention.In the accompanying drawings:
Fig. 1 is a kind of general flow chart for curing U.S. chat robots intension recognizing method of the present invention；
Fig. 2 is a kind of structural schematic diagram for curing U.S. chat robots intention assessment device of the present invention.
Specific embodiment
In order to be clearer and more clear technical problems, technical solutions and advantages to be solved, tie below
Closing accompanying drawings and embodiments, the present invention will be described in further detail.It should be appreciated that specific embodiment described herein is only used
To explain the present invention, it is not intended to limit the present invention.
As shown in Figure 1, a kind of doctor's U.S. chat robots intension recognizing method of the invention comprising following steps:
A. obtain training corpus, according in training corpus text data and corresponding intention labels be trained and anticipated
Figure recognition classifier；
B. current text data to be identified is pre-processed and is inputted in trained intention assessment classifier, institute
It states intention assessment classifier and returns to intention assessment result；
C. judge whether the intention assessment result is effective as a result, if so, the current text data of output is corresponding
Intention assessment result；If it is not, the corresponding effective intention assessment result of the text data for then exporting one group；
D. according to the effective intention assessment of output as a result, corresponding answer is returned to visitor.
Wherein, the text data in the training corpus may be from the dialog history record of the visitor collected, described
Dialog history record includes voice record or text entry；The current text data to be identified, upper one group of textual data
According to may be from text data made of the voice data conversion of visitor's input, or arise directly from the text of visitor's data
Data.
It is right when being trained in the step a to the text data in the training corpus or in the step b
When the current text data or upper one group of text data carry out Forecasting recognition, further to the text data
Pre-processed: the pretreatment includes that word segmentation processing and vectorization are handled, to obtain term vector, and the term vector is inputted
It is trained in the intention assessment classifier or Forecasting recognition.In the present embodiment, the word segmentation processing be can be used
The method that IKAnalyzer segmenter, ICTCLAS segmentation methods, Ansj Chinese word segmentation or stammerer segment is segmented, this implementation
Using stammerer participle technique in example；The vectorization processing is to be turned the text data segmented using word2vec model
Turn to term vector.
Wherein, the stammerer participle technique realizes centering sentence and is split by word granularity, supports three kinds points
Word mode: first is that accurate model, it is intended to sentence most accurately be cut, text analyzing is suitble to；Second is that syntype, institute in sentence
What is had can all scan at the word of word, and speed is very fast, but not can solve ambiguity；Third is that search engine mode,
On the basis of accurate model, to long word cutting again, recall rate is improved, is segmented suitable for search engine.It can also support simultaneously
Traditional font participle and Custom Dictionaries.
The Word2Vec word embedded technology considers the context relation between word, each word is mapped to one
A vector carries out vectorization to text data to realize, obtains term vector.Word2vec is broadly divided into CBOW
(Continuous Bag of Words) and Skip-Gram both of which.CBOW is to speculate target words from original statement；And
Skip-Gram is exactly the opposite, is to deduce original statement from target words.CBOW is proper to toy data base, and Skip-
Gram is performed better than in large-scale corpus, can be selected according to actual needs.
In the present embodiment, the effective intention assessment is as a result, refer to corresponding to the current text data
Intention labels；The invalid intention assessment is as a result, refer to without other knowledges other than output result or the intention labels
Other result.Wherein, the intention labels include: description symptom, consulting price, consulting address, determine subscription time, consulting item
Mesh, consulting course for the treatment of curative effect, the preferential activity of consulting, consulting side effect, consulting materials instrument treatment method, consulting contact method.Such as
Shown in table 1:
Table 1- visitor is intended to list
In the present embodiment, in the step a, the training of the intention assessment classifier further comprises:
A1. training corpus described in cutting extracts therein 10% and is used as testing material, and residue 90% is used as training corpus；
A2. it will be trained in described 90% training corpus input Random Forest model, the Random Forest model packet
Include multiple decision trees, voted by each decision tree classification results, and according to voting results export intention assessment as a result,
Obtain intention assessment classifier；
A3. the intention assessment classifier is tested using described 10% testing material, and by extensive index
The intention assessment classifier is assessed；
A4. output meets the intention assessment classifier of evaluation index.
Wherein, random forest (Random Forest) be one include multiple decision trees classifier, and its output
Classification be by set the classification of output individually mode depending on.
Also, in the step a, in the training corpus text data and corresponding intention labels be trained
Before, digital mapping first is carried out to the intention labels, the intention labels in the training corpus is mapped to unique corresponding
Number.
The training process of the present embodiment is exemplified below:
1) prepare training corpus, including text data and the corresponding intention labels of text data, as shown in table 2:
List is expected in table 2- training
2) text divides, and text is divided into word list using participle technique, as shown in table 3:
Table 3- segments list
3) text vector is term vector to the word2vec model conversation that the text segmented utilizes pre-training good, such as
Shown in table 4:
Table 4- term vector list
4) intention labels map, and the intention labels in training corpus are mapped to number, as shown in table 5:
Table 5- intention labels mapping table
5) cutting training corpus extracts 10% training corpus as testing material (training set), for monitoring classifier effect
Fruit, residue 90% are used as training corpus (training set)；
Table 6- training set
Table 7- test set
| Intention labels | Text data | Term vector | Number |
| Seek advice from price | Price/how to calculate | [0.8, 0.05, 0.09, 0.01] | 0 |
| Consulting address | You/hospital/at which | [0.08, 0.5, 0.1, 0.01] | 1 |
6) training set of table 6 is put into Random Forest model and is trained, obtains classifier by classifier training；
7) model evaluation assesses classifier by extensive index (including accuracy), that is, with the test of table 7
The data of collection are input to trained model file and are predicted, obtain the prediction result of test set, extensive for monitoring model
Ability；
The prediction result of table 8- test set
According to prediction result, Accuracy value is 100%, illustrates that all data all predicts correctly in test set, after can general
Model output.
8) output category device.
It is first right before being pre-processed to current text data to be identified in the step b in the present embodiment
The text data carries out word counting, judges the text quantity in the text data whether less than 3, if so, directly returning
Return invalid intention assessment result；Otherwise, word segmentation processing further is carried out to the text data and vectorization handles to obtain
Term vector, and the term vector is inputted in the intention assessment classifier and carries out Forecasting recognition.
The prediction process of the present embodiment is exemplified below:
1) text data of visitor's input is obtained；
Table 9- current text data to be identified
| Text data |
| You are good |
| Could you tell me? |
| I thinks the price that double-edged eyelid are cut under consulting |
| Can the better address of your hospitals tell that I am lower? |
2) text determines, judges the Chinese character quantity in the text data whether less than 3, if so, exporting "None", otherwise
It continues to execute in next step；
Table 10- text decision table
| Text data | Prediction result |
| You are good | Nothing |
| Could you tell me? | |
| I thinks the price that double-edged eyelid are cut under consulting | |
| Can the better address of your hospitals tell that I am lower? |
3) text divides, and is divided into list to text using participle technique, as shown in table 11:
Table 11- text divides table
4) text vector is term vector to the word2vec model conversation that the text segmented utilizes pre-training good, such as
Shown in table 12:
Table 12- term vector list
5) Intention Anticipation, using term vector as input, into trained classifier, classifier can return to intention assessment knot
Fruit, as shown in table 13:
Table 13- Intention Anticipation table
6) corresponding answer is returned to visitor according to intention assessment result.
Such as have following session operational scenarios, when the visitor the problem of are as follows: " 3 points of Korean style? " if not carrying out intention assessment, machine
Device people may be due to that can not understand that error return is replied or directly given to this meaning without can be carried out.And it is incorporated in the present invention
After intention assessment, before robot can detect visitor refer to consulting price, and then deducibility visitor may inquire be
Price then returns to corresponding pricing information.
Chat process is exemplified below:
1) visitor: you are good
2) visitor: could you tell me?
3) robot: { intention assessment result: " nothing " } you are good,, can what may I ask help you?
4) visitor: I thinks the price that double-edged eyelid are cut under consulting
5) robot: intention assessment result: ["None", " seeks advice from price "] } probably in 1000-2000 range
6) visitor: 3 points of Korean style?
7) robot: intention assessment result: ["None", " seeks advice from price ", " nothing "] 3 points of Korean style may slightly expensive one
Point, approximately 5000 or so.
8) robot: intention assessment result: ["None", " seek advice from price ", " nothing "] you however come us when free this,
Under we arrange the doctor of profession to instruct to you.
Can 9) visitor: the better address of your hospitals tell that I am lower?
10) robot: intention assessment result: ["None", " seek advice from price ", " nothing ", " consulting address "] address xx save
The area xx of the city the xx mansion the xx building x.
({ ... } is to hide content, is the ordered list for carrying out intention assessment to the reply of visitor above, actually chatted
Journey is not shown)
Using the intent of the present invention recognition methods, the case where carrying out intention assessment to every text data above
Under, the intention assessment of current text data is judged as a result, if it is " nothing ", returns to the meaning of an intention assessment non-' without '
Otherwise figure recognition result directly returns to the intention assessment of current text data as a result, and according to intention assessment result robot
Provide corresponding reply.In the present embodiment, in text data above-mentioned, two text datas, meaning are hereinbefore generated
Consulting price that figure recognition result is sequentially respectively ["None", " "], it is assumed that the current text data of visitor are as follows: " Korean style
3 points? ", it is intended that recognition result are as follows: nothing.Then the intention assessment of intention assessment non-' without ' is returned as a result, i.e. upper one
The intention assessment result of the non-nothing of item: consulting price.Then the intention assessment according to consulting price, robot provide corresponding reply.
As shown in Fig. 2, the present invention also provides a kind of U.S. chat robots intention assessment devices of doctor comprising:
It is intended to training module, for obtaining training corpus, according to the text data and corresponding intention mark in training corpus
Label are trained to obtain intention assessment classifier；
Intention Anticipation module, for being pre-processed to current text data to be identified and inputting trained intention
In recognition classifier, the intention assessment classifier returns to intention assessment result；
Recognition result judgment module, for judging whether the intention assessment result is effective as a result, if so, output is worked as
The corresponding intention assessment result of preceding text data；If it is not, the corresponding effective intention of the text data for then exporting one group is known
Other result.
It should be noted that all the embodiments in this specification are described in a progressive manner, each embodiment weight
Point explanation is the difference from other embodiments, and the same or similar parts between the embodiments can be referred to each other.
For device embodiment, since it is basically similar to the method embodiment, so being described relatively simple, related place referring to
The part of embodiment of the method illustrates.
Also, herein, the terms "include", "comprise" or its any other variant are intended to the packet of nonexcludability
Contain, so that the process, method, article or equipment for including a series of elements not only includes those elements, but also including
Other elements that are not explicitly listed, or further include for elements inherent to such a process, method, article, or device.
In the absence of more restrictions, the element limited by sentence "including a ...", it is not excluded that including the element
Process, method, article or equipment in there is also other identical elements.In addition, those of ordinary skill in the art can manage
Solution realizes that all or part of the steps of above-described embodiment may be implemented by hardware, and can also be instructed by program relevant
Hardware is completed, and the program can store in a kind of computer readable storage medium, and storage medium mentioned above can be with
It is read-only memory, disk or CD etc..
The preferred embodiment of the present invention has shown and described in above description, it should be understood that the present invention is not limited to this paper institute
The form of disclosure, should not be regarded as an exclusion of other examples, and can be used for other combinations, modifications, and environments, and energy
Enough in this paper invented the scope of the idea, modifications can be made through the above teachings or related fields of technology or knowledge.And people from this field
The modifications and changes that member is carried out do not depart from the spirit and scope of the present invention, then all should be in the protection of appended claims of the present invention
In range.
Claims (10)
1. a kind of U.S. chat robots intension recognizing method of doctor, which comprises the following steps:
A. obtain training corpus, according in training corpus text data and corresponding intention labels be trained to obtain intention and know
Other classifier；
B. current text data to be identified is pre-processed and is inputted in trained intention assessment classifier, the meaning
Figure recognition classifier returns to intention assessment result；
C. judge whether the intention assessment result is effective as a result, if so, exporting the current corresponding intention of text data
Recognition result；If it is not, the corresponding effective intention assessment result of the text data for then exporting one group.
2. the U.S. chat robots intension recognizing method of a kind of doctor according to claim 1, it is characterised in that: the step
To the current textual data when being trained in a to the text data in the training corpus or in the step b
According to or upper one group of text data carry out Forecasting recognition when, word segmentation processing and vector further are carried out to the text data
Change handles to obtain term vector, and the term vector inputted in the intention assessment classifier and is trained or Forecasting recognition.
3. the U.S. chat robots intension recognizing method of a kind of doctor according to claim 2, it is characterised in that: the participle
Processing is carried out using the method for IKAnalyzer segmenter, ICTCLAS segmentation methods, Ansj Chinese word segmentation or stammerer participle
Participle；The vectorization processing, is to convert term vector for the text data segmented using word2vec model.
4. the U.S. chat robots intension recognizing method of a kind of doctor according to claim 1, it is characterised in that: the step
In a, the training of the intention assessment classifier further comprises:
A1. training corpus described in cutting extracts therein 10% and is used as testing material, and residue 90% is used as training corpus；
A2. it will be trained in described 90% training corpus input Random Forest model, the Random Forest model includes more
A decision tree votes to classification results by each decision tree, and exports intention assessment as a result, obtaining according to voting results
Intention assessment classifier；
A3. the intention assessment classifier is tested using described 10% testing material, and by extensive index to institute
Intention assessment classifier is stated to be assessed；
A4. output meets the intention assessment classifier of evaluation index.
5. the U.S. chat robots intension recognizing method of a kind of doctor according to claim 1, it is characterised in that: the step
In a, to the text data in the training corpus and before corresponding intention labels are trained, first to the intention labels into
Intention labels in the training corpus are mapped to unique corresponding number by the mapping of line number word.
6. the U.S. chat robots intension recognizing method of a kind of doctor according to claim 1, it is characterised in that: the step
In b, before pre-processing to current text data to be identified, word counting, judgement first are carried out to the text data
Whether the text quantity in the text data is less than 3, if so, directly returning to invalid intention assessment result；Otherwise, into one
It walks and the text data progress word segmentation processing and vectorization is handled to obtain term vector, and the term vector is inputted into the intention
Forecasting recognition is carried out in recognition classifier.
7. the U.S. chat robots intension recognizing method of a kind of doctor according to any one of claims 1 to 6, it is characterised in that:
It further include step d, according to the effective intention assessment of output as a result, corresponding answer is returned to visitor.
8. the U.S. chat robots intension recognizing method of a kind of doctor according to any one of claims 1 to 6, it is characterised in that:
The effective intention assessment is as a result, refer to intention labels corresponding to the current text data；Described is invalid
Intention assessment as a result, refer to without output result or the intention labels other than other recognition results.
9. the U.S. chat robots intension recognizing method of a kind of doctor according to any one of claims 1 to 6, it is characterised in that:
The intention labels include: description symptom, consulting price, consulting address, determine subscription time, consulting item, the consulting course for the treatment of
Curative effect, the preferential activity of consulting, consulting side effect, consulting materials instrument treatment method, consulting contact method.
10. a kind of U.S. chat robots intention assessment device of doctor characterized by comprising
Be intended to training module, for obtaining training corpus, according in training corpus text data and corresponding intention labels into
Row training obtains intention assessment classifier；
Intention Anticipation module, for being pre-processed to current text data to be identified and inputting trained intention assessment
In classifier, the intention assessment classifier returns to intention assessment result；
Recognition result judgment module, for judging whether the intention assessment result is effective as a result, if so, output is current
The corresponding intention assessment result of text data；If it is not, the corresponding effective intention assessment knot of the text data for then exporting one group
Fruit.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201810856548.4A CN109325106A (en) | 2018-07-31 | 2018-07-31 | A kind of U.S. chat robots intension recognizing method of doctor and device |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201810856548.4A CN109325106A (en) | 2018-07-31 | 2018-07-31 | A kind of U.S. chat robots intension recognizing method of doctor and device |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN109325106A true CN109325106A (en) | 2019-02-12 |
Family
ID=65263937
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201810856548.4A Pending CN109325106A (en) | 2018-07-31 | 2018-07-31 | A kind of U.S. chat robots intension recognizing method of doctor and device |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN109325106A (en) |
Cited By (11)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN110136699A (en) * | 2019-07-10 | 2019-08-16 | 南京硅基智能科技有限公司 | A kind of intension recognizing method based on text similarity |
| CN110147445A (en) * | 2019-04-09 | 2019-08-20 | 平安科技（深圳）有限公司 | Intension recognizing method, device, equipment and storage medium based on text classification |
| CN110162775A (en) * | 2019-03-11 | 2019-08-23 | 腾讯科技（深圳）有限公司 | Determine the method, apparatus and computer equipment of intention assessment accuracy |
| CN110347789A (en) * | 2019-06-14 | 2019-10-18 | 平安科技（深圳）有限公司 | Text is intended to intelligent method for classifying, device and computer readable storage medium |
| CN111177377A (en) * | 2019-12-20 | 2020-05-19 | 厦门快商通科技股份有限公司 | Medical and beauty training data evaluation method, device, equipment and storage medium |
| CN111221955A (en) * | 2020-01-09 | 2020-06-02 | 厦门快商通科技股份有限公司 | Visitor intention data pre-extraction method and system based on small amount of data |
| CN111611391A (en) * | 2020-06-17 | 2020-09-01 | 厦门快商通科技股份有限公司 | Method, device, equipment and storage medium for classifying conversation |
| WO2020211008A1 (en) * | 2019-04-17 | 2020-10-22 | 深圳市欢太科技有限公司 | Speech recognition method and apparatus, storage medium and electronic device |
| CN113762372A (en) * | 2021-08-30 | 2021-12-07 | 中国电子科技集团公司第十五研究所 | Method and device for identifying organization members in instant messaging information |
| WO2022121163A1 (en) * | 2020-12-11 | 2022-06-16 | 平安科技（深圳）有限公司 | User behavior tendency identification method, apparatus, and device, and storage medium |
| US11423227B2 (en) | 2020-02-13 | 2022-08-23 | International Business Machines Corporation | Weak supervised abnormal entity detection |
Citations (5)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN104598532A (en) * | 2014-12-29 | 2015-05-06 | 中国联合网络通信有限公司广东省分公司 | Information processing method and device |
| CN106325113A (en) * | 2015-06-26 | 2017-01-11 | 北京贝虎机器人技术有限公司 | Robot control engine and system |
| CN107175674A (en) * | 2017-05-23 | 2017-09-19 | 上海棠棣信息科技股份有限公司 | A kind of robot interactive method and device |
| KR20170107231A (en) * | 2016-03-15 | 2017-09-25 | 서울대학교산학협력단 | Upper limb rehabilitation robot module and upper limb rehabilitation robot system for precision control by image processing |
| CN107562816A (en) * | 2017-08-16 | 2018-01-09 | 深圳狗尾草智能科技有限公司 | User view automatic identifying method and device |
-
2018
- 2018-07-31 CN CN201810856548.4A patent/CN109325106A/en active Pending
Patent Citations (5)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN104598532A (en) * | 2014-12-29 | 2015-05-06 | 中国联合网络通信有限公司广东省分公司 | Information processing method and device |
| CN106325113A (en) * | 2015-06-26 | 2017-01-11 | 北京贝虎机器人技术有限公司 | Robot control engine and system |
| KR20170107231A (en) * | 2016-03-15 | 2017-09-25 | 서울대학교산학협력단 | Upper limb rehabilitation robot module and upper limb rehabilitation robot system for precision control by image processing |
| CN107175674A (en) * | 2017-05-23 | 2017-09-19 | 上海棠棣信息科技股份有限公司 | A kind of robot interactive method and device |
| CN107562816A (en) * | 2017-08-16 | 2018-01-09 | 深圳狗尾草智能科技有限公司 | User view automatic identifying method and device |
Cited By (14)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN110162775A (en) * | 2019-03-11 | 2019-08-23 | 腾讯科技（深圳）有限公司 | Determine the method, apparatus and computer equipment of intention assessment accuracy |
| CN110147445A (en) * | 2019-04-09 | 2019-08-20 | 平安科技（深圳）有限公司 | Intension recognizing method, device, equipment and storage medium based on text classification |
| WO2020211008A1 (en) * | 2019-04-17 | 2020-10-22 | 深圳市欢太科技有限公司 | Speech recognition method and apparatus, storage medium and electronic device |
| CN113348502A (en) * | 2019-04-17 | 2021-09-03 | 深圳市欢太科技有限公司 | Voice recognition method, voice recognition device, storage medium and electronic equipment |
| CN110347789A (en) * | 2019-06-14 | 2019-10-18 | 平安科技（深圳）有限公司 | Text is intended to intelligent method for classifying, device and computer readable storage medium |
| WO2020248366A1 (en) * | 2019-06-14 | 2020-12-17 | 平安科技（深圳）有限公司 | Text intention intelligent classification method and device, and computer-readable storage medium |
| CN110136699A (en) * | 2019-07-10 | 2019-08-16 | 南京硅基智能科技有限公司 | A kind of intension recognizing method based on text similarity |
| CN111177377A (en) * | 2019-12-20 | 2020-05-19 | 厦门快商通科技股份有限公司 | Medical and beauty training data evaluation method, device, equipment and storage medium |
| CN111221955A (en) * | 2020-01-09 | 2020-06-02 | 厦门快商通科技股份有限公司 | Visitor intention data pre-extraction method and system based on small amount of data |
| US11423227B2 (en) | 2020-02-13 | 2022-08-23 | International Business Machines Corporation | Weak supervised abnormal entity detection |
| CN111611391A (en) * | 2020-06-17 | 2020-09-01 | 厦门快商通科技股份有限公司 | Method, device, equipment and storage medium for classifying conversation |
| CN111611391B (en) * | 2020-06-17 | 2022-08-23 | 厦门快商通科技股份有限公司 | Method, device, equipment and storage medium for classifying conversation |
| WO2022121163A1 (en) * | 2020-12-11 | 2022-06-16 | 平安科技（深圳）有限公司 | User behavior tendency identification method, apparatus, and device, and storage medium |
| CN113762372A (en) * | 2021-08-30 | 2021-12-07 | 中国电子科技集团公司第十五研究所 | Method and device for identifying organization members in instant messaging information |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN109325106A (en) | A kind of U.S. chat robots intension recognizing method of doctor and device | |
| CN108427707B (en) | Man-machine question and answer method, device, computer equipment and storage medium | |
| US11468241B2 (en) | Techniques to add smart device information to machine learning for increased context | |
| CN108595696A (en) | A kind of human-computer interaction intelligent answering method and system based on cloud platform | |
| CN111444344B (en) | Entity classification method, entity classification device, computer equipment and storage medium | |
| CN107330011A (en) | The recognition methods of the name entity of many strategy fusions and device | |
| CN110727779A (en) | Question-answering method and system based on multi-model fusion | |
| CN106709754A (en) | Power user grouping method based on text mining | |
| CN107038480A (en) | A kind of text sentiment classification method based on convolutional neural networks | |
| CN108984655B (en) | Intelligent customer service guiding method for customer service robot | |
| CN111309887B (en) | Method and system for training text key content extraction model | |
| CN110033281A (en) | A kind of method and device that intelligent customer service is converted to artificial customer service | |
| CN108052505A (en) | Text emotion analysis method and device, storage medium, terminal | |
| CN110781204B (en) | Identification information determining method, device, equipment and storage medium of target object | |
| CN110019729B (en) | Intelligent question-answering method, storage medium and terminal | |
| CN110472057B (en) | Topic label generation method and device | |
| US20210073628A1 (en) | Deep neural network training method and apparatus, and computer device | |
| CN111368096A (en) | Knowledge graph-based information analysis method, device, equipment and storage medium | |
| CN108920603B (en) | Customer service guiding method based on customer service machine model | |
| CN110222192A (en) | Corpus method for building up and device | |
| CN108509588B (en) | Lawyer evaluation method and recommendation method based on big data | |
| CN117763126A (en) | Knowledge retrieval method, device, storage medium and apparatus | |
| CN112527985A (en) | Unknown problem processing method, device, equipment and medium | |
| CN115905187B (en) | Intelligent proposition system oriented to cloud computing engineering technician authentication | |
| CN109284376A (en) | Cross-cutting news data sentiment analysis method based on domain-adaptive |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| PB01 | Publication | ||
| PB01 | Publication | ||
| SE01 | Entry into force of request for substantive examination | ||
| SE01 | Entry into force of request for substantive examination | ||
| RJ01 | Rejection of invention patent application after publication | ||
| RJ01 | Rejection of invention patent application after publication |
Application publication date: 20190212 |