CN104331523B - A kind of question sentence search method based on conceptual object model - Google Patents
A kind of question sentence search method based on conceptual object model Download PDFInfo
- Publication number
- CN104331523B CN104331523B CN201410713510.3A CN201410713510A CN104331523B CN 104331523 B CN104331523 B CN 104331523B CN 201410713510 A CN201410713510 A CN 201410713510A CN 104331523 B CN104331523 B CN 104331523B
- Authority
- CN
- China
- Prior art keywords
- question sentence
- concept
- attribute
- relation
- similarity
- 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.)
- Active
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
Landscapes
- Engineering & Computer Science (AREA)
- Physics & Mathematics (AREA)
- Theoretical Computer Science (AREA)
- Mathematical Physics (AREA)
- Computational Linguistics (AREA)
- Data Mining & Analysis (AREA)
- Databases & Information Systems (AREA)
- General Engineering & Computer Science (AREA)
- General Physics & Mathematics (AREA)
- Information Retrieval, Db Structures And Fs Structures Therefor (AREA)
Abstract
A kind of question sentence search method based on conceptual object model, in turn includes the following steps:According to the characteristics of research restriction field question sentence, according to conceptual object model knowledge representation method, field concept object model is set up;User inputs text question sentence and determined;Similarity Measure mode based on conceptual object model definition question sentence;Similarity Measure mode based on question sentence carries out question sentence retrieval.This method can carry out semantic analysis and improve efficiency, reduce time complexity.
Description
Technical field
The present invention relates to automatically request-answering system research field, more particularly relate to the phase of question sentence in automatically request-answering system
Calculated like degree and question sentence retrieval research.
Background technology
The Similarity Measure difficulty of Chinese sentence is very big, and conventional method is mainly included based on TF-IDF methods and based on semanteme
Algorithm.TF-IDF methods based on vector space model are mainly used in large-scale file retrieval, are only included when sentence
When word is enough, the method just has relatively good effect.Due to limit field question sentence it is short and small and comprising vocabulary quantity it is few,
So application effect is not good enough.TF-IDF methods only considered the statistical property of word within a context, not enterprising in semantic level
Row analysis, it is impossible to the characteristic of Chinese is embodied well, with certain limitation.Based on semantic algorithm:Due to semantic data
The limitation in storehouse, efficiency of algorithm is low so that the real-time of question answering system is deteriorated, and a certain language is needed based on semantic Similarity Measure
The support of adopted knowledge dictionary, and the mainly world knowledge included in knowledge dictionary, and contain for a certain professional domain knowledge package
It is less, lack the support to professional domain keyword, less effective calculated to the Question sentence parsing in field.
Therefore the present invention is calculated by the relevant feature to domain knowledge and current statement similarity has the problem of progress
Research and analyse, on the basis of conceptual object model is set up with reference to certain semantic analysis, it is proposed that based on conceptual object model
Question sentence parsing computational methods, this method can carry out the analysis of field concept Object Semanteme and improve efficiency of algorithm, and reduction is calculated
Method time complexity.
The content of the invention
Simple semantic analysis can be carried out it is an object of the invention to overcome the deficiencies of the prior art and provide one kind and is carried
High efficiency of algorithm, reduces the sentence retrieval method based on conceptual object model of Algorithms T-cbmplexity, successively including following step
Suddenly:
Step 1:According to the characteristics of research restriction field question sentence, according to conceptual object model knowledge representation method, neck is set up
Domain conceptual object model;
Step 2:User inputs text question sentence and determined;
Step 3:Character string stream process, extracts concept, object and the attribute in question sentence, utilizes concept or object factory question sentence
Theme, using concept or the attribute description question sentence focus of object, question sentence is expressed as conceptual object model representation;
Step 4:It is specially that Question sentence parsing is calculated based on the Similarity Measure mode of conceptual object model definition question sentence
It is divided between four parts, object between Similarity Measure, attribute Similarity Measure and remainder similarity between Similarity Measure, relation
Calculate, last weighted calculation obtains the Similarity Measure mode of question sentence;
Step 5:Similarity Measure mode based on question sentence carries out question sentence retrieval.
Preferably, it is described according to conceptual object model knowledge representation method, set up field concept object model, specific steps
For:Field concept, domain object, concept attribute and object properties and they it are therefrom extracted by analysis field question sentence feature
Between relation, obtain field concept set, domain object set, attribute set and set of relationship, through analyze set of relationship in
The inheritance between relation, concept and concept, concept between the representation of contextual definition relation, wherein concept and attribute with
Relation between object is individually represented, finally sets up field concept object model, specific as follows:
Step 1.1:Field concept is extracted, field concept set Concepts={ C are set up1,C2,…Cn};
Step 1.2:Domain object is extracted, domain object set Objects={ O are set up1,O2,…,Om};
Step 1.3:Concept attribute and object properties are extracted, attribute set Attributes={ A are set up1,A2,…,Ak};
Step 1.4:Extract the relation between field concept, domain object, concept attribute and object properties, opening relationships collection
Close Relations={ R1,R2,…,Rl};
Step 1.5:Relation, concept and the concept between relation, concept and attribute between defined notion, attribute and object
Between inheritance, concept and object between relation and other relations representation, set up field concept object model.
Preferably, the step 3 is comprised the following steps that:
Step 3.1:Using Forward Maximum Method method, all spectra object in user's question sentence is found out according to domain object storehouse,
If question sentence comprises only a domain object, 3.2 are gone to step, 3.3 are otherwise gone to step;
Step 3.2:For domain object, according to field concept object model, the attribute for obtaining its concept belongs to as object
Property, the object properties that user's question sentence is included are found out using Forward Maximum Method method, if not finding object properties in question sentence,
A default property is provided, the conceptual object for eventually forming inquiry question sentence represents Q={ Qobject,Qattribute,Qrest};
Step 3.3:For comprising two and during two or more domain object, first according to field concept object mould
Type, judges the relation between object, if between an object and other objects being the relation of object and attribute, then it is assumed that question sentence
Only comprising an object, remaining is the attribute of object, goes to step 3.2;If concept and another object belonging to an object
Affiliated concept is hyponymy, then using the next object as unique domain object, goes to step 3.2;Other situations are gone to step
3.4;
Step 3.4:For question sentence include multiple fields object in the case of, question sentence be expressed as domain object, object properties,
Object relationship and other four parts, Q={ Qobject,Qattribute,Qrelation,Qrest, obtain the conceptual object mould of user's question sentence
Formula representation.
Preferably, the Question sentence parsing in the step 4 calculates the similarity meter being decomposed between the object that question sentence is included
The Similarity Measure between Similarity Measure, object relationship, remainder Similarity Measure four between calculation, object properties
Point;The Similarity Measure of corresponding two set is regarded in the calculating of wherein each several part as, chooses one from a set successively
Individual element calculates similarity with the element in another set respectively, picks out the element pair of maximum similarity, and circulation is until the
One collection is combined into sky;Then the similarity for these elements pair picked out is added, divided by first is gathered the element included
Number;The result calculated respectively based on two set is finally averagely obtained to the phase between the similarity of two set, question sentence
Calculate and obtained by four some numerical results weighted calculations like degree.
Preferably, the step 4 is comprised the following steps that:
Step 4.1:Object part Similarity Measure:
Sim (O in formulau,Ov) it is object OuAnd OvSimilarity, if same object, its value is 1, is otherwise 0, Ou
And OvIt is Q' respectivelyobj,QobjIn relation, n1 and m1 are Q' respectivelyobj,QobjThe number of middle relation;
Step 4.2:Attribute section Similarity Measure:
Sim (A in formulau,Av) it is attribute AuWith attribute AvSimilarity, if same attribute, its value is 1, otherwise for
0, attribute AuWith attribute AvIt is Q' respectivelyatt,QattIn relation, n2 and m2 are Q' respectivelyatt,QattThe number of middle relation;
Step 4.3:Similarity Measure between relation:
Sim (R in formulau,Rv) it is relation RuWith relation RvSimilarity, if same relation, its value is 1 to be otherwise
0, RuAnd RvIt is Q' respectivelyrel,QrelIn relation, n3 and m3 are Q' respectivelyrel,QrelThe number of middle relation;
Step 4.4:Remainder Similarity Measure:
Sim (W in formulau,Wv) it is word WuWith word WvSimilarity, its computational methods be based on semantic computation mode, WuAnd WvPoint
It is not Q'resAnd QresIn word, n4 and m4 are Q' respectivelyrestAnd QrestThe number of middle word;
Step 4.5:The Similarity Measure of final question sentence:
SIM (Q', Q)=a1SIMqobj(Q'obj,Qobj)+a2SIMqatt(Q'att,Qatt)
+a3SIMqrel(Q'rel,Qrel)+a4SIMqres(Q'res,Qres)
Wherein, a1+a2+a3+a4=1, a1, a2, a3, a4The distribution coefficient of each similarity is represented respectively.
Preferably, the object in question sentence represents the theme of question sentence, and its attribute represents the focus of question sentence, wherein a1=0.3, a2
=0.4, a3=0.15, a4=0.15.
Preferably, if lacking certain part in the object model representation of question sentence, the coefficient of this part press than
Example is assigned in other coefficients.
The question sentence is retrieved for retrieving the question sentence similar to user's question sentence, is retrieved by Similarity Measure and sequence
As a result.
The question sentence is retrieved for retrieving the question sentence similar to user's question sentence, is retrieved by Similarity Measure and sequence
As a result.
Brief description of the drawings
Fig. 1 conceptual object model schematics
Fig. 2 Similarity Measure logic charts
Embodiment
Describe the specific implementation of the present invention in detail by taking colleges and universities' enrolment consultation field as an example below, it is necessary to it is pointed out here that
It is to implement to be only intended to further illustrating for the present invention below, it is impossible to be interpreted as limiting the scope of the invention, the field
Some nonessential modifications and adaptations that person skilled in the art makes according to the invention described above content to the present invention, still fall within this
The protection domain of invention.
Explanation is defined to the object in domain knowledge, concept, attribute etc. first.
It is that people will carry out puing question to the anything of understanding to define 1 object, and it can be specific things or take out
The event of elephant.
Defining 2 concepts, there is the abstract of object of same characteristic features to be exactly concept.
Defining 3 attributes is used to describe concept or a certain method characteristic of object.
In colleges and universities' enrolment consultation field, " specialty " is exactly a concept, and " statistics ", " sociology " are exactly two right
As " enrollment " is exactly object " statistics ", the attribute of " sociology ".
In colleges and universities' enrolment consultation field, concept therein, object, attribute difference formalized description are manually extracted as follows:
Concept set:Concepts={ C1,C2,…Cn} (1)
Object set:Objects={ O1,O2,…,Om} (2)
Attribute set:Attributes={ A1,A2,…,Ak} (3)
Relation between justice is former between reference WordNet notional words and in HowNet is (for example:Hyponymy, synonymous pass
System, part-whole relationship etc.) relation defined between field concept is designed, and by manually marking the relation between concept.Obtain
All set of relationship, formalization representation is as follows:
Set of relationship:Relations={ R1,R2,…,Rl} (4)
In these relations, inheritance between relation, concept and the concept between concept and attribute is found through analysis, general
Read and be related to that this three classes Relationship Comparison is special between object, individually represent, remaining relation set of relationship
Relations is represented.
Using the concept set above obtained, object set, attribute set, set of relationship as conceptual object model concept
Node, Object node, attribute node, relation node.Wherein, rectangle represent concept node, round rectangle represent Object node, it is ellipse
Circle represents that attribute node, rhombus represent relation node.Line between node is using following rule:Arrow is used between upper subordinate concept
Line is connected, arrow point between subordinate concept, concept and its attribute connected with the line of one end band point, concept and example object it
Between connected, connected with straight line between remaining node with the line of two ends band point.The attribute of object can be obtained from the concept belonging to it,
Do not described in model.It is specifically shown in Figure of description 1.
The conceptual object model form of object and Attribute Recognition and question sentence represents that step is as follows:
Need first to carry out word segmentation processing for the inquiry question sentence of user, the present invention utilizes the Words partition system of the Chinese Academy of Sciences
ICTCLAS, then removes stop words, obtains user's inquiry question sentence keyword and represents Q={ W1, W2 ..., Wn }.Then identification is asked
Object and its attribute in sentence, specific recognizer are as follows:
Input:User's question sentence keyword represents Q={ W1, W2 ..., Wn }.
Output:The object model of user's question sentence is represented.
Step 1:Using Forward Maximum Method method, all spectra object in user's question sentence is found out according to domain object storehouse, if
Question sentence comprises only a domain object, goes to step 2, otherwise goes to step 3.
Step 2:For domain object, according to field concept object model, the attribute for obtaining its concept is used as the category of object
Property, the object properties that user's question sentence is included are found out using maximum forward matching algorithm, if not finding object properties in question sentence,
One default property is then provided.Such as object is " accounting speciality ", then its default property is " brief introduction " attribute.Eventually form and look into
The conceptual object of inquiry sentence represents Q={ Qobject,Qattribute,Qrest}。
Step 3:In the case of comprising two or more domain object, formality according to conceptual object model, judge object it
Between relation, if between an object and other objects being the relation of object and attribute, then it is assumed that question sentence only comprising one it is right
As remaining is the attribute of object, goes to step 2.If on concept belonging to an object and the concept belonging to another object be
The next relation, then using the next object as unique domain object, go to step 2.Other situations go to step 4.
Step 4:When question sentence includes multiple fields object, question sentence is expressed as domain object, object properties, right
As relation and other four parts, Q={ Qobject,Qattribute,Qrelation,Qrest}。
By above-mentioned algorithm, the object model for obtaining user's inquiry question sentence is represented.
The Similarity Measure that user is inquired about between question sentence and candidate's question sentence collection is decomposed into the phase between the object that question sentence is included
Like the Similarity Measure between the Similarity Measure between degree calculating, object properties, object relationship, remainder Similarity Measure
Four parts.
The Similarity Measure of corresponding two set is regarded in the calculating of each several part as, chooses one from a set successively
Individual element respectively with another set in element calculate similarity, the element pair for the maximum similarity selected, circulation until
First collection is combined into sky.Then the similarity for these elements pair picked out is added, divided by first is gathered the element included
Number.The result calculated respectively based on two set is finally averagely obtained to the similarity of two set.Each several part is similar
Degree calculates specific as follows:
Step 4.1:Object part Similarity Measure:
Sim (O in formulau,Ov) it is object OuAnd OvSimilarity, if same object, its value is 1, is otherwise 0, Ou
And OvIt is Q' respectivelyobj,QobjIn relation, n and m are Q' respectivelyobj,QobjThe number of middle relation;
Step 4.2:Attribute section Similarity Measure:
Sim (A in formulau,Av) it is attribute AuWith attribute AvSimilarity, if same attribute, its value is 1, otherwise for
0.Attribute AuWith attribute AvIt is Q' respectivelyatt,QattIn relation, n and m are Q' respectivelyatt,QattThe number of middle relation;
Step 4.3:Similarity Measure between relation:
Sim (R in formulau,Rv) it is relation RuWith relation RvSimilarity, if same relation, its value is 1 to be otherwise
0。RuAnd RvIt is Q' respectivelyrel,QrelIn relation, n and m are Q' respectivelyrel,QrelThe number of middle relation;
Step 4.4:Remainder Similarity Measure:
Sim (W in formulau,Wv) it is word WuWith word WvSimilarity, its computational methods be based on Hownet semantic computation, WuAnd WvPoint
It is not Q'resAnd QresIn word, n and m are Q' respectivelyrestAnd QrestThe number of middle word;
Step 4.5:Similarity calculating method between final question sentence is as follows
SIM (Q', Q)=a1SIMqobj(Q'obj,Qobj)+a2SIMqatt(Q'att,Qatt)
+a3SIMqrel(Q'rel,Qrel)+a4SIMqres(Q'res,Qres)
Wherein, a1+a2+a3+a4=1, found by research, the object in question sentence represents the theme of question sentence, its attribute is represented
The focus of question sentence, this two parts are to describe user to put question to the main aspect being intended to, and user belongs to more focused on certain aspect of object
Property, therefore a is set herein1=0.3, a2=0.4, a3=0.15, a4=0.15, if in the object model representation of question sentence
Lack certain part, then the coefficient of this part is prorated in other coefficients.
The algorithm steps of question sentence retrieval.
In order to improve question sentence effectiveness of retrieval, question and answer are all expressed as conceptual object model to the question sentence of concentration in advance, had
Body question sentence searching algorithm is as follows:
Input:User inquires about question sentence
Output:Most like preceding 5 question sentences
Step 1:User's inquiry question sentence participle, stop words processing and synonym processing;
Step 2:User inquires about question sentence object and its attribute extraction;
Step 3:The similarity of user's inquiry question sentence and candidate's question sentence is calculated, if similarity is more than the threshold value of setting,
The question sentence is put into candidate result to concentrate, the similarity with next candidate's question sentence is then calculated, until candidate's question sentence collection is sky.
Step 4:Obtained candidate result collection is arranged from big to small by similarity, preceding 5 question sentences are taken as the knot of return
Really.
Although for illustrative purposes, it has been described that illustrative embodiments of the invention, those skilled in the art
Member it will be understood that, can be in form and details in the case of the scope and spirit for not departing from invention disclosed in appended claims
The upper change for carrying out various modifications, addition and replacement etc., and all these changes should all belong to appended claims of the present invention
Each step in protection domain, and claimed each department of product and method, can be in any combination
Form is combined.Therefore, to disclosed in this invention embodiment description be not intended to limit the scope of the present invention,
But for describing the present invention.Correspondingly, the scope of the present invention is not limited by embodiment of above, but by claim or
Its equivalent is defined.
Claims (6)
1. a kind of question sentence search method based on conceptual object model, it is characterised in that in turn include the following steps:
Step 1:According to the characteristics of research restriction field question sentence, according to conceptual object model knowledge representation method, field is set up general
Read object model;
Step 2:User inputs text question sentence and determined;
Step 3:Character string stream process, extracts concept, object and the attribute in question sentence, utilizes concept or the master of object factory question sentence
Question sentence, using concept or the attribute description question sentence focus of object, is expressed as conceptual object model representation by topic;
Step 4:It is specially that Question sentence parsing is calculated to be divided into based on the Similarity Measure mode of conceptual object model definition question sentence
Similarity Measure and remainder similarity meter between Similarity Measure, relation between Similarity Measure, attribute between four parts, object
Calculate, last weighted calculation obtains the Similarity Measure mode of question sentence;
Step 5:Similarity Measure mode based on question sentence carries out question sentence retrieval;
The step 3 is comprised the following steps that:
Step 3.1:Using Forward Maximum Method method, all spectra object in user's question sentence is found out according to domain object storehouse, if asking
Sentence comprises only a domain object, goes to step 3.2, otherwise goes to step 3.3;
Step 3.2:For domain object, according to field concept object model, the attribute of its concept is obtained as object properties, profit
The object properties that user's question sentence is included are found out with Forward Maximum Method method, if not finding object properties in question sentence, are provided
One default property, the conceptual object for eventually forming inquiry question sentence represents Q={ Qobject,Qattribute,Qrest};
Step 3.3:For comprising two and during two or more domain object, first according to field concept object model, sentencing
Relation between disconnected object, if between an object and other objects being the relation of object and attribute, then it is assumed that question sentence is only wrapped
Containing an object, remaining is the attribute of object, goes to step 3.2;If belonging to the concept and another object belonging to an object
Concept be hyponymy, then using the next object as unique domain object, go to step 3.2;Other situations go to step 3.4;
Step 3.4:When question sentence includes multiple fields object, question sentence is expressed as domain object, object properties, object
Relation and other four parts, Q={ Qobject,Qattribute,Qrelation,Qrest, obtain the conceptual object pattern table of user's question sentence
Show form.
2. a kind of question sentence search method based on conceptual object model as claimed in claim 1, it is characterised in that:It is described according to
Conceptual object model knowledge representation method, sets up field concept object model, concretely comprises the following steps:Pass through analysis field question sentence feature
Field concept, domain object, concept attribute and object properties and the relation between them are therefrom extracted, field concept collection is obtained
Conjunction, domain object set, attribute set and set of relationship, through analyzing the representation of the contextual definition relation in set of relationship,
The relation between inheritance, concept and the object between relation, concept and concept wherein between concept and attribute is individually
Represent, finally set up field concept object model, it is specific as follows:
Step 1.1:Field concept is extracted, field concept set Concepts={ C are set up1,C2,…Cn};
Step 1.2:Domain object is extracted, domain object set Objects={ O are set up1,O2,…,Om};
Step 1.3:Concept attribute and object properties are extracted, attribute set Attributes={ A are set up1,A2,…,Ak};
Step 1.4:Extract the relation between field concept, domain object, concept attribute and object properties, opening relationships set
Relations={ R1,R2,…,Rl};
Step 1.5:Between relation, concept and the concept between relation, concept and attribute between defined notion, attribute and object
The representation of relation and other relations between inheritance, concept and object, sets up field concept object model.
3. a kind of question sentence search method based on conceptual object model as claimed in claim 1, it is characterised in that:The step
Question sentence parsing in 4, which is calculated, is decomposed into the Similarity Measure between the object that question sentence is included, the similarity between object properties
Similarity Measure, the part of remainder Similarity Measure four between calculating, object relationship;The calculating of wherein each several part is regarded as
Be it is corresponding two set Similarity Measures, successively from one set in choose an element respectively with another set in
Element calculates similarity, picks out the element pair of maximum similarity, and circulation is combined into sky until first collection;Then picking out
The similarity of these elements pair is added, divided by first is gathered the element number included;Base finally will be combined into respectively with two collection
The Similarity Measure that the result that plinth is calculated averagely is obtained between the similarity of two set, question sentence is weighted by four some numerical results
Calculating is obtained.
4. a kind of question sentence search method based on conceptual object model as claimed in claim 3, it is characterised in that:In question sentence
Object represents the theme of question sentence, and its attribute represents the focus of question sentence, wherein a1=0.3, a2=0.4, a3=0.15, a4=0.15.
5. a kind of question sentence search method based on conceptual object model as claimed in claim 3, it is characterised in that:If question sentence
Object model representation in lack certain a part, then the coefficient of this part is prorated in other coefficients.
6. a kind of question sentence search method based on conceptual object model as claimed in claim 1, it is characterised in that:The question sentence
Retrieve for retrieving the question sentence similar to user's question sentence, by Similarity Measure and sequence obtains retrieval result.
Priority Applications (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201410713510.3A CN104331523B (en) | 2014-11-27 | 2014-11-27 | A kind of question sentence search method based on conceptual object model |
Applications Claiming Priority (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201410713510.3A CN104331523B (en) | 2014-11-27 | 2014-11-27 | A kind of question sentence search method based on conceptual object model |
Publications (2)
Publication Number | Publication Date |
---|---|
CN104331523A CN104331523A (en) | 2015-02-04 |
CN104331523B true CN104331523B (en) | 2017-07-28 |
Family
ID=52406250
Family Applications (1)
Application Number | Title | Priority Date | Filing Date |
---|---|---|---|
CN201410713510.3A Active CN104331523B (en) | 2014-11-27 | 2014-11-27 | A kind of question sentence search method based on conceptual object model |
Country Status (1)
Country | Link |
---|---|
CN (1) | CN104331523B (en) |
Families Citing this family (7)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN104820681A (en) * | 2015-04-17 | 2015-08-05 | 清华大学 | Response method and system for online Q&A service |
CN104915396A (en) * | 2015-05-28 | 2015-09-16 | 杭州电子科技大学 | Knowledge retrieving method |
CN105653671A (en) * | 2015-12-29 | 2016-06-08 | 畅捷通信息技术股份有限公司 | Similar information recommendation method and system |
CN107423432B (en) * | 2017-08-03 | 2020-05-12 | 当家移动绿色互联网技术集团有限公司 | Method and system for distinguishing professional problems and small talk problems by robot |
CN107766498B (en) * | 2017-10-19 | 2022-01-07 | 北京百度网讯科技有限公司 | Method and apparatus for generating information |
CN110543553A (en) * | 2019-07-31 | 2019-12-06 | 平安科技(深圳)有限公司 | question generation method and device, computer equipment and storage medium |
CN110704292B (en) * | 2019-10-15 | 2020-11-03 | 中国人民解放军海军大连舰艇学院 | Evaluation method for display control interface design |
Citations (1)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN101286161A (en) * | 2008-05-28 | 2008-10-15 | 华中科技大学 | Intelligent Chinese request-answering system based on concept |
-
2014
- 2014-11-27 CN CN201410713510.3A patent/CN104331523B/en active Active
Patent Citations (1)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN101286161A (en) * | 2008-05-28 | 2008-10-15 | 华中科技大学 | Intelligent Chinese request-answering system based on concept |
Non-Patent Citations (2)
Title |
---|
基于HowNet语义相似度的FAQ研究;贾可亮,樊孝忠,张禹;《计算机应用》;20070930;全文 * |
基于问句语义表征的中文问句相似度计算方法;陈康,樊孝忠,刘杰,贾可亮;《北京理工大学学报》;20071231;全文 * |
Also Published As
Publication number | Publication date |
---|---|
CN104331523A (en) | 2015-02-04 |
Similar Documents
Publication | Publication Date | Title |
---|---|---|
CN104331523B (en) | A kind of question sentence search method based on conceptual object model | |
Young et al. | Augmenting end-to-end dialogue systems with commonsense knowledge | |
Saleena | An ensemble classification system for twitter sentiment analysis | |
CN106294593B (en) | In conjunction with the Relation extraction method of subordinate clause grade remote supervisory and semi-supervised integrated study | |
CN106570708B (en) | Management method and system of intelligent customer service knowledge base | |
CN107515877B (en) | Sensitive subject word set generation method and device | |
Prusa et al. | The effect of dataset size on training tweet sentiment classifiers | |
KR102288249B1 (en) | Information processing method, terminal, and computer storage medium | |
CN106326440B (en) | A kind of man-machine interaction method and device towards intelligent robot | |
WO2019080863A1 (en) | Text sentiment classification method, storage medium and computer | |
CN106844346A (en) | Short text Semantic Similarity method of discrimination and system based on deep learning model Word2Vec | |
CN110826337A (en) | Short text semantic training model obtaining method and similarity matching algorithm | |
CN109960786A (en) | Chinese Measurement of word similarity based on convergence strategy | |
CN110532379B (en) | Electronic information recommendation method based on LSTM (least Square TM) user comment sentiment analysis | |
CN110175221B (en) | Junk short message identification method by combining word vector with machine learning | |
CN103593412B (en) | A kind of answer method and system based on tree structure problem | |
CN107239512B (en) | A kind of microblogging comment spam recognition methods of combination comment relational network figure | |
CN111177310A (en) | Intelligent scene conversation method and device for power service robot | |
CN112819023A (en) | Sample set acquisition method and device, computer equipment and storage medium | |
CN110297888A (en) | A kind of domain classification method based on prefix trees and Recognition with Recurrent Neural Network | |
CN108108354A (en) | A kind of microblog users gender prediction's method based on deep learning | |
CN104008187A (en) | Semi-structured text matching method based on the minimum edit distance | |
CN106874397B (en) | Automatic semantic annotation method for Internet of things equipment | |
CN105912525A (en) | Sentiment classification method for semi-supervised learning based on theme characteristics | |
CN112699232A (en) | Text label extraction method, device, equipment and storage medium |
Legal Events
Date | Code | Title | Description |
---|---|---|---|
C06 | Publication | ||
PB01 | Publication | ||
C10 | Entry into substantive examination | ||
SE01 | Entry into force of request for substantive examination | ||
GR01 | Patent grant | ||
GR01 | Patent grant |