CN106095762A - A kind of news based on ontology model storehouse recommends method and device - Google Patents
A kind of news based on ontology model storehouse recommends method and device Download PDFInfo
- Publication number
- CN106095762A CN106095762A CN201610081578.3A CN201610081578A CN106095762A CN 106095762 A CN106095762 A CN 106095762A CN 201610081578 A CN201610081578 A CN 201610081578A CN 106095762 A CN106095762 A CN 106095762A
- Authority
- CN
- China
- Prior art keywords
- news
- information
- knowledge
- dimension
- attribute
- 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
- G06F16/90—Details of database functions independent of the retrieved data types
- G06F16/95—Retrieval from the web
- G06F16/953—Querying, e.g. by the use of web search engines
- G06F16/9535—Search customisation based on user profiles and personalisation
Landscapes
- Engineering & Computer Science (AREA)
- Databases & Information Systems (AREA)
- Theoretical Computer Science (AREA)
- Data Mining & Analysis (AREA)
- Physics & Mathematics (AREA)
- General Engineering & Computer Science (AREA)
- General Physics & Mathematics (AREA)
- Information Retrieval, Db Structures And Fs Structures Therefor (AREA)
- Machine Translation (AREA)
Abstract
The invention provides a kind of news based on ontology library and knowledge mapping and recommend method and device.Said method comprising the steps of: (1) pretreatment, including punctuate, participle, part-of-speech tagging etc.；(2) news body identification, is identified the news ontology information in input newsletter archive；(3) key message excavates, and excavates the key message (key word, crucial body etc.) in input news chapter in conjunction with the news ontology information identified；(4) attribute obtains, the information obtained based on preceding step, and inquiry knowledge mapping obtains knowledge point property information, and expands according to knowledge mapping relation, excavates attribute information further；(5) dimension generates, and the information (body, key word, attribute etc.) obtained based on preceding step generates recommendation dimension；(6) recommend knowledge to recall, based on recommending dimension, in news documents storehouse and knowledge base, carry out recommending the acquisition of knowledge.
Description
Technical field
The present invention relates to natural language processing, intelligent recommendation technical field, particularly relate to a kind of based on news ontology knowledge
The intelligent news of storehouse and knowledge mapping recommends method and device thereof.
Background technology
Internet era, the mobile Internet the most just risen, big data age, news recommend be news read
A key function in service, to promoting Consumer's Experience, increases user's viscosity and has greatly effect.
In prior art, news recommended technology typically has two kinds.
One is content-based recommendation technology, and the vector space model being based primarily upon word bag realizes Documents Similarity weighing apparatus
Amount.Another is collaborative filtering based on user behavior, is mainly based upon user's historical behavior data, carries out potential use
The prediction of family news interested.
Problem of the prior art is:
One, technology based on commending contents, is based only on key word information, it is impossible to find the similarity of Deep Semantics.To in language
Polysemy, adopted many words phenomenon cannot be carried out well modeling；
Two, based on commending contents technology, all documents recommended obtain based on a unified method for measuring similarity
(e.g., typically taking Top 10).Recommend document and original text and recommend all to there is great similarity between document, being the most all not
With the mutual reprinting in source, user can not be brought effective information gain；
Three, collaborative filtering based on user behavior, there is also above-mentioned two problems.It addition, the problem that there is also cold start-up.
Because this technical scheme needs historical behavior based on user to be trained modeling.For there is no the field of user behavior data
Scape, it is difficult to effectively carry out.
Prior art (comprising above two) is difficult to solve to recommend the degree of depth and the problem of range.The degree of depth is recommended to refer to input
News documents, it is impossible to obtain its background knowledge and association cause and effect information.Range is recommended to refer to for input document, it is impossible to horizontal stroke
Carry out recommending (similar incidents that such as, Paris the most probably expands to China) to relevant news or knowledge.And recommend the degree of depth with
And recommendation range, the professional journalists (such as editor, reporter etc.) for specialty is particularly important.They need people in routine duties
Work take considerable time energy the related information knowledge of the degree of depth, range is carried out collect excavate, also result in this kind of work to from
Dealer's experience accumulation and the high request of the extensive degree of knowledge so that news working is relatively costly.
Summary of the invention
How the technical problem to be solved in the present invention is to inputting one or more news documents, automatically recommends dimension rich
Rich, it is provided simultaneously with certain depth and the relevant document of range or knowledge.
Said method comprising the steps of: (1) pretreatment, including punctuate, participle, part-of-speech tagging etc.；(2) news body is known
Not, the news ontology information in input newsletter archive is identified；(3) key message excavates, in conjunction with the news identified
Key message (key word, crucial body etc.) in input news chapter is excavated by ontology information；(4) attribute obtains, base
In the information that preceding step obtains, inquiry knowledge mapping obtains knowledge point property information, and opens up according to knowledge mapping relation
Exhibition finds, excavates attribute information further；(5) dimension generates, information (body, key word, the attribute obtained based on preceding step
Deng) generate and recommend dimension；(6) recommend knowledge to recall, based on recommending dimension, recommend in news documents storehouse and knowledge base
The acquisition of knowledge.
Described device includes following unit: (1) pretreatment unit, it is achieved the input punctuate of text, participle, part of speech
Mark etc.；(2) news body recognition unit, it is achieved the news ontology information identification process in input newsletter archive；(3) crucial letter
Breath excavates unit, it is achieved the automatic mining of the key message (key word, crucial body etc.) in news chapter；(4) attribute obtains
Unit, it is achieved the knowledge of knowledge mapping inquiry and knowledge based figure genealogical relationship is expanded, and obtains correlation attribute information；(5) dimension
Signal generating unit, it is achieved recommend the generation of dimension, the output information (body, key word, attribute etc.) of the unit that continues before being mainly based upon
It is optimized combination, recommends dimension to generate；(6) knowledge is recommended to recall unit, it is achieved recommendation dimension based on the unit output that front continues
Degree, carries out recommending the acquisition of knowledge in news documents storehouse and knowledge base.
The intelligent news based on news ontology library and knowledge mapping that the embodiment of the present invention provides recommends method and device,
By news body identification and the attribute excavation of knowledge based collection of illustrative plates, input text words can be departed from and limit, it is thus achieved that extensively
Related information very abundant on degree and the degree of depth；By these related informations, then based on dimension generating algorithm, in document sets and
Knowledge mapping carries out recalling of relevant documentation and knowledge point, it is possible to obtain there is the information of bigger breadth and depth.Make big
Many users can enjoy the information service that content is extensive, associate deeply and have knowledge content, and especially, obtain employment people to Journalism
From the point of view of scholar, its work efficiency can be greatly enhanced, promote its business output level.
Accompanying drawing explanation
Fig. 1 is that a kind of news based on ontology model storehouse recommends method flow diagram；
Fig. 2 is news ontology library schematic diagram；
Fig. 3 is that news body finds flow chart；
Fig. 4 is a kind of news knowledge mapping schematic diagram；
Fig. 5 is a kind of news recommendation apparatus frame diagram based on ontology model storehouse；
Fig. 6 is that a kind of news knowledge mapping attribute expands schematic diagram.
Detailed description of the invention
For making the object, technical solutions and advantages of the present invention clearer, below in conjunction with specific embodiment, and reference
Accompanying drawing, the present invention is described in further detail.
Fig. 1 is the schematic diagram that the intelligent news that the present invention proposes recommends method.
Intelligent news as shown in Figure 1 recommends the input of method 100 to be news documents 111, and can be one can also be
Many." news documents " mentioned here refers to the newsletter archive that various media are published, and concrete form can be form web page, also
It can be the form (such as xml form) of plain text or other any half structure.
Step S120, carries out pretreatment to news documents 111, including punctuate process, participle, part-of-speech tagging.Punctuate processes
It is that text 111 is disconnected with fullstop, is divided into multiple sentence and carries out subsequent treatment.Participle, part-of-speech tagging carry out word Chinese text
Language cutting processes, and gives the part of speech of each word mark its correspondence upper, and part of speech includes noun, verb etc..
Step S121, to pretreated text 111, carries out body identification, and " body " here refers to relate in text
And the concept in the News Field arrived.Such as " theme of news ", " news region ", " media event " etc..And these bodies tool
Having levels property.
Fig. 2 is news ontology library schematic diagram.
As in figure 2 it is shown, the ground floor child node of news ontology knowledge base comprises " theme of news ", " news region ", " news
Event ", " newsmaker ", " other " etc..The most each child node is again a subtree.Length is limited, and Fig. 2 simply local shows
It is intended to.For example, " media event " can be divided into again " delay event ", " unexpected incidents ", " other events " etc..Its
In, " accident " can be layered again segmentation further.Can be sub-divided into the most in this example " media event-> accident->
Burst occurred events of public safety-> social security events-> the attack of terrorism-> shooting incident ".
Each the corresponding one group of attribute definition of body node, for describing news concept corresponding to this body at Xin Wen Bao
The main points paid close attention to during road.In fig. 2, represent, with one group of ellipse, the attribute that node is corresponding.Such as, " attack of terrorism " this
The attribute of body point is (" attacker ", " person of being attacked ", " time ", " place ") in the present embodiment.In other are embodied as, can
It is customized with the application demand according to embodiment.
Body child node acquiescence inherits the attribute of body father node, it is also possible to carry out increasing, deleting according to the demand of being embodied as
Remove, revise.
Fig. 3 is news body identification process figure.
News body recognition methods 300 as shown in Figure 3, can enter with part-of-speech tagging result through the participle of pretreatment S120
Row body identification.
Step S311, carries out synonym extension to each vocabulary, utilizes synonym table to find synonym.Such as " capital " is
The synonym in " Beijing ", " loss " is the synonym of " loss ".
Step S312, carries out bottom body discovery.Utilize a kind of expression way of context discovery of vocabulary or one
The complex concept that words represent.Such as " within * hour, two days " is time concept, and " having waited two days " is " the most timely " concept.
This discovery procedure is supported based on bottom ontology knowledge base, carries out pattern match with regular expression.Common pattern is by manually
Sum up, it is also possible to be aided with the automatic discovery technique of machine, but technique is not emphasis of the present invention, does not do and launches.
Step S313, carries out body and traces back, recalled by upper strata body.As, the bottom that above-mentioned steps is found
Body "Natural disaster", can recall "Natural disaster-> Emergent Public Events-> accident-> media event " this body tree
Path.
It addition, each body node, should there is the definition of attribute mutually.Such as " natural disaster " this body point, right
Attribute is answered to have " date occurs ", " scene ", " origin cause of formation ", " the condition of a disaster scale ", " rescue of dealing with problems arising from an accident ", " instructions from the higher level ", " later stage shadow
Ring " etc..
Step S122, carries out key message excavation to the text after body identification.Described key message comprises key
Word, crucial body.Key word is from the vocabulary of appearance in input text, the body letter that crucial body obtains from S121 step
Breath.Key word excavates the classical way of (document having is referred to as keyword abstraction, and English is keyword extraction) such as
TF*IDF
: t represents the vocabulary currently considering to be scored, and d represents the document of current consideration, and n represents to concentrate in unitary document and comprises t's
Number of files, N represents the number of files of unitary document collection.In specific embodiment in the present invention, this algorithm is optimized, examines
Consider the information of body.
The computational methods of the present embodiment are:
Wherein,WithFor weight parameter, be used for adjusting between original vocabulary TFIDF information and body TFIDF information is important
Property, it is preferable that it is 0.5.O (t) is the body of all correspondences of t.Represent the number of levels of difference between t and o.Employing factorial is fallen
As the form of weighting, number guarantees that raise its disturbance degree declines along with the level of abstraction of body.
Assume that comprising a vocabulary t in the input newsletter archive of the present embodiment is " Paris attack of terrorism ", and it is at body
Cognitive phase identification obtains ontology information" media event-> accident-> burst occurred events of public safety-> social safety thing Part-> the attack of terrorism-> shooting incident ", then the weight increment of t is by " shooting incident " this body(1/(1+1)!)TFIDF (" shooting incident "), and the weight increment of t is by " attack of terrorism " this body(1/(1+2)!) TFIDF (" attack of terrorism ").
The optimization method that the present embodiment is carried can be solved the statistic that many words synonym causes divided by the information of comprehensive body
The problem dissipated, it is also possible to consider the mutual gain of same or like semantic vocabulary to a certain extent, to excavate more adduction
The key message of reason.
The excavation of crucial body is similar with the excavation of key word, and only, the excavation of crucial body only considers that its upper strata is originally
The body impact on it, without the impact considering concrete vocabulary.
This step can also use other key word method for digging such as TextRank, and can be combined this similarly
The optimization of body knowledge.
Step S123, the information obtained according to preceding step, the acquisition of attribute is carried out based on news knowledge mapping.Before described
The information that face step obtains comprises the vocabulary comprised in input text, the body identified and the crucial letter excavated
Breath.Preferably, the key message (key word and crucial body) excavated is utilized to carry out in news ontology knowledge collection of illustrative plates
Inquiry, finds the knowledge card of correspondence, obtains attribute information therein.
Fig. 4 is news knowledge mapping schematic diagram.
Described news knowledge mapping uses general RDF tlv triple (entity 1, relation, entity 2) to be described, but its
It is that the professionalism according to News Field is described.Entity in tlv triple can be that an entity is (such as name, place name, machine
Structure name etc.), it is also possible to it is media event.
The pattern (level and attribute list) that the representation of knowledge defines according to the news ontology library shown in Fig. 2 is carried out.Assume
Input text can obtain " Paris " in preceding step, " shooting incident ", " ISIS ", the relevant information such as " French ", these letters
Breath, as inquiry, can find the entity of correspondence in knowledge mapping.
Each entity is to there being property value.Then according to the relation between these property values and binding entity and entity,
Continue to inquire about further in knowledge mapping, more multiple entity can be recalled, and then obtain more attribute information.This mistake
Journey is referred to as attribute expanding course.Fig. 6 demonstrates the process that an attribute is expanded.
Step S124, the information obtained according to preceding step, carry out recommending the generation of dimension.Described recommendation dimension refers to document
Classification or label.Such as " China+attack of terrorism " this dimension represents the attack of terrorism occurred in the range of China
The relevant information of event.
And " Chinese " therein, " attack of terrorism " is exactly the information obtained in preceding step.Recommend the generation of dimension
Mainly one item of information combines preferred process.
Preferably, the input information of this step comprises the key word obtained from input text, the crucial letter such as body, attribute
Breath item.The dimension of output is i.e. the combination of these items of information.Preferably, dimension generates and follows following guideline:
1) item of information of name entity, such as people's name, can be separately as a dimension；
2) group item is the most, and mark is the highest.Such as " China+attack of terrorism " is better than " attack of terrorism "；
3) heterogeneous information item combination, mark is the highest.Such as " new three plates+sports industry " is better than " new three plates+additional issue "；
4) scoring for key message in preceding step is combined.
It is exactly a process being combined at all items of information that the present embodiment dimension generates process, in order to prevent combining
Many, use search stack strategy to carry out beta pruning.K the dimension only keeping score the highest.Preferably, k is set as 5.
Step S125, the some recommendation dimensions generated according to S124, retrieve in overall news documents respectively, call together
Return relevant documentation, form the recommendation information that each dimension is corresponding.Preferably, it is also possible to according to the single dimensional information generated in dimension
The retrieval carrying out knowledge point in news knowledge mapping is recalled.
Fig. 5 is intelligence news recommendation apparatus schematic diagram.
Intelligence news recommendation apparatus 500 includes 4 processing units and a modeling unit.
Modeling unit 510, is responsible for the news ontology library required for whole device, the management of news knowledge mapping, Yong Huke
With thus unit, news ontology library, news knowledge mapping increased, delete, the operation such as amendment.
Pretreatment unit 521, is responsible for providing the preprocessing function such as participle, part-of-speech tagging.Non-structured text 111 is carried out
Pretreatment, including punctuate process, participle, part-of-speech tagging.It is that text 111 is disconnected with fullstop that punctuate processes, and is divided into multiple sentence
Carry out subsequent treatment.Participle, part-of-speech tagging carry out word segmentation process Chinese text, and give each word mark its correspondence upper
Part of speech, part of speech includes noun, verb etc..
Body recognition unit 522, is responsible for the pre-processed results according to unit 521 output, carries out the identification of news body.First
First each vocabulary is carried out synonym extension, utilize synonym table to find synonym.Then, based on matching regular expressions pattern,
Find bottom body.Operation is traced back, it is thus achieved that body path finally by body tree.
Key message excavates unit 523, is responsible for excavating the key message in input text 111.Such as key word and key
Body.Key word excavates the classical way of (document having is referred to as keyword abstraction, and English is keyword extraction)
Such as TFIDF:
T represents the vocabulary currently considering to be scored, and d represents the document of current consideration, and n represents to concentrate in unitary document and comprises t's
Number of files, N represents the number of files of unitary document collection.In specific embodiment in the present invention, this algorithm is optimized, examines
Consider the information of body.
The carried computational methods of the present invention are:
Wherein,WithFor weight parameter, be used for adjusting between original vocabulary TFIDF information and body TFIDF information is important
Property, it is preferable that it is 0.5.O (t) is the body of all correspondences of t.Represent the number of levels of difference between t and o.Employing factorial is fallen
As the form of weighting, number guarantees that raise its disturbance degree declines along with the level of abstraction of body.
Assume that comprising a vocabulary t in the input newsletter archive of the present embodiment is " Paris attack of terrorism ", and it is at body
Cognitive phase identification obtains ontology information" media event-> accident-> burst occurred events of public safety-> social safety thing Part-> the attack of terrorism-> shooting incident ", then the weight increment of t is by " shooting incident " this body(1/(1+1)!)TFIDF (" shooting incident "), and the weight increment of t is by " attack of terrorism " this body(1/(1+2)!) TFIDF (" attack of terrorism ").
The excavation of crucial body is similar with the excavation of key word, and only, the excavation of crucial body only considers that its upper strata is originally
The body impact on it, without the impact considering concrete vocabulary.
This step can also use other key word method for digging such as TextRank, and can be combined this similarly
The optimization of body knowledge.
Attribute acquiring unit 524, is responsible for knowledge based collection of illustrative plates and carries out the acquisition of attribute.The information that preceding step obtains comprises
Vocabulary, the body identified and the key message excavated comprised in input text.
These information are inquired about by news ontology knowledge collection of illustrative plates, finds the knowledge card of correspondence, obtain genus therein
Property information.Each entity is to there being property value.Then according to the relation between these property values and binding entity and entity, continue
Continue and inquire about further in knowledge mapping, more multiple entity can be recalled, and then obtain more attribute information.
Dimension signal generating unit 525, is responsible for the information obtained according to preceding step, carries out recommending the generation of dimension.Recommend dimension
The generation of degree is mainly an item of information and combines preferred process.
Preferably, the input information of this step comprises the key word obtained from input text, the crucial letter such as body, attribute
Breath item.The dimension of output is i.e. the combination of these items of information.Preferably, dimension generates and follows following guideline:
1) item of information of name entity, such as people's name, can be separately as a dimension；
2) group item is the most, and mark is the highest.Such as " China+attack of terrorism " is better than " attack of terrorism "；
3) heterogeneous information item combination, mark is the highest.Such as " new three plates+sports industry " is better than " new three plates+additional issue "；
4) scoring for key message in preceding step is combined.
It is exactly a process being combined at all items of information that the present embodiment dimension generates process, in order to prevent combining
Many, use search stack strategy to carry out beta pruning.K the dimension only keeping score the highest.Preferably, k is set as 5.
Recommend knowledge to recall unit 526, be responsible for the some recommendation dimensions generated according to unit 525, respectively in overall news
Document is retrieved, recalls relevant documentation, form the recommendation information that each dimension is corresponding.Preferably, it is also possible to according to generation
Single dimensional information in dimension carries out the retrieval of knowledge point in news knowledge mapping and recalls.
The above, the only detailed description of the invention of the present invention, but protection scope of the present invention is not limited thereto, and any
Those familiar with the art, in the technical scope that the invention discloses, can readily occur in change or replace, should contain
Cover within protection scope of the present invention.Therefore, protection scope of the present invention should be as the criterion with described scope of the claims.
Claims (8)
1. a news based on ontology library and knowledge mapping recommends method, it is characterised in that comprise the following steps:
Input newsletter archive is carried out news body identification；
Input newsletter archive is carried out key message excavation；
Attribute acquisition is carried out based on news knowledge mapping；
Carry out recommending dimension to generate based on the key message excavated from input newsletter archive and attribute information；
Recalling and exporting of recommendation information is carried out according to recommending dimension.
2. news body identification as claimed in claim 1, it is characterised in that comprise the steps of
Each vocabulary is carried out synonym extension, utilizes synonym table to find synonym；
According to canonical match pattern, news ontology library is retrieved, find bottom body；
According to the hierarchical structure of body tree, each bottom body is traced back, it is thus achieved that Ontology Matching path.
3. news ontology library as claimed in claim 2, it is characterised in that:
Towards the distinguishing hierarchy in Journalism field, such as it is divided into " media event ", " newsmaker ", " theme of news " etc. big
Class and each big class are divided into again some groups；The classification frequently involved in one news report of each body node on behalf；Often
The all corresponding attribute list of this body node individual, some the fundamental points in this news category corresponding.
4. key message as claimed in claim 1 excavates, it is characterised in that:
Evaluation methodology combines lexical information and ontology information:
Key word excavation classical way such as TFIDF:
T represents the vocabulary currently considering to be scored, and d represents the document of current consideration, and n represents to concentrate in unitary document and comprises t's
Number of files, N represents the number of files of unitary document collection, in specific embodiment in the present invention, is optimized this algorithm, examines
Consider the information of body；The present invention is carried calculating term weight formula:
Wherein, α and β is weight parameter, is used for adjusting the importance between original vocabulary TFIDF information and body TFIDF information,
Preferably, 0.5 it is；O (t) is the body of all correspondences of t；Lo represents the number of levels of difference between t and o；Employing factorial is reciprocal
Form as weighting guarantees that raise its disturbance degree declines along with the level of abstraction of body.
5. news knowledge mapping as claimed in claim 1, it is characterised in that:
Use general RDF tlv triple (entity 1, relation, entity 2) to be described, but it is that the specialty according to News Field is special
Property is described；
Entity in tlv triple can be an entity (such as name, place name, mechanism's name etc.), it is also possible to is media event；
The pattern (level and attribute list) that the representation of knowledge defines according to the news ontology library described in claim 3 is carried out.
6. attribute as claimed in claim 1 obtains, it is characterised in that:
Utilize the key message excavated to inquire about in news ontology knowledge collection of illustrative plates, find the knowledge card of correspondence, obtain
Take attribute information therein；
There is attribute expanding course；Each entity is to there being property value, then according to these property values and binding entity and reality
Relation between body, continues to inquire about further in knowledge mapping, can recall more multiple entity, and then obtain more genus
Property information.
7. recommendation dimension as claimed in claim 1 generates, it is characterised in that:
It is the preferred process of combination that a key message item obtained based on claim 4 is carried out；
It follows following guideline:
1) item of information of name entity, such as people's name, can be separately as a dimension；
2) group item is the most, and mark is the highest；Such as " China+attack of terrorism " is better than " attack of terrorism "；
3) heterogeneous information item combination, mark is the highest；Such as " new three plates+sports industry " is better than " new three plates+additional issue "；
4) scoring for key message in preceding step is combined.
8. a news recommendation apparatus based on ontology library and knowledge mapping, it is characterised in that including:
Modeling unit, is responsible for the news ontology library required for whole device, the management of news knowledge mapping, and user can be thus single
The operations such as news ontology library, news knowledge mapping are increased, delete by unit, amendment；
Pretreatment unit, is responsible for providing the preprocessing function such as participle, part-of-speech tagging；
Body recognition unit, is responsible for carrying out the identification of news body；
Key message excavates unit, is responsible for excavating the key message in input text；
Attribute acquiring unit, is responsible for knowledge based collection of illustrative plates and carries out the acquisition of attribute；
Dimension signal generating unit, is responsible for the information obtained according to preceding units, carries out recommending the generation of dimension；
Recommend knowledge to recall unit, be responsible for the recommendation dimension generated according to preceding units, carry out in overall news documents respectively
Retrieval, recalls relevant documentation, forms the recommendation information that each dimension is corresponding, according to the single dimensional information generated in dimension in news
The retrieval carrying out knowledge point in knowledge mapping is recalled.
Priority Applications (2)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201610081578.3A CN106095762A (en) | 2016-02-05 | 2016-02-05 | A kind of news based on ontology model storehouse recommends method and device |
| CN201610966184.6A CN106570144B (en) | 2016-02-05 | 2016-11-04 | The method and apparatus of recommendation information |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201610081578.3A CN106095762A (en) | 2016-02-05 | 2016-02-05 | A kind of news based on ontology model storehouse recommends method and device |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN106095762A true CN106095762A (en) | 2016-11-09 |
Family
ID=58536015
Family Applications (2)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201610081578.3A Pending CN106095762A (en) | 2016-02-05 | 2016-02-05 | A kind of news based on ontology model storehouse recommends method and device |
| CN201610966184.6A Active CN106570144B (en) | 2016-02-05 | 2016-11-04 | The method and apparatus of recommendation information |
Family Applications After (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201610966184.6A Active CN106570144B (en) | 2016-02-05 | 2016-11-04 | The method and apparatus of recommendation information |
Country Status (1)
| Country | Link |
|---|---|
| CN (2) | CN106095762A (en) |
Cited By (26)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN106776862A (en) * | 2016-11-28 | 2017-05-31 | 北京奇艺世纪科技有限公司 | A kind of game video searching method and device |
| CN106844322A (en) * | 2017-01-22 | 2017-06-13 | 百度在线网络技术（北京）有限公司 | Intelligent article generation method and device |
| CN106874345A (en) * | 2016-12-23 | 2017-06-20 | 中国科学院自动化研究所 | Media event information extraction method based on object of planning figure |
| CN107330007A (en) * | 2017-06-12 | 2017-11-07 | 南京邮电大学 | A kind of Method for Ontology Learning based on multi-data source |
| CN107391549A (en) * | 2017-06-05 | 2017-11-24 | 北京百度网讯科技有限公司 | News based on artificial intelligence recalls method, apparatus, equipment and storage medium |
| CN107633005A (en) * | 2017-08-09 | 2018-01-26 | 广州思涵信息科技有限公司 | A kind of knowledge mapping structure, comparison system and method based on class teaching content |
| CN107895056A (en) * | 2017-12-29 | 2018-04-10 | 百度在线网络技术（北京）有限公司 | A kind of information recommendation method, device, electronic equipment and storage medium |
| CN108121760A (en) * | 2017-11-23 | 2018-06-05 | 南京邮电大学 | A kind of mining analysis towards OGC geographic information services data is with recommending method |
| CN108241621A (en) * | 2016-12-23 | 2018-07-03 | 北京国双科技有限公司 | The search method and device of legal knowledge |
| CN108491502A (en) * | 2018-03-21 | 2018-09-04 | 腾讯科技（深圳）有限公司 | A kind of method, terminal, server and the storage medium of news tracking |
| CN108510110A (en) * | 2018-03-13 | 2018-09-07 | 浙江禹控科技有限公司 | A kind of water table trend analysis method of knowledge based collection of illustrative plates |
| CN108600337A (en) * | 2018-03-30 | 2018-09-28 | 上海乂学教育科技有限公司 | A kind of best learning Content automatic push method |
| CN109033358A (en) * | 2018-07-26 | 2018-12-18 | 李辰洋 | News Aggreagation and the associated method of intelligent entity |
| CN109101495A (en) * | 2018-08-27 | 2018-12-28 | 上海宝尊电子商务有限公司 | A kind of fashion world document creation method based on image recognition and knowledge mapping |
| CN109145119A (en) * | 2018-07-02 | 2019-01-04 | 北京妙医佳信息技术有限公司 | The knowledge mapping construction device and construction method of health management arts |
| CN109285597A (en) * | 2018-10-08 | 2019-01-29 | 北京健康有益科技有限公司 | A kind of dietotherapy recipe recommendation method, apparatus and readable medium |
| CN109614603A (en) * | 2018-12-12 | 2019-04-12 | 北京百度网讯科技有限公司 | Method and apparatus for generating information |
| CN109635194A (en) * | 2018-12-12 | 2019-04-16 | 北京百度网讯科技有限公司 | Method and apparatus for generating information |
| CN109657043A (en) * | 2018-12-14 | 2019-04-19 | 北京百度网讯科技有限公司 | Automatically generate the method, apparatus, equipment and storage medium of article |
| CN109977291A (en) * | 2019-03-20 | 2019-07-05 | 武汉市软迅科技有限公司 | Search method, device, equipment and storage medium based on physical knowledge map |
| CN110245243A (en) * | 2019-06-20 | 2019-09-17 | 北京百度网讯科技有限公司 | The method and apparatus of news retrieval, electronic equipment, computer-readable medium |
| CN110427465A (en) * | 2019-08-14 | 2019-11-08 | 北京奇艺世纪科技有限公司 | A kind of content recommendation method and device based on word knowledge mapping |
| CN110704743A (en) * | 2019-09-30 | 2020-01-17 | 北京科技大学 | Semantic search method and device based on knowledge graph |
| CN111091454A (en) * | 2019-11-05 | 2020-05-01 | 新华智云科技有限公司 | Financial public opinion recommendation method based on knowledge graph |
| CN112015908A (en) * | 2020-08-19 | 2020-12-01 | 新华智云科技有限公司 | Knowledge graph construction method and system, and query method and system |
| CN112559768A (en) * | 2020-12-11 | 2021-03-26 | 北京中科汇联科技股份有限公司 | Short text mapping and recommendation method |
Families Citing this family (8)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN108268450B (en) * | 2018-02-27 | 2022-04-22 | 百度在线网络技术（北京）有限公司 | Method and apparatus for generating information |
| CN109165350A (en) * | 2018-08-23 | 2019-01-08 | 成都品果科技有限公司 | A kind of information recommendation method and system based on deep knowledge perception |
| CN111353836B (en) * | 2018-12-20 | 2023-07-07 | 百度在线网络技术（北京）有限公司 | Commodity recommendation method, device and equipment |
| CN110162710B (en) * | 2019-05-28 | 2022-06-21 | 北京搜狗科技发展有限公司 | Information recommendation method and device under input scene |
| CN111125372A (en) * | 2019-12-12 | 2020-05-08 | 中汇信息技术（上海）有限公司 | Text information publishing method and device, readable storage medium and electronic equipment |
| CN111291265B (en) * | 2020-02-10 | 2023-10-03 | 青岛聚看云科技有限公司 | Recommendation information generation method and device |
| CN113761214A (en) * | 2020-06-05 | 2021-12-07 | 智慧芽信息科技(苏州)有限公司 | Information flow extraction method, device and equipment |
| CN113032578B (en) * | 2021-03-23 | 2022-12-06 | 平安科技（深圳）有限公司 | Information pushing method and device based on hotspot event and computer equipment |
Family Cites Families (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| KR20110070057A (en) * | 2009-12-18 | 2011-06-24 | 한국전자통신연구원 | Natural language based travel recommendation apparatus and method using location and theme information |
| CN103455487B (en) * | 2012-05-29 | 2018-07-06 | 腾讯科技（深圳）有限公司 | The extracting method and device of a kind of search term |
| CN104573054B (en) * | 2015-01-21 | 2018-06-01 | 杭州朗和科技有限公司 | A kind of information-pushing method and equipment |
-
2016
- 2016-02-05 CN CN201610081578.3A patent/CN106095762A/en active Pending
- 2016-11-04 CN CN201610966184.6A patent/CN106570144B/en active Active
Cited By (38)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN106776862B (en) * | 2016-11-28 | 2021-07-23 | 北京奇艺世纪科技有限公司 | Game video searching method and device |
| CN106776862A (en) * | 2016-11-28 | 2017-05-31 | 北京奇艺世纪科技有限公司 | A kind of game video searching method and device |
| CN106874345A (en) * | 2016-12-23 | 2017-06-20 | 中国科学院自动化研究所 | Media event information extraction method based on object of planning figure |
| CN106874345B (en) * | 2016-12-23 | 2024-02-27 | 中国科学院自动化研究所 | News event information extraction method based on planning-target diagram |
| CN108241621B (en) * | 2016-12-23 | 2019-12-10 | 北京国双科技有限公司 | legal knowledge retrieval method and device |
| CN108241621A (en) * | 2016-12-23 | 2018-07-03 | 北京国双科技有限公司 | The search method and device of legal knowledge |
| CN106844322A (en) * | 2017-01-22 | 2017-06-13 | 百度在线网络技术（北京）有限公司 | Intelligent article generation method and device |
| CN107391549A (en) * | 2017-06-05 | 2017-11-24 | 北京百度网讯科技有限公司 | News based on artificial intelligence recalls method, apparatus, equipment and storage medium |
| CN107391549B (en) * | 2017-06-05 | 2021-06-11 | 北京百度网讯科技有限公司 | Artificial intelligence based news recall method, device, equipment and storage medium |
| US11238097B2 (en) | 2017-06-05 | 2022-02-01 | Beijing Baidu Netcom Science And Technology Co., Ltd. | Method and apparatus for recalling news based on artificial intelligence, device and storage medium |
| CN107330007A (en) * | 2017-06-12 | 2017-11-07 | 南京邮电大学 | A kind of Method for Ontology Learning based on multi-data source |
| CN107633005B (en) * | 2017-08-09 | 2020-11-10 | 广州思涵信息科技有限公司 | Knowledge graph construction and comparison system and method based on classroom teaching content |
| CN107633005A (en) * | 2017-08-09 | 2018-01-26 | 广州思涵信息科技有限公司 | A kind of knowledge mapping structure, comparison system and method based on class teaching content |
| CN108121760A (en) * | 2017-11-23 | 2018-06-05 | 南京邮电大学 | A kind of mining analysis towards OGC geographic information services data is with recommending method |
| CN107895056A (en) * | 2017-12-29 | 2018-04-10 | 百度在线网络技术（北京）有限公司 | A kind of information recommendation method, device, electronic equipment and storage medium |
| CN108510110A (en) * | 2018-03-13 | 2018-09-07 | 浙江禹控科技有限公司 | A kind of water table trend analysis method of knowledge based collection of illustrative plates |
| CN108491502B (en) * | 2018-03-21 | 2022-02-08 | 腾讯科技（深圳）有限公司 | News tracking method, terminal, server and storage medium |
| CN108491502A (en) * | 2018-03-21 | 2018-09-04 | 腾讯科技（深圳）有限公司 | A kind of method, terminal, server and the storage medium of news tracking |
| CN108600337A (en) * | 2018-03-30 | 2018-09-28 | 上海乂学教育科技有限公司 | A kind of best learning Content automatic push method |
| CN109145119A (en) * | 2018-07-02 | 2019-01-04 | 北京妙医佳信息技术有限公司 | The knowledge mapping construction device and construction method of health management arts |
| CN109033358B (en) * | 2018-07-26 | 2022-06-10 | 李辰洋 | Method for associating news aggregation with intelligent entity |
| CN109033358A (en) * | 2018-07-26 | 2018-12-18 | 李辰洋 | News Aggreagation and the associated method of intelligent entity |
| CN109101495A (en) * | 2018-08-27 | 2018-12-28 | 上海宝尊电子商务有限公司 | A kind of fashion world document creation method based on image recognition and knowledge mapping |
| CN109285597A (en) * | 2018-10-08 | 2019-01-29 | 北京健康有益科技有限公司 | A kind of dietotherapy recipe recommendation method, apparatus and readable medium |
| CN109614603A (en) * | 2018-12-12 | 2019-04-12 | 北京百度网讯科技有限公司 | Method and apparatus for generating information |
| CN109635194A (en) * | 2018-12-12 | 2019-04-16 | 北京百度网讯科技有限公司 | Method and apparatus for generating information |
| CN109657043A (en) * | 2018-12-14 | 2019-04-19 | 北京百度网讯科技有限公司 | Automatically generate the method, apparatus, equipment and storage medium of article |
| CN109977291A (en) * | 2019-03-20 | 2019-07-05 | 武汉市软迅科技有限公司 | Search method, device, equipment and storage medium based on physical knowledge map |
| CN110245243B (en) * | 2019-06-20 | 2022-02-01 | 北京百度网讯科技有限公司 | News retrieval method and device, electronic equipment and computer readable medium |
| CN110245243A (en) * | 2019-06-20 | 2019-09-17 | 北京百度网讯科技有限公司 | The method and apparatus of news retrieval, electronic equipment, computer-readable medium |
| CN110427465A (en) * | 2019-08-14 | 2019-11-08 | 北京奇艺世纪科技有限公司 | A kind of content recommendation method and device based on word knowledge mapping |
| CN110427465B (en) * | 2019-08-14 | 2022-03-04 | 北京奇艺世纪科技有限公司 | Content recommendation method and device based on word knowledge graph |
| CN110704743A (en) * | 2019-09-30 | 2020-01-17 | 北京科技大学 | Semantic search method and device based on knowledge graph |
| CN110704743B (en) * | 2019-09-30 | 2022-02-18 | 北京科技大学 | Semantic search method and device based on knowledge graph |
| CN111091454A (en) * | 2019-11-05 | 2020-05-01 | 新华智云科技有限公司 | Financial public opinion recommendation method based on knowledge graph |
| CN112015908A (en) * | 2020-08-19 | 2020-12-01 | 新华智云科技有限公司 | Knowledge graph construction method and system, and query method and system |
| CN112559768A (en) * | 2020-12-11 | 2021-03-26 | 北京中科汇联科技股份有限公司 | Short text mapping and recommendation method |
| CN112559768B (en) * | 2020-12-11 | 2023-02-17 | 北京中科汇联科技股份有限公司 | Short text mapping and recommendation method |
Also Published As
| Publication number | Publication date |
|---|---|
| CN106570144A (en) | 2017-04-19 |
| CN106570144B (en) | 2018-07-27 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN106095762A (en) | A kind of news based on ontology model storehouse recommends method and device | |
| CN108763333B (en) | Social media-based event map construction method | |
| CN109189942B (en) | Construction method and device of patent data knowledge graph | |
| Inzalkar et al. | A survey on text mining-techniques and application | |
| Hsu | Content-based text mining technique for retrieval of CAD documents | |
| Kowalski | Information retrieval architecture and algorithms | |
| CN107609052A (en) | A kind of generation method and device of the domain knowledge collection of illustrative plates based on semantic triangle | |
| CN104281702B (en) | Data retrieval method and device based on electric power critical word participle | |
| Fan et al. | Project-based as-needed information retrieval from unstructured AEC documents | |
| Chawla et al. | Product opinion mining using sentiment analysis on smartphone reviews | |
| CN102609512A (en) | System and method for heterogeneous information mining and visual analysis | |
| JP2005526317A (en) | Method and system for automatically searching a concept hierarchy from a document corpus | |
| CN102184262A (en) | Web-based text classification mining system and web-based text classification mining method | |
| CN104679867B (en) | Address method of knowledge processing and device based on figure | |
| CN101710343A (en) | Body automatic build system and method based on text mining | |
| CN111177591A (en) | Knowledge graph-based Web data optimization method facing visualization demand | |
| CN112231494B (en) | Information extraction method and device, electronic equipment and storage medium | |
| CN103678412A (en) | Document retrieval method and device | |
| CN102043793A (en) | Knowledge-service-oriented recommendation method | |
| CN104346382B (en) | Use the text analysis system and method for language inquiry | |
| Jafari et al. | Unsupervised keyword extraction for hashtag recommendation in social media | |
| Jiang et al. | Research on BIM-based Construction Domain Text Information Management. | |
| Chitrakala et al. | Concept-based extractive text summarization using graph modelling and weighted iterative ranking | |
| Rao et al. | Enhancing multi-document summarization using concepts | |
| Sulaiman et al. | An object properties filter for multi-modality ontology semantic image retrieval |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| C06 | Publication | ||
| PB01 | Publication | ||
| C02 | Deemed withdrawal of patent application after publication (patent law 2001) | ||
| WD01 | Invention patent application deemed withdrawn after publication |
Application publication date: 20161109 |