CN109359240A - A method of automation search - Google Patents
A method of automation search Download PDFInfo
- Publication number
- CN109359240A CN109359240A CN201811070856.0A CN201811070856A CN109359240A CN 109359240 A CN109359240 A CN 109359240A CN 201811070856 A CN201811070856 A CN 201811070856A CN 109359240 A CN109359240 A CN 109359240A
- Authority
- CN
- China
- Prior art keywords
- search
- user
- scene
- text
- block chain
- 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
Abstract
The invention discloses a kind of methods of automation search, comprising the following steps: step 1: creation user search intent computing system, the user search intent computing system, including subsystem: creating the personal scene event map of user；Before search, one of the duplication of user's operation, shearing, paste event or multiple combinations；Before search, specific event that user contacts in equipment；Before search, message, voice, video, e-mail Content of communciation based on user；Query based on user；Based on nearest hot news event；New knowledge point；Based on contact person；Taking pictures based on user, video；Personal scene search record storehouse based on user；Conjecture based on deep learning；Step 2: obtaining the search intention of user；Step 3: generating search text based on being intended to；Step 4: appliance for releasing single frame shows a no less than search text, clicks and open search and webpage；Directly display the searched page of search text.
Description
Technical field
Invention is related to the method and technology field of search, specially a kind of method of automation search.
Background technique
Search engine can be used to the particular document in locating documents set.In addition, search engine can be used to determine
Particular keywords or phrase in the document of position.One or more indexes can be used to position particular document, key in search engine
Word or phrase.In addition, search engine can execute boolean sum other operations during positioning specific information.In order to design one
Kind is very necessary the method for searching for the stronger automation search of quicker, more accurate and purpose.
Summary of the invention
A kind of method for being designed to provide automation search of invention, to solve mentioned above in the background art ask
Topic.
To achieve the above object, invention provides the following technical solutions: a method of automation search, including following step
It is rapid:
Step 1: creation user search intent computing system, the user search intent computing system, including subsystem:
(1) the personal scene event map of user is created；
(2) before searching for, one of the duplication of user's operation, shearing, paste event and multiple combinations；
(3) before searching for, specific event that user contacts in equipment；
(4) before searching for, message, voice, video, e-mail Content of communciation based on user；
(5) based on the query of user；
(6) based on nearest hot news event；
(7) it is based on new knowledge point；
(8) contact person, contact method and address list satellite information are based on；The address list satellite information include: unit,
Location and remark information；
(9) taking pictures based on user, video；
(10) the personal scene search record storehouse based on user；
(11) based on the probabilistic forecasting of deep learning；
(12) document content edited based on user.
Step 2: obtaining the search intention of user；
Step 3: generating search text based on being intended to；Text is searched in the generation, comprising:
(1) existing text is directlyed adopt；
(2) in existing text basis, optimization generates search text；It is initial by entry to existing text first
URL is parsed, and then page link resolver is based on the HTML rule of definition by other entries URL included in the page
It saves into the library entry URL, crawl request is issued to the URL of parsing later, the data that will acquire are sent to knowledge data parsing
Device is responsible for knowledge data duplicate removal required for obtaining in each entry page by knowledge data resolver, finally to filtering after
Knowledge data carry out triple store process of aggregation, reach optimization generate search text；
(3) intelligence generates search text；
The intelligence generates search text
1. picture, audio, video, animation are converted into text；
2. long sentence, chapter text are intelligently generated by text summarization technique and meet preset search key
Number of words threshold value；Automatic abstract process includes three basic steps:
S1, text analyzing process: long sentence, chapter text are analyzed and processed, identify redundancy；
The selection and extensive process of S2, content of text: recognizing important information from document, by taking passages formula digest and understanding
The method compressed text of formula digest, forming digest indicates；Extracts type digest is made of the segment extracted in original text, understanding type
As long as digest is formed after reorganizing to original text content；
The conversion and generating process of S3, digest: realizing the recombination to textual content or generates digest according to semantic expressiveness, and
Ensure the continuity of digest；
Step 4: appliance for releasing single frame shows a no less than search text, clicks and open search and webpage；Or/and it directly displays
Searched page based on search text.
Preferably, the personal scene event map step of the creation user includes:
Step 1: presetting multi-level scene tag library, and preset the association probability between corresponding scene logical relation, scene
Sequential relationship, chain type expressing in series relationship between computation rule and scene；
Step 2: the base station used by user mobile phone and the dynamic multilayer for obtaining acquisition and calculating user of global position system
Secondary scene tag value, corresponding storage are raw according to the preset scene logical relation into preset multi-level scene tag library
At individual subscriber scene；
Step 3: being based on multiple independent individual subscriber scenes, according to the association probability computation rule between scene, calculate pre-
If association probability between multiple independent individual subscriber scenes in the period, according to the sequential relationship between preset scene, chain
Formula expressing in series relationship generates individual subscriber scene map.
Preferably, the personal scene search record storehouse is stored as block chain Cloud Server, including privately owned block chain equipment,
Publicly-owned block chain equipment, node server, central server, hardware firewall, block chain memory module, sending module and reception
Module；Wherein privately owned block chain equipment, publicly-owned block chain equipment are publicly-owned and privately owned by block chain memory module storage individual's
Event acquisition data, be sent in node server by sending module distribution in real time, pass through receiving module receive it is real
When the event data that acquires and uploaded in central server by node server, the Cloud Server further includes hardware fire prevention
Wall；Data are divided into block chain backup module, block chain isolation module and block chain node module in block chain memory module, will
Existing block chain data are isolated into individual module according to function, and block chain node module is arranged in node server, support
Across a network access, when run-time error occurs in module, block chain backup module, block chain isolation module are carried out for single module
Upgrade restoration updating.
Preferably, the conjecture based on deep learning includes: to be applied to personal scene search record storehouse, to the number of generation
According to filtering, removing artefact pretreatment operation, the people of data, time, weather, place, gender, occupation, age, reality are then extracted
When etc. characteristic parameters, carry out the principal component analysis of PCA correlation, extract principal component characteristic parameter, then train identification model, selection
Radial basis function is the support vector machines of kernel function as Emotion identification model.
Compared with prior art, advantageous effect of the invention is: presetting multi-level scene library by setting up, generates of user
People's scene；Theme quantity is abundant during studying obtained user model, solves the sparsity of initial data；Creation is personal
Contextual data improves the effective percentage of function recommendation, so that some multidimensional new feature scenes so that function recommends to have more specific aim
Information supplement recommends to obtain a large number of users big data rapidly by function；It is connected to by the information of various dimensions morphologic correlation isomery
A topological network, message sample obtained from together are comprehensive；Effectively user is helped quickly to find interested and high quality letter
Breath, semantic association provides background knowledge, and to search for quicker, more accurate and purpose stronger, has advanced accurate pre-
Survey effect.It can be with the particular document in precise positioning collection of document.In addition, the search engine can further be used to positioning text
Particular keywords or phrase in shelves and particular document, keyword or phrase are positioned using one or more index.Quickly
Boolean sum other operations are executed during positioning specific information.The data of generation are filtered, remove artefact etc. with pretreatment
Operation, the characteristic parameters such as the people for then extracting data, time, weather, place, gender, occupation, age, real-time, carries out PCA phase
The principal component analysis of closing property, extracts principal component characteristic parameter, then trains identification model, selects radial basis function for the branch of kernel function
Vector machine is held as Emotion identification model, not only increases promotion effect, accuracy, by the feature of preceding PCA principal component analysis
Pre-treatment is conducive to the convergence of data.
Detailed description of the invention
Fig. 1 is the flow diagram of a kind of method of automation search；
Fig. 2 is that the optimization of a kind of method of automation search generates search text flow diagram.
Specific embodiment
Following will be combined with the drawings in the embodiments of the present invention, and technical solution in the embodiment of the present invention carries out clear, complete
Site preparation description, it is clear that described embodiments are only a part of the embodiments of the present invention, instead of all the embodiments.It is based on
Embodiment in the present invention, it is obtained by those of ordinary skill in the art without making creative efforts every other
Embodiment shall fall within the protection scope of the present invention.
Embodiment 1:
Fig. 1 and Fig. 2 are please referred to, invention provides a kind of technical solution: a method of automation search, including following step
It is rapid:
Step 1: creation user search intent computing system, the user search intent computing system, including subsystem:
(1) the personal scene event map of user is created；
(2) before searching for, one of the duplication of user's operation, shearing, paste event and multiple combinations；
(3) before searching for, specific event that user contacts in equipment；
(4) before searching for, message, voice, video, e-mail Content of communciation based on user；
(5) based on the query of user；
(6) based on nearest hot news event；
(7) it is based on new knowledge point；
(8) contact person, contact method and address list satellite information are based on；The address list satellite information include: unit,
Location and remark information；
(9) taking pictures based on user, video；
(10) the personal scene search record storehouse based on user；
(11) based on the probabilistic forecasting of deep learning；
(12) document content edited based on user.
Step 2: obtaining the search intention of user；
Step 3: generating search text based on being intended to；Text is searched in the generation, comprising:
(1) existing text is directlyed adopt；
(2) in existing text basis, optimization generates search text；It is initial by entry to existing text first
URL is parsed, and then page link resolver is based on the HTML rule of definition by other entries URL included in the page
It saves into the library entry URL, crawl request is issued to the URL of parsing later, the data that will acquire are sent to knowledge data parsing
Device is responsible for knowledge data duplicate removal required for obtaining in each entry page by knowledge data resolver, finally to filtering after
Knowledge data carry out triple store process of aggregation, reach optimization generate search text；
(3) intelligence generates search text；
The intelligence generates search text
1. picture, audio, video, animation are converted into text；
2. long sentence, chapter text are intelligently generated by text summarization technique and meet preset search key
Number of words threshold value；Automatic abstract process includes three basic steps:
S1, text analyzing process: long sentence, chapter text are analyzed and processed, identify redundancy；
The selection and extensive process of S2, content of text: recognizing important information from document, by taking passages formula digest and understanding
The method compressed text of formula digest, forming digest indicates；Extracts type digest is made of the segment extracted in original text, understanding type
As long as digest is formed after reorganizing to original text content；
The conversion and generating process of S3, digest: realizing the recombination to textual content or generates digest according to semantic expressiveness, and
Ensure the continuity of digest；
Step 4: appliance for releasing single frame shows a no less than search text, clicks and open search and webpage；Or/and it directly displays
Searched page based on search text.
Preferably, the personal scene event map step of the creation user includes:
Step 1: presetting multi-level scene tag library, and preset the association probability between corresponding scene logical relation, scene
Sequential relationship, chain type expressing in series relationship between computation rule and scene；
Step 2: the base station used by user mobile phone and the dynamic multilayer for obtaining acquisition and calculating user of global position system
Secondary scene tag value, corresponding storage are raw according to the preset scene logical relation into preset multi-level scene tag library
At individual subscriber scene；
Step 3: being based on multiple independent individual subscriber scenes, according to the association probability computation rule between scene, calculate pre-
If association probability between multiple independent individual subscriber scenes in the period, according to the sequential relationship between preset scene, chain
Formula expressing in series relationship generates individual subscriber scene map.
Preferably, the personal scene search record storehouse is stored as block chain Cloud Server, including privately owned block chain equipment,
Publicly-owned block chain equipment, node server, central server, hardware firewall, block chain memory module, sending module and reception
Module；Wherein privately owned block chain equipment, publicly-owned block chain equipment are publicly-owned and privately owned by block chain memory module storage individual's
Event acquisition data, be sent in node server by sending module distribution in real time, pass through receiving module receive it is real
When the event data that acquires and uploaded in central server by node server, the Cloud Server further includes hardware fire prevention
Wall；Data are divided into block chain backup module, block chain isolation module and block chain node module in block chain memory module, will
Existing block chain data are isolated into individual module according to function, and block chain node module is arranged in node server, support
Across a network access, when run-time error occurs in module, block chain backup module, block chain isolation module are carried out for single module
Upgrade restoration updating.
Embodiment 2:
The method for creating multi-level event and scene TuPu method, comprising:
Step 1: presetting multi-level event tag library and multi-level scene tag library, and preset corresponding scene event logic
The association probability computation rule and scene between association probability computation rule, event between relationship, scene logical relation, scene
Map method for building up, event map method for building up；The multi-level event tag, including level-one event tag and multistage subordinate's thing
Part label；Level-one event tag includes: main body event and object event；The main body event refers in terminal generation
Event；The object event refers to that terminal passes through the external event occurred of audiovisual identifying system detecting identification；
Specifically, in the present embodiment, multi-level event tag library is preset to described, does following design:
The main body event refers to that main body event is level-one event tag in the event that terminal occurs, it includes
Secondary event label: it chats, make a phone call, using APP, browsing；In secondary event label includes three-level event mark using APP
Label: wechat is used.
The object event refers to that terminal passes through the external thing occurred of audiovisual identifying system detecting identification
Part, object event are level-one event tags；It includes secondary event labels: having a meeting, goes window-shopping, watches movie, sings；Secondary event mark
Watching movie comprising three-level event tag in label: science fiction film is seen；Science fiction film includes level Four event mark to seeing in three-level event tag
Label: A Fanda is seen.
Step 2: terminal identifies the multi-level scene tag value for obtaining and/or calculating user by audiovisual,
Corresponding storage, according to the preset scene logical relation, generates user's into the preset multi-level scene tag library
Personal scene is based on multiple independent individual subscriber scenes, according to the association probability computation rule between scene, calculates preset time
Association probability between multiple independent individual subscriber scenes generates user according to preset scene map method for building up in section
People's scene map；The multi-level scene tag, including level-one scene tag and multistage subordinate's scene tag, the level-one scene
Label includes four people, time, place, weather labels；The scene map method for building up refers to multiple only in certain time
Vertical individual subscriber scene establishes chain type expressing in series figure based on forward and reverse sequential relationship, and in adjacent individual subscriber
The probability that its association occurs is marked on side between scene；The preset time period, can be customized by users；
The sequential relationship includes: Before: scene occurs before another scene；After: scene is at another
Occur after scape；Includes: one scene includes another scene；Is_Included: one scene is by another scene packet
Contain；During: one scene keeps a state whithin a period of time；Simultane-ous: occur simultaneously；Iafter: scene
Occur immediately following another scene, and they are not overlapped, uninterruptedly；Ibe-fore scene occurs before another scene, and it
Be not overlapped, uninterruptedly；1Iaentity: Same Scene is indicated；Begins: one scene starts that another scene is caused to start；
Ends: one scene terminates that another scene is caused to terminate；BegunBy: one scene due to another scene starts, with
Begins is opposite；Ended_By: one scene terminates due to another scene terminates, opposite with Ends；
The calculation method of the sequential relationship are as follows: while acquiring scene, acquisition scene occur at the beginning of point and
End time point is calculated by the time and determines sequential relationship；The determination method of the forward and reverse sequential relationship: according to as above
13 kinds of sequential relationships that calculation method obtains, by the relationship for indicating the adjacent appearance of scene in time shaft；The chain type series connection table
Up to referring to: based on forward and reverse sequential relationship, the chain type expressing in series figure between multiple independent individual subscriber scenes is established,
And probabilistic information is marked on adjacent side；
Embodiment 3
The method that timing creates personal scene various dimensions characteristic spectrum, comprising:
Step 1: presetting multi-level scene tag library, and preset the association probability between corresponding scene logical relation, scene
Sequential relationship, chain type expressing in series relationship between computation rule and scene, the multi-level scene tag, including level-one
Scene tag and multistage subordinate's scene tag, the level-one scene tag includes four people, time, place, weather labels；
Specifically, in the present embodiment, multi-level scene tag library is preset to described, does following design:
1. people is level-one scene tag, corresponding tag entry is preset；It includes second level scene tags: gender/occupation/wedding
Relation by marriage state/age/health status/mood, presets corresponding tag entry；Doctor in second level scene tag includes three-level scene
Label: post/position presets corresponding tag entry, and logically library is built in subordinate relation arrangement to all preset tag entries.
2. the time is level-one scene tag；It includes second level scene tags: state/birthday/working day in season/year；Two
Working day in grade scene tag includes three-level scene tag: go to work the period in the morning/working period in the afternoon；In three-level scene tag
The working period in the morning include level Four scene tag: just the next period of working period/fastly, preset corresponding tag entry, owned
Logically library is built in subordinate relation arrangement to preset tag entry.
3. place is level-one scene tag；It includes second level scene tags: family/company/dining room/cinema；Second level field
Dining room in scape label includes three-level scene tag: western-style restaurant/Chinese Restaurant；Chinese Restaurant in three-level scene tag includes level Four field
Scape label: corresponding tag entry is preset in Hunan cuisine shop/Sichuan cuisine shop, and all preset tag entries logically arrange by subordinate relation
Column build library.
4. weather is that it includes second level scene tags for level-one scene tag: fair weather/not fair weather/climate disaster；Second level
Fair weather does not include three-level scene tag: meteorology/temperature/wind-force in scene tag, presets corresponding tag entry, all default
Tag entry logically subordinate relation arrangement build library.
Step 2: the base station used by user mobile phone or/and global position system are moved and are obtained and/or calculated user
Multi-level scene tag value, corresponding storage is into preset multi-level scene tag library, according to the preset scene logic
Relationship generates individual subscriber scene, the multi-level scene tag value of the people, by preset user's representation data library or/and reality
When calculate and obtain, the multi-level scene tag value of the time automatically obtains, the multi-level scene tag in the place by system
Value, the base station used by user mobile phone or/and global position system are dynamic to be obtained, the multi-level scene tag value of the weather, by
System automatically obtains, the scene logical relation, is to define scene mark between each scene tag in multi-level scene tag library
After label obtain corresponding scene tag value, for forming the rule of description individual subscriber scene；
Embodiment 4
Preferably, the conjecture based on deep learning includes: to be applied to personal scene search record storehouse, to the number of generation
According to filtering, removing artefact pretreatment operation, the people of data, time, weather, place, gender, occupation, age, reality are then extracted
When etc. characteristic parameters, carry out the principal component analysis of PCA correlation, extract principal component characteristic parameter, then train identification model, selection
Radial basis function is the support vector machines of kernel function as Emotion identification model.
Working principle: multi-level scene library is preset by setting up, generates the personal scene of user；Study obtained user's mould
Theme quantity is abundant during type, solves the sparsity of initial data；Personal contextual data is created, so that function is recommended more
Have specific aim, improve function recommendation effective percentage so that some multidimensional new feature scene informations supplement recommended by function it is fast
Speed obtains a large number of users big data；A topological network obtained from being linked together as the information of various dimensions morphologic correlation isomery
Network, message sample are comprehensive；Effectively user is helped quickly to find interested and high quality information, semantic association provides background and knows
It is stronger to know quicker, the more accurate and purpose of search, there is advanced accurate prediction effect.It can be with precise positioning document
Particular document in set.In addition, the search engine can further be used to particular keywords or phrase in locating documents
Particular document, keyword or phrase are positioned with one or more indexes are used.Quickly during positioning specific information
Execute the other operations of boolean sum.The data of generation are filtered, remove the pretreatment operations such as artefact, then extract data people,
Time, weather, place, gender, occupation, the age, in real time etc. characteristic parameters, carry out the principal component analysis of PCA correlation, extract it is main at
Divide characteristic parameter, then trains identification model, selecting radial basis function is the support vector machines of kernel function as Emotion identification mould
Type not only increases promotion effect, accuracy, and the feature pre-treatment by preceding PCA principal component analysis is conducive to the convergence of data.
Quicker, the more accurate and purpose of method search of automation search is stronger, has advanced accurate
Prediction effect.
Obviously, various changes and modifications can be made to the invention without departing from essence of the invention by those skilled in the art
Mind and range.In this way, if these modifications and changes of the present invention belongs to the range of the claims in the present invention and its equivalent technologies
Within, then the present invention is also intended to include these modifications and variations.
Claims (4)
1. a kind of method of automation search, it is characterised in that: the following steps are included:
Step 1: creation user search intent computing system, the user search intent computing system, including subsystem:
(1) the personal scene event map of user is created；
(2) before searching for, one of the duplication of user's operation, shearing, paste event and multiple combinations；
(3) before searching for, specific event that user contacts in equipment；
(4) before searching for, message, voice, video, e-mail Content of communciation based on user；
(5) based on the query of user；
(6) based on nearest hot news event；
(7) it is based on new knowledge point；
(8) contact person, contact method and address list satellite information are based on；The address list satellite information include: unit, address with
Remark information；
(9) taking pictures based on user, video；
(10) the personal scene search record storehouse based on user；
(11) based on the probabilistic forecasting of deep learning；
(12) document content edited based on user.
Step 2: obtaining the search intention of user；
Step 3: generating search text based on being intended to；Text is searched in the generation, comprising:
(1) existing text is directlyed adopt；
(2) in existing text basis, optimization generates search text；First to existing text by the initial URL of entry into
Row parsing, then page link resolver is saved other entries URL included in the page based on the HTML rule of definition
Into the library entry URL, crawl request is issued to the URL of parsing later, the data that will acquire are sent to knowledge data resolver, by
Knowledge data resolver is responsible for knowledge data duplicate removal required for obtaining in each entry page, finally to filtered knowledge
Data carry out triple store process of aggregation, reach optimization and generate search text；
(3) intelligence generates search text；
The intelligence generates search text
1. picture, audio, video, animation are converted into text；
2. long sentence, chapter text intelligently to be generated to the number of words for meeting preset search key by text summarization technique
Threshold value；Automatic abstract process includes three basic steps:
S1, text analyzing process: long sentence, chapter text are analyzed and processed, identify redundancy
Information；
The selection and extensive process of S2, content of text: recognizing important information from document, by taking passages formula digest and understanding formula text
The method compressed text plucked, forming digest indicates；Extracts type digest is made of the segment extracted in original text, understanding type digest
As long as being formed after being reorganized to original text content；
The conversion and generating process of S3, digest: realizing the recombination to textual content or generates digest according to semantic expressiveness, and ensures
The continuity of digest；
Step 4: appliance for releasing single frame shows a no less than search text, clicks and open search and webpage；Or/and it directly displays and is based on
Search for the searched page of text.
2. a kind of method of automation search as described in claim 1, it is characterised in that: the personal scene of the creation user
Event map step includes:
Step 1: presetting multi-level scene tag library, and preset the calculating of the association probability between corresponding scene logical relation, scene
Sequential relationship, chain type expressing in series relationship between rule and scene；
Step 2: the base station used by user mobile phone and the dynamic multi-level field for obtaining acquisition and calculating user of global position system
Scape label value, corresponding storage, according to the preset scene logical relation, are generated and are used into preset multi-level scene tag library
Family individual's scene；
Step 3: multiple independent individual subscriber scenes are based on, according to the association probability computation rule between scene, when calculating default
Between association probability between multiple independent individual subscriber scenes in section, according to the sequential relationship between preset scene, chain type string
Join relationship between expression, generates individual subscriber scene map.
3. a kind of method of automation search as described in claim 1, it is characterised in that: individual's scene search record storehouse
Be stored as block chain Cloud Server, including privately owned block chain equipment, publicly-owned block chain equipment, node server, central server,
Hardware firewall, block chain memory module, sending module and receiving module；Wherein privately owned block chain equipment, publicly-owned block chain are set
The standby publicly-owned and privately owned event acquisition data personal by the storage of block chain memory module, are distributed by sending module in real time
Formula is sent in node server, is received the event data acquired in real time by receiving module and is uploaded to by node server
In central server, the Cloud Server further includes hardware firewall；It is standby to be divided into block chain for data in block chain memory module
Existing block chain data are isolated into individual mould according to function by part module, block chain isolation module and block chain node module
Block, block chain node module are arranged in node server, support across a network access, when run-time error occurs in module, block
Chain backup module, block chain isolation module carry out upgrading restoration updating for single module.
4. a kind of method of automation search as described in claim 1, it is characterised in that: the conjecture based on deep learning
Include: to be applied to personal scene search record storehouse, the data of generation are filtered, remove artefact pretreatment operation, are then extracted
The people of data, time, weather, place, gender, occupation, age, the characteristic parameters such as in real time, carry out PCA correlation principal component point
Principal component characteristic parameter is extracted in analysis, then trains identification model, selects radial basis function for the support vector machines conduct of kernel function
Emotion identification model.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201811070856.0A CN109359240A (en) | 2018-09-07 | 2018-09-07 | A method of automation search |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201811070856.0A CN109359240A (en) | 2018-09-07 | 2018-09-07 | A method of automation search |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN109359240A true CN109359240A (en) | 2019-02-19 |
Family
ID=65350694
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201811070856.0A Pending CN109359240A (en) | 2018-09-07 | 2018-09-07 | A method of automation search |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN109359240A (en) |
Cited By (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN110113635A (en) * | 2019-04-25 | 2019-08-09 | 广州智伴人工智能科技有限公司 | A kind of method and system of automatic broadcasting PUSH message |
| CN112269899A (en) * | 2020-10-20 | 2021-01-26 | 西安工程大学 | Video retrieval method based on deep learning |
-
2018
- 2018-09-07 CN CN201811070856.0A patent/CN109359240A/en active Pending
Cited By (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN110113635A (en) * | 2019-04-25 | 2019-08-09 | 广州智伴人工智能科技有限公司 | A kind of method and system of automatic broadcasting PUSH message |
| CN110113635B (en) * | 2019-04-25 | 2021-05-25 | 广州智伴人工智能科技有限公司 | Method and system for automatically playing push message |
| CN112269899A (en) * | 2020-10-20 | 2021-01-26 | 西安工程大学 | Video retrieval method based on deep learning |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN100405371C (en) | Method and system for abstracting new word | |
| US9235653B2 (en) | Discovering entity actions for an entity graph | |
| CN102880687B (en) | Based on individual interactive data retrieval method and the system thereof of label technique | |
| WO2016150083A1 (en) | Information input method and apparatus | |
| CN100595765C (en) | Medium player based key words content issue method and system | |
| CN104516949B (en) | Web data treating method and apparatus, inquiry processing method and question answering system | |
| CN102737022B (en) | Method and device for acquiring and searching relevant knowledge information | |
| CN110377908B (en) | Semantic understanding method, semantic understanding device, semantic understanding equipment and readable storage medium | |
| CN102054033A (en) | Expression search engine, expression management system using same and expression management method | |
| CN104991943A (en) | Music searching method and apparatus | |
| CN105718585B (en) | Document and label word justice correlating method and its device | |
| CN103020293A (en) | Method and system for constructing ontology base in mobile application | |
| CN104978332B (en) | User-generated content label data generation method, device and correlation technique and device | |
| CN103530429B (en) | Webpage content extracting method | |
| CN103546634A (en) | Handhold equipment theme control method and handhold equipment theme control device | |
| CN104978314A (en) | Media content recommendation method and device | |
| CN103399855B (en) | Behavior intention determining method and device based on multiple data sources | |
| CN104111935A (en) | Server and microblog pushing method and system | |
| CN109918656A (en) | A kind of live streaming hot spot acquisition methods, device, server and storage medium | |
| CN103390044A (en) | Method and device for identifying linkage type POI (Point Of Interest) data | |
| CN110807326A (en) | Short text keyword extraction method combining GPU-DMM and text features | |
| CN109359240A (en) | A method of automation search | |
| KR101682659B1 (en) | Method for customized news alarm based on keyword and management server for news search for the same | |
| CN105279159B (en) | The reminding method and device of contact person | |
| CN109255385A (en) | A kind of method, apparatus and its application automatically creating multi-level event and scene TuPu method |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| PB01 | Publication | ||
| PB01 | Publication | ||
| WD01 | Invention patent application deemed withdrawn after publication |
Application publication date: 20190219 |
|
| WD01 | Invention patent application deemed withdrawn after publication |